A Quick Approach To Statistics by G.R.pashA

By: Rafaqat

http://stat9943.blogspot.com
A Quick Approach to Statistics with Questions and Answers

1111111111
Contents

Preface
By: Rafaqat
Chapter 1: Introductory Statistics

Basic Statistics 1
.. Measures of Central Tendency
I
7 i
Measures of DiSpersion '
Questions-Answers 9 1
Exercises 11
18
~
J Chapter 2: Basic Probability Theory
Basics in Probability 23
Some Probability Rules
23
Counting Rules 28
Questions-Answers 29
Exercises 31
34
Chapter 3: Random Vari~bles
Exercises 46
48
Chapter 4: Discrete rrobability Distributions
Questions~Answers 49
Exercises 57
59
Chapter 5: Continuous Probability Distributions 63 !
Questions-Answers
Exercises 81
85 1
i
A QMJa Approach to Stat/sJJa wJJh Questions and Answers
Chapter 6: Regression & Correlation 91

Regression 91
Correlation 94
Exercises I0 I
Chapter 7: Sampling 105

. Basics of Sampling 105
Sampling Techniques 101
Bias and Errors 111
By: Rafaqat
Exercises 125
Chapter 8: Statistical Inference 127

Exercises 140
Chapter 9: Design and Analysis of Experiment 143

Exercises 151
Chapter 10: Analysis of Time Series 153

Exercises 164
Chapter 11: Index Numben 167

Questions-~wcrs 174
Exercises 177
Chapter 12: Nonpatametric Statistics 181

Questions-Answers. 185
Exercises 189
Answers to Exercises 195
Bibliography 201
Subject Index 203
rs A Quick Approach to Statistics with Questions and Answers
Chapter 1
. 11111111
Introductory Statistics
;
Basic Statistics
Statistics
By: Rafaqat
Statistics is the collection of methods- for planning experiments, obtaining

data and then organizing, summarizing, presenting, analyzing, interpreting
and drawing conclusions based on clata.
Statistical Methods
Statistical methods are those ways that are used to collect, present, analyze,
and interpret quantitative data.
Type of Statistics
There are two major types of Statistics: Descriptive Statistics and Infere,.ntial
Statistics:
Descriptive Statistics
It consists of methods for organizing and summarizing information in a
presentable and effective way.
Inferential Statistics -
It consists of methods of drawing conclusions about a population based on
information obtained from a sample of the population.
Data
A collection of fatts from which conclusions may be drawn is referred as
data.
Observation -
Any sort of recording of information:1i; called observation.
Chapter 1: IntroductoryStatistlcs
r
A. Quick Approach to Statistics with Questions and Answers
Types of Data
Generally, data can be classified by their nature and 'fay of collection.
Types of Data (Nature) .
Qualitative Data
Qualitative (or Categorical or Attribute) data can be separated into different
categories tl:tat are distinguished by some non-numerical characteristics. For
example, gender of person, blood type, and eye color etc.
Quantitative Data
Quantitative data consist of numbers representing counts or measurements
such as number of patients in a hospital, ages ofa group of persons, data
about height and weight of individuals ety
Quantitative data can be further classified into discrete and .continuous data.
By: Rafaqat
All type of c.ount data are referred as discrete data where measured data are
referred as continuous data.
Discrete Data . .
Data obtained by categorizing subjects so that there is a distinct interval
between any two possible values e.g_., number of patients in a hospital and
number of chairs in a room etc.
Continuous Data
Continuous data resuh from infinitely many possible values that can be
associated with points on a continuous sCale in such a way that there are no
gaps or interruptions. For example, data about height and weight of
individuals etc.
,
Types of Data (Collection)
Primary Data .
The data collected directly from pc0ple and organizations via questionnaires
or surveys before being analyzed to reach conclusions concerning the issues
covered in the.questionnaire or survey.
Secondary Data
The data that have undergone any ~ort of treatment by statistical methods.
In other words, the data that have already been assembled, having been
coltected for some other purpose., are referred as secondary data. Sources
include census reports, trade publications, and subscription services.
Chapter I: Introductory Statistics 2
A Quick Approai:h to Statistics with Questions and Answers
Scales of Measurement
Nominal Scale
)n nomin.al scale is categorized by data that consists of names, labels or
categories only. Such data cannot be arranged in an ordering scheme. For
example, gender; "male" and "female", response; "yes" or "no",-etc.
Ordinal Scale
The ordinal scale involves data that may be arranged in some order but
differences between data values either cannot be determined or are
meaningless. For example, in a sample of 36 'stereo speakers, 12 were rated
good, 16 wei;e rated better and 8 were rated best.
Interval Scale
By: Rafaqat
The interval scale is like the ordinal scale with the additional property that
meaningful amount of differences between data can be determmedJ
However, there is no inherent (natural) zero starting point. Interval scal6
take the notion ofranking items in order one step further, s\nce the distance
between adjacent poinfs on the scale are equal. For instance, the Fahrenheit
scale is an interval scale, since each degree is equal but there is no absolute
zero point. This means that although we can add and subtract degrees (I 00
is 10 warmer than 90), we cannot multiply values or create ratios (100 is
not twice as warm as 50). What is important in determining whether a
scale is considered interval or not is the underlying intent regarding the
equal intervals: although in' an IQ scale, the intervals are not necessarily
equal {e:g. the difference between 105 and 11.0 is not really .the same as
between 80 and 85), behavioral scientists are willing to assume that most of
their measures are interval scales as this allows the calculation of averages -
mode, median and mean -, the range and standard deviation. ,.
Ratio Scale
The ratio scale is the intervat scale modified to include the inherent zero
stlrting point. FOJ: values at this level, differences and ratios are meaningful}
Ratio scales are the most sophisticated of scales, since it incorporates all tlie
characteristics of nominal, ordinal and interval scales. As a resuft, a large
number of descriptive calc\Jlations are applicable such as when respondents
are asked for their age, height, income etc.
Chapter I: Introductory Statistics J
Presentation of Data
Classification
A classification is the separation or ordering of objects into classes where
classes are categories for grouping data.
Tabulation
Tabulation is placement of data into rows and columns with suitable heads
and subheads.
Frequency
The frequency of a particular class is the number of original scores that fall
into that class. Simply, frequency is the number of times that a repeated
observation occurs.
By: Rafaqat
Cumulative Frequency
The cumulative frequency for ~ class is the sum of the frequencies of that
class and all the previous classes.
Relative Frequency
The relative frequency of a particular class can be found by dividing by the
class frequency by the total of all frequencies.
Grouped data
The data presented in the form of a frequency distribution are called
grouped data.
Frequency Distribution
The division of counts (frequencies) of number of scores that fall into each
class (category) is called frequency distributio!Vln other words, a listing of
cla$ses and their frequencies is called frequency distribution. A table that
represents classes along with their respective class frequencies is called
frequency table.
Class Limits
The values or numbers specifying a class are called class limi~ The
smallest value specifying a cl?SS is called lower class limit while the largest
value for specifying a class is called upper class limi9
A Quick Approach to Statistics wiih Questions and Answers
Class Boundaries
Class boundaries are the numbers used to separate classes, but without the
gaps created by class limits:) They are obtained by increasing the upper class
limits and decreasing the-1'ower class limits by the same amount so that
there are no gaps between consecutive classes. These boundaries are also
called precise limits or tnie limits. ,
Class Mark (Midpoints)

Class marks or midpoints of the classes are obtained.by adding lower class
limits to the corresponding upper class limits and dividing by 2.
Class Interval (Width)

Class interval or class width is the difference between two consecutive
lower class limits or two consecutive lower class boundaries.
By: Rafaqat
In the table below, class interval (width)= 196-191=201-196=5.
Cholesterol
Class Class Cumulative Relative
Level Frequency
Boundaries Mark Frequency Frequency
(Class Limits)
191- 195 190.S- 195.S 193 l l 1/25 = 0.04
196-200 195.5-200.5 198 3 3+1 =4 3125=0.12
201-205 200.S - 205.5 203 4 4+4 =8 0.16
206-210 205.5-210.5 208 7 7+8 =IS 0.28
211-215 210.5-215.5 213 5 5+15 = 20 0.20
216-220 215.5 - 220.5 218 4 4+20 = 24 0.16
221-225 220.5 - 225.5 223 l 1+24 = 25 0.04.
Total 25 1.00
Graph
A drawing representing the relationship between data sets is called the
graph.
Histogram
A graph that displays the classes on the horizontal axis and the frequencies
of the classes on the vertical axis is called histogram. The frequency of each
class is represented by a vertical bar whose height is proportional to the
frequency of that class.
Chapter 1: Introductory Statistics s .

Frequen~y Polygon
Polygon means closed shape. A frequency polygon is obtained by joining
the mid points of the adjacent bars of histograms with straight lines and then
joining the both ends w.ith X-axis by assuming class frequencies zero at
those pointS.
Frequency Curve
When frequency polygon is constructed for large numbers of observations
and small class intervals, a smo'othed curve can be approximated that is
referred as frequency curve.
Cumulative Frequency Polygon (Ogive)

A cumulative frequency polygon or ogive is .obtained by plotting
cumulative frequencies of classes along Y-axis while upper class boundaries
along X-axis.
By: Rafaqat
Charts
It is the plotting data and showing results of a process over a period of time
(day, month, etc.)
Diagram
A diagram is a simplified and structured visual representation of concepts,
ideas, constructions, relations, statistical data, anatomy, etc. used in all
aspects of human activities to visualize and clarify the topic.
\"-a.,.. l: Introductory Statisdcs

6
II rs
{ A Quick Approach to Statistics with Questions and Answers
.I
I
f
i
1 Measures of Central Tendency
'
\
Central Tendency
The general level, characteristic, or typical value that is representative of the
majority of ca5es is referred as central or average value in general. The
tendency of observations to gather around the central. part of data is called
central tendency.
Measure of Central Tendency

By: Rafaqat
A measure of central tendency is a value at the center or middle of a data

set.
. Arithmetic Mean
It is a type of average (measure of central tendency), which is defined as the
sum of all the values in a set of numerical data divided by total number of
observations in that data se:.J This is the mast commonly used measure of
central tendency and is simply called M_ean. It is labeled as either
(lowercase Greek letter "mu") to denote a population mean or X (X-bar) to
denote a sample mean.
Weighted Mean
An average of means calculated by weighting each individual mean
according to the number of data points that made up that individual mean.
Geometric Mean
A mean of n objects that is computed by taking the n-th root of the product
of the n terms. A measure of the central tendency of a data set that
minimizes the effects of extreme values.
Harmonic Mean
It is reciprocal of the mean of reciprocal values.
Median
The median of a set of scores is the middle value when the scores are
arranged in order of increasing (cir decreasing) magnirude. The median is
often denoted by X (X-tilde ).
Chapter 1: Introductory Statistics 7
r
F Mode
The mode is the value that has the largest frequency in a data set. When two
scores occur with the same greatest frequency, each one is Mode and the
data set is called bimodal. When a data set has more than two. modes, it is
called multimodal.
Empirical Relation among Mean, Median, and Mode

There is an empirical relation between mean, median and mode that is
Mode = 3 Median - 2 Mean.
Quartile
A quartile is any of the 3 values which divide the sorted data set into 4
equal parts, so that each part represents I/4th of the sample or population. It
is denoted by Q; (i = I, 2, 3). The second quartile is, obviously, equal to
By: Rafaqat
median ..
Deciles
A decile is any of the 9 values which divide the sorted data set into I 0 equal
parts, so that each part represents Ill Otb of the sample or population. It is
denoted by D, (i = I, 2, .., 9). The 5th decile is, obviously, equal to median.
Percentile
A percentile is any of the 99 values which divide the sorted data set into ~ 00
eqwlJ parts. so that each parrrepr.esents 1/1 OOth of the sample or population.
It is denoted by P; (i = I, 2, ... , 99). The 50th percentile is, obviously, equal
to median.
Measures of Dispersion
Dispersion
Dispersion indicates variation while a descriptive measure that indicates the
amount of variation in a data set is called measure ofdispersion.
Range
The range is the length of the smallest interval which contains all the data. It
is defined to be the difference between the largest and the smallest value of
a data set.
Mean Deviation
By: Rafaqat
The absolute deviation of an element of a data set is the absolute difference

between that element and a given point (usually mean or median). The
average of such deviations is called mean deviation.
Quartile Deviation _
Quartile deviation is the half of'the difference between first and the third
quartile.
Variance
A measure of the variation shown by a set of observations and defined as
the mean of the squares deviations of all the observations from their mean.
It is usually denoted by d (sigma-square) for population and s2 for sample.
Standard Deviation
The positive square root of the variance (defined above).
Coefficient of Variation (CV)

It is used to describe the standard deviation relative to mean. It is expressed
in percentage. It allows us to compare variability of data sets with different
measurement units (such as. centimeters versus minutes).
Z-Score (Standard Score)

It is the number of standard deviations that .a given value x is above or
below the mean. It is simply referred as the standardized variable.
Chapter I: Introductory Statistics

'
Moments
A moment designates the power to which deviations are raised before
averaging them. Moments determine the shape and location of a
distribution.
Skewness
A distribution of data is symmetric if the left half of its histogram is roughly
a mirror image of its right half. The lack of symmetry ot departure from
symmetry is called skewness. A skewed distribution extends more to one
side than the other. If it has longer right tail, it is called positively skewed
distribution. if it has longer left tail, it is called negatively skewed
distribution.
It is important to note that
Mean = Median = Mode; Symmetric Distribution.
By: Rafaqat
Mean 2'.: Median ;:: Mode; Positively Skewed

Mean ::; Median ::; Mode; Negatively Skewed
A measure to describe the degree of skewness is called coefficient of
skewness .
.Kurtosis
Kurtosis descnl>es the extent to which a frequency distribution of scores is
bunched around the center or spread toward the endpoints. Simply
speaking, it measures the degree of peakedness or flatness of a unimodel
distribution.
A measure to describe the degree of kurtosis is called coefficient of
kurtosis.
A Ql!ick Approach to Statistics with Que~tions and Answers
Questions-Answers
Q.J What is the main objective of Statistics while .applying on
numerical data?
Ans. The main objective of Statistics is summarization of numerical
data.
Q.2 What is meant by interpretati{Jn of data?

Ans. Interpretation of the data means drawing of conclusions from the
analysis arid comparisons made with the data.
Q.3 Can accurate data be misused? Explain. .

Ans. Certainly, any treatment of accurate data that are not comparable
as if they were comparable is misused. Any example would be a
By: Rafaqat
comparison of the annual number of automobile. accidents in the

Punjab and Sindh. to determine which province has the more
careful or better.. driver. Also, presentation of only part of the facts
could lead to the misuse.
Q.4 Men have more auto accidents than women; therefore women
are better drivers than men. True or False? Explain.
Ans. This is false" because more men than women driVe cars and for the
same span of time, say one year, the total hours driven by men are
more than by the females; therefore, men would have more
accidents ev,en if they are equally good drivers. But if we take
average number of road accidents per average number of driven
hours for both the genders then the situation may be comparable.
Q.5 Why is the presentation of the data necessary?

Ans. Presentation of the data refers to a meaningful and orderly
assembly and display of data so that the significant attributes of
the data can be determined
Q.6 Naf111! three Pakistani Government agencies that are good source
of the data.
Ans. Some that might be named are, Federal Bureau of statistics,
Punjab Bureau of Statistics, Sindh Bureau of Statistics, NWFP
Bureau of Statistics, Balochistan Bureau of Statistics, Pakistan
Census Organization, Agriculture Census Organization etc.
Q. 7 Why should an. anafyst know the accuracy of the data he plans to
analyze? .
Ans. If the analyst does not know the accuracy of his data he cannot
trust any analysis and interpretation that are made with them.
Q.8 Why do we construct a histogram?

Ans. We construct histogram in order to assess the shape of the
frequency distribution (to aSsess the degree of skewness and
kurtosis)
Q.9 Discuss the advantages of the polygon over the histogram.

Ans. A polygon provides the true shape of a distribution better than a
histogram. A polygon is easier to construct and two polygons can
be compared on the same 'chart.
By: Rafaqat
Q.10 In what ways 11fretplmcy polygon is different from a histogram?

Ans. Most commonly representations of the frequency distribution take
-the fonn ofthe.freqilency polygon, which is a line graph instead
of the bar- type graph of the histogram. It sketches an outline. of
the data pattern more clearly.
Q.11 Olr "'""' kind of data, the use of Arithmetic mean is inost
milllble?
AIU. If there is less variation in data (likely to b.e homogeneous) and the
observations are equally weighted then Arithmetic Mean is the
most suitable measure of central tendency.. In other words,' if all
the observations have ~ame importance or there are no extreme
values (outliers), Arithmetic Mean is preferable.
Q.12 Give chief characteristics of arithmetic mean, media11, mode,

geometric mean and harmonic mean.
Ans. Some important characteristics of averages are:
Arithmetic Mean:
Each value in the group to be averaged directly I
influences the magnitude of the arithmetic mean.
The arithmetic mean is subject to algebraic manipulation;
for the number of values from which it was computed will
give their sum.
The arithmetic mean can be computed even if all that is

known is the total of a group offigures and the number of
figures in the group.
Median:
The median is a positional average, being the center
figure in an arrayed list of figures . . After it has been
determined, its value cannot be changed, changing values. '
of the other figures, unless the changes result in placing a
new figure at the center of the list.
The median cannot be treated algebraically like the
arithmetic mean, geometric mean and harmonic mean.
Mode:
The mode, being that figure which appears most often in a
By: Rafaqat
group, is not changed in value by changes in the ot}J.er

figures.
The mode cannot be treated algebraically like the
arithmetic mean, geometric mean and harmonic mean.
Geometric Mean:
All of the values being averaged directly affect the
magnitude of the geometric mean.
The geometric mean is subject to algebraic manipulation.
The geometric mean gives the large values less weight
than arithmetic mean; therefore it will be smaller than the
arithmetic mean of the same figures.
The geometric me.an will be zero if any of the values in the .
series to be averaged is zero.
When negative values are used, the geometric mean does
not exist.
/armonic Mean:
Each figure being averaged directly affects the magnitude
of the harmonic mean.
The harmonic mean gives less weight to the larger figures
that do the arithmetic and geometric means; therefore, it
will be smaller than these means when all three are
computed/ram ihe same figures.
The harmonic merm of a set of ratios may be
appropriaiely calculated if the numerators ofthe fractions
. ~ . _ _ _ _..,fr_o_m_w_h_ic_h_th_e_r_a_tr_o_s_w_e_r_e_c_o_m...p_u_te_d_a_r_e_t_h_e_s_a_m_e_
. ..._ _.
., 'r 1: Introductory Statistics 13
A Qukk Approach to Statistics with Questions and Answers
If any of the values of the variable is zero, the harmonic

mean cannot be determined
Q.13 Under what general conditions each of the five averages the most
appropriate to use for a given group offigures~
Ans. General conditions under which the five averages would be used
are;
Arithmetic mean used as an average when the group of
figures is quite homogeneous.
Median used as an average of data that are highly
skewed
Mode used when it is a highly predominate figure in a
group.
Geometric Mean used primarily as an average ofratios of
By: Rafaqat
chang~
Harmonic Mean used when the numerators of the
fractions used to compute the ratios were the same or
nearly so. 1
Q.14 Why cannot the arithmetic mean be computed from an open-end

frequency distribution?
Ans. An arithmetic mean cannot be computed from an open-end
frequency distribution because the midpoints of the open-end
classes cannot. be determined but which are needed for the
computation.
Q.15 . Why ah average computed from a frequency diStribution is not

exactly the $ame as that computed from the original figures used
to construct the distribution?
Ans. The individual values are lost in the groups therefore an averag
computed from a frequency distribution is an estimate of tho,
average from the same data in the ungrouped form. In this case.
we assume that only the mid-points have the relevant class
frequencies rather than original values so it makes difference in
the average calculation for grouped and ungrouped data. This
difference is called Grouping Error. However, in most cases the
estimates are accurate enough to be most useful.
Q.16 When computed from the same data, which of the following
averages will be the largest, second largest, and the smalli:st:
arithmetic mean, geometric mean, and harmonic mean? Whyr
Ans. The arithmetic mean would be the largest and the harmonic mean
would be smallest. The geometric mean would be the smaller than .
the arithmetic mean because it gives less weight to the larger
values. The harmonic mean, however, gives less weight to the
larger values than does the geometric mean.
Q.17 Why is the median usually a better average for a highly skewed
frequency distribution than the arithmetic mean?
Ans. The extremes in the highly skewed distribution will distort or bias
an arithmetic mean in .their direction. The median, however, begin
a positional average will not receive the same bias.
Q.18 In order to use the formula for the mode:

By: Rafaqat
Mode= L+ (f. - fo) xh

(/. - fo)+(f. + /,)
Why must the interval of the model class and the two adjacent
classes be the same?
Ans. Jn the.mode formula the h' represents the interval, which is
common to the model class, and the classes that immediately
precede and follow the model class.
Q.19 Ifyour chairman of the department asked you to plan alt evening
for the sons of the department employees and toldyoli''thtit the':
average age~ of the sons was 14, what more would yoli'wantto
know about the ages before you started planning?
Ans. One would need information on the variations in the ages of the
sons. If their ages all are in the range 13-15, the entertainmen't
plan would be entirely different than it would be if the ages were
distributedfrom ages 6 to 18..
Q.20 A man who stated that he manufactured lid for Jhe glass jars was
asked what size lids he manufactured. He replied, '1the typical
size". To which average was he referring?
Ans. This would be the mode. The ariihmetic mean size and median size
may not fit any glass jar.
Q.21 Could a man six feet tall drawn while crossing a (iver with an
average depth of two feet?
Ans. Of course, the question was asked to get a little humor into a
sometimes dry subject. The question should make a student realize
how misleading an average can be when nothing is known about
the values that makeup the average.
Q.11 In 1003, the mean (arithmetic) monthly income of the families in

a particular city was Rs. 10,900. The median family income was
Rs. 18,100. If you were planning a large sales campaign in that
particular city for the product that would be purchased by
families in the middle and upper-middle income groups, which of
the two averages would be more useful? Why?
Ans. The median would be because one knows that one-half of the
families had incomes of the Rs. J8,200 or more and would
therefore have some estimate of the size ofthe market. The mean of
By: Rafaqat
Rs. 10,900 cannot indicate a market size ai all because nothing is

known about the values that are averaged
Q.13 Why do we need measure of dispersion even when we have

measure of central tendency?
Ans. The measure of central tendency, such as mean, median, mode etc.,
do not reveal the whole picture of the distribution of a data set.
Two data sets with the same mean may have completely different
spreads. Therefore we need measure of dispersion, in addition to
measure of central tendency, in order to assess the clear picture of
the'data.
For example, consider two groups of students h<!Ving marks in a
class test as follows:
Group-I: J2, 5, JO, 8, 3, 22
Group-I/: JO, 8, 9, J J, J2
Both groups have equal mean marks i.e., JO but obviously, in
Group-I, there is more variation in marks. Jn this situation, just
mean is not enough to evaluate the group performance.
Q.14 For small sample size of 5 or less, which measure of dispersion is

preferable; Range or Standard Deviatio11?.
Ans. For small samples like ofsize 5, range and standard deviation give
the same interpretation but due to easy computations, Range is
preferred in such cases.
Q.25 Why is the measurement of the . dispersion computed for a

frequency distributiqn not equal to that computed for the original
figures from which the distribution was constructed?
Chapter/: Introductory Statistics 16
Ans. Due to same Grouping Error as discussed in answering for above

Q-15.
Q.26 When computed from the same data, which will be larger, the
mean deviation or the standard deviation? Why?
Ans. The standard deviation. Becaiise it gives more weight to the large
values.
Q.27 Which measure of the dispersion, is most widely used?

Ans.. The standard deviation is used more than any other measure
because of its application to the analysis ofthe sample data.
Q.28 How can we compare the variation of the two or more

distributions if: (i). the arithmetic means of the distribution are
By: Rafaqat
the same? (ii). if they are different?

Ans. When the arithmetic means ofthe two or more distributions are the
same, a comparison of absolute measures of dispersions. will
enable one to see which is the most varied If the arithmetic means
differ, measures ofrelative dispersion must be compared
Q.29 What are the measures of relative dispersion? How are they
used?
Ans. Measures of the relative dispersion are absolute measures of
dispersion expressed as a fraction or percent of some base, usually
the arithmetic mean. They are used to compare the variations of
two or more sets of the data that are expressed in the different
units or two or more sets of data expressed in the same units but
which have different arithmeti,c means.
Q.30 Define mesokurtic, platykurtic, and leptokurtlc.

Ans. Mesokurtic means middle peakedness (betweeR flat and highly
peaked, usually called normal), platykurtic means "flat topped"
and leptokurtic means highly peaked
Exercises
Exercise-Lt (MCQs')
Q. l Statistics is used in the situations of

(a) uncertainty
(b) dealing aggregate data
(c) variability
(d) All three (a), (b) & (c)
By: Rafaqat
Q.2 Indicate the following for what the type of the data described
below is nominal: .
(a) Team scores in a cricket match.
(b) Daily temperatures in degree Celsius.
(c) Room numbers in the Holiday Irmho'1. _
(d) Identification oftfie children who luJve chicken pox.
Q.3 Indicate the discrete variable for the variables given below:
(a) Batting average of the Pakistan cricket team.
i
(b) Number of children of each of 1,000 married graduates of
a public university.
(c) Average heights of the students of the lst year Class.
(d) Daily hours of the sunshine during the period from .
September 21 to. December 21.
Q.4 Indicate for which of the following data, the shape of the curve
will be normal?
(a) A computer printout shows the current checking account
balances for all the checking customers o.f the national
bank.
(b) Diagnostic reading test scores are tabulated for all the low
graders in a school of a distri.ct.
(c) a
Gifted students in creative writing class are given the
verbal subtest of an intelligent test.
(d) A nation wide mathematics exam is given to 1-,000
students at a college wi_th a selective admission policy.
A Quick Approacli to Statistics witli Questions and Answers
Q.5 Indicate for which of the following set of computations, the

distribution to be approximately positively skewed?
(a) Mean is 79.3, median is 75.4 and mode is 72.
(b) Mean is 25,6, median is 24.9 and mode is 24.
(c) Mean is 128.74, median is 12.68 and mode is 135.
(d) Mean is 50.3, median is 49.6 and mode is 50.
By: Rafaqat
Chapter I: Introductory Stalistics 19
A Quick Approach to StatiStics with Questions and Answers
Exercise 1.2 (True/False)

Read the following statements carefully and indiq1te which statement is
"True" or "False":
1. Any one who assembles numerical data is a Statistician.

2. As a general rule more detailed descriptions of data can be found
in the primary source than in a secondary source.
3. Discrete data can. be expressed only in. the whole numbers.
4. Continuous data are jtist ratio-scale data.
5. The clas_s width of a frequency distribution is of ~qua! size.
By: Rafaqat
6, The class intervals of the frequency distribution shoulet

ordinarily be equal to either IO or 5.
7. The more classes in the frequency distribution the better the data
are summarized.
8. Open-ended frequency distribution should always be avoided.
9. The sum of the class frequencies is equal to the number of
observations made.
1O. Relative frequencies can never be greater than I.
11. Cumulative frequencies. never decrease as the corresponding
observed values increases.
12. A frequency polygon touches the horizontal axis just once.
13. If we arrange the observations in a data set from highest to
lowest; the data point lying in the middle is the median of the
data set.
14. The value most often repeated in the data set is called the
arithmetic mean.
15. Extreme values in a data set have a strong effect on the median.
16; The difference between the highest and the lowest observations
in.a data set, called the quartile range.
17. The range is more stable measure of variability than the inter-
quartile range .
. 18. The median is less influenced by chance than the mode because
the median take into account the entire distribution.
19. In a symmetric unimodel distribution, the arithmetic mean is
equal to the median.
20. The standard deviation is the square of the variance.
21. If the data of an experiment is measured in meters, then the
By: Rafaqat
standard deviation is reported as meters squared.

22. The variance, like the standard deviation, takes into account
every observation in the data set.
23. The coefficient ofVariatioQis an absolute measure of dispersion.
24. . When calculating the average rate of debt expansion for a
company, the correct mean to use is the arithmetic mean.
25. The geometric mean is an average.
26. When there are a few extremely large items in the series being
averaged the .geometric mean will be . the larger than the
arithmetic mean,
27. The median can be estimated from a cumulative frequency
curve.
28. The reciprocal of the harmonic mean is equal to the arithmetic
mean.
29.. In symmetrical distribution, the mean, median, and mode are
equal; provided that the mode is a single number.
30. .The 50th percentile is equal to the median only in the
symmetrical samples.
Chapter I: Introductory statistics 21
31. For a positive skewed distribution, the mode must be smaller

than the median.
32. The quartile deviation is a measure of a scatter of the first half
- the data.
33. A leptokurtic curve is more skewed than a platykurtic curve: .
34. The mean deviation cannot be computed fr.om an open-ended
frequency distribution.
35. One cannot"determine how much a distribtition is varied by its
standard deviation alone.
36. The coefficient of variation is a useful measure for comparing
By: Rafaqat
degree of skewness in two or more frequency distributions.

37. Histograms are better than polygons for comparing two
frequency distributions on the same chart.
38. The frequency polygon is a special type;;ofthe
,.. time series line
chart.
39. Charts should usually be higher than they are wide.
40. Diagrams are less accurate than tabies.
http://stat9943.blogspot.com i
t
111111 Chapter 2
Basic Probability Theory
B~sics in Probability
By: Rafaqat
Experiment . .
The process of obtaining an observation is called an experiment. Hitting a
target, checking the boiling point of a liquid, taking examination for a
student, conducting interviews for some jobs, tossing of a coin, rolling of a
die, hitting a ball of a batsman, sale of so)lle products, chemical reaction of
elements, are few examples of experiments.
Trial
A single performance of an experiment is called trial. If a batsman plays a
single ball, if a bowler bowls, if a student solves a single question, single
rolling of a die, all these are the examples of a trial.
Outcome
An outcome is the result of an experiment. Each possible distinct result of
an experiment is refei:red as outcome. Hitting or not hitting a target, making
some scores, leaving a ball or being out for a batsman, boiling of water on
l 00C are the examples of outcomes relevant to the experiments discussed
above.
Random Experiment
An expttriment is called random experiment if its outcomes cannot be
predicted in advance even if it is performed under similar conditions. Any
random experiment has the following properties:
(i) It has at least two outcomes.
(ii) The number of all possible outcomes is known in advance.
(iii) It can be repeated any number of times under similar conditions.
Chapter 2: Basic Probability Theory 23
Among above exampl~s of experiments, hitting a target, ta.king examination

for a student, tossing of a coin, rolli.ng of a die, hitting a ball of a batsman,
sale of some product are examples of Random Experiments while checking
the boiling point of a liquid and chemical reaction of elements 'Nill be
.nonrandom experiments.
Sample Space
The s~t of all possible outcomes of a random experiment is called a sample
space. It is usually denoted by Q (omega).
In the random experiment in which a student takes a examination, suppose
the result of examination can be in form of grades, 'A', 'B', 'C', 'D' and 'F'
then Q = {'A',"B', 'C',,'D','F'}..-
In rolling a die, Q = { l, 2, 3, 4, 5, 6}.
By: Rafaqat
Event
Any subset of sample space is called event. It is a coflection of outcomes of
an experiment. Events may be either simple or composite. Formally, if an
event consists of single sample point of a sample space, it is called
elementary or simple event and in case of two or more sample points, it is
referred as composite event.
Simple Events: A student passes an examination, a batsman makes shot for
six, a die show a number 2, etc.
Composite Event: A motor accident for rash driving and failure of brakes, a
ball results in I run and a run-out during an over in a cricket inatch, et-c.
Event Space
A set of all events relevant to a sample space is called event space. It is
usually denoted by ;t
Equally Likely Events

Two events A 1 and A 2 relevant to a sample space are said to be equally
likely if their probabilities of occurrences are equal. In other words, if e\ ent
A 1 is as likely to occur as Ai does then A 1 and Ai are said to be equally likely
events.
Mutually .Exclusive or Disjoint Events

. Two events A 1 and Ai relevant to a samirle space are said to be mutually
exclusive or disjoint if they cannot occur simultaneously. Thus tossing of a
coin is assoCiated with mutually exclusive events. If the event 'head' occurs,
the event 'tail' eannot occur at the same time. In above example of grading
Chap~er 2: Basic Ptobability Theory 24
A Quick Approach to Statistics with Questions anti Answers
of a student in an examination, if all the five grades refer to five different

events then they can not occur simultaneously 11s a student can take,
obviously any one of the grade, so these events are also disjoint.
Mathematically, if A 1 and Ai are two mutually exclusive events then
P(A 1 nA 2 )=0.
Collectively Exhaustive Events

The events A 1 A 2, , Ak are said to be collectively exhaustive if all A;'s
( i = 1, 2, ... , k) are Disjoint and A 1 u A2 u ... Ak = n.
In above example of grades, all the five events are collectively exhaustive.
Complementary Events
Complementary event for an event A is the event that A does not occur. For
event A it is denoted by A1 or Ac. For example, passing of a student is
By: Rafaqat
complementary event to the failing of that student.
Independent Events
Two events A 1 and Ai relevant to a sample space are said to be statistically
independent ifthe occurrence of A 1 does not affect the probability of
occurrence or non-occurrence of Ai.'
Symbolically,
=
P(A 1 nA 2) P(A 1). P(Ai).
The passing (or failing) of one student is statistically independent to the
passing or failing of other student(s) in the same examination, score on .a
current ball is statistically independent to the result of previous ball in a
cricket match, are the examples of statistically independent events.
Probability
Probability may be defined as the likelihood of the occurrence of an event.
A probability provides a quantitative description of the likely occurrence of
a particular event. In other words, it is a numerical measure of uncertainty.
Probability is conventionally expressed on a scale from 0 to I; a rare event
has a probability close to 0, a very common event has a probability close to
1.
Subjective Probability
A subjective probability des.crioes an individual's personal judgment about
how likely a particular event is to occur~ It is not based on any precise
computations but is often a reasonable assessment by a knowledgeable
person. A person's subjective probability of an event describes his/her
degree of belief in the event. For example, a cricket expert says that there
are more than 80% chances that the team A will win the tournament. A
planning minister guesses that at least 3/4 villages of the country will be
supplied electricity by the end of next year.
Objective Probability
A probability that can be established theoretically or from historical data:
The objective probability has following main approaches to define
Probability:
(i) Classical (Priori) Definition of Probability
(ii) Relative -Frequency Definition of Probability
(iii) Axiomatic Definition of Probability
By: Rafaqat
Classical (Priori) Definition of Probability

If an experiment can produce n different mutually exclusive results all of
which are equally likely, and if m of these results are considered favorable
(or result in event A), then the probability of a favorable result (or the
probability of event A) is min.
In tossing of coin, the probability of occurring 'head' or 'tail' and the
probability of selection a female student among a group of 20 male and 30
female students, follow classical definition of probability.
Relative-Frequency Definition of Probability

This definition of probability states that if in n tri.als, an event A occurs nA
times, its probability, P(A) isapproximately nA/n:
P(A)-= nA!n,
provided that n is sufficiently large and the ratio nA/n is nearly const~mt as n
increases.
The probability of having number 6 on an irregular die, the probability of
survival of a 40 years old Hepatitis C patient, the probability for wining of a
tournament by a team, are the examples 'of relative frequency definition of
probability.
Axiomatic Definition of Probability

Let n be the Sample Space, the probability of an event A is, by definition, a
number P(A) assigned to A. This number satisfies the following three
axioms:
(i) P(A) is a non-negative number; P(A) ~ 0.
(ii) Probability of the event n (sure or certain event) is equal
to l;
J'(O.) = 1.
A Quick Approach to Statistics witli Questions and Answers
(iii) If two events A1 and Ai have no common element (i.e.

Mutually Exclusive) then
P(Ai u Ai)= P(Ai) + P(A2)
Conditional Probability ..
In many situations, once more information becomes available; we are able
to revise our estimates for the probability of further outcomes or events
happening. For example, 'suppose you go out for lunch at the same place
and time every Friday and you are served lunch within 15 minutes with
probability 0.9. How.ever, given that you notice that the restaurant is
exceptionally .busy, the probability of being served lunch within 15 minutes
may reduce to 0.7, This is the conditional probability of being served lunch
within 15 minutes given that the restaurant is exceptionally busy. The effect
of such information is to reduce the sample space by exc1uding some
By: Rafaqat
outcomes as being impossible which before receiving the information were

believed possible.
Formally, the conditional probability of A given Bis the probability ofevent
A occurrin_g, given that event B has already occurred. It can be found by
dividing the probability of events A and B both occurring by the probability
of event B as shown below
P( A I B) = P( A(""\ B) .
P(B)
It is obvious to note that if A and B ~re statistically independent then
P(A I B) = P(A). .
Some Probability Rules
AdditiQn Rule of Probability

If A 1 and Ai be any two events relevant to a sample space n, then
P(A I u Ai ) = P(A 1) + P(Ai) - P(A I (J Ai ) .
In words, the probability that either event A 1or event Ai or both events occur
equals the probability that event A 1 occurs plus the probability that event Ai
occurs minus the probability that bo~h occur.
Generally, for n events:
. ,.
P(LJ A,)=
iI
I P(A.)- L P(A,r.A )+
;al 1<1
1 L P(A,ri A1 riAt)- ... +(-t)"i P(f) A.)
i<j<lt i..al
By: Rafaqat
Multiplication Rule of Probability

For any two events A 1 and Ai relevant to a sample space n, the
multiplication rule is a result used to determine the probability that two
events, A 1 and Ai. both occur. This rule follows the definition of conditional
probability as
P(Ai 11 Ai)= P(Ai).P(Ai I Ai) .
Generally, for n events: .
n
P(Ai 11 Airi ... 11 A.)= P(Ai).P(Ai I Ai).P(AJ I Ai Ai) .... 11 P(A. I Ai 11Az11 . 11
Law of Total Probability

Suppose, A 1 and Ai be mutually exclusive and exhaustive events with non-
zero probabilities then for any event B (with non-zero probability)
P(B) = P_(Ai).P(B I Ai)+ P(Az).P(B I Ai).
P(B) =I P(A.).P(B I A,).
iI
1 ,,
1
Baye's Theorem
Suppose, A 1 and Ai be mutually exClusive and exhaustive events with non-
zero probabilities then for any event B (with non-zero probability)
P(Ai I B) = P(A,).P(B I A)
P(AJP(B I A)+ P(Ai).P(B I Ai)
28 f
Chapter 2: Basic ProbabUity Theory
I
I-
I
I
.P(A;I B) P(A.).P(B I A.) ' (i =J,2, ...,n)

I
iI
P(A.).P( B I AJ
Counting Rules
Rule of Multiplication
If there are K procedures and ith procedure may be performed inn; ways (i
=I, 2, ... , k) then all the k pr0cedures may be performed in n 1 x n 2 x. ... x ni.
ways. For example, a person has 3 different pairs of shoes and 4 differ~nt
pairs of socks then he may use all of these pairs in J x .4 = 12 different
ways.
By: Rafaqat
Rule of Addition .
If there are K procedures and ith procedure may be performed in n; ways (i
= l, 2, ... , k) then the number of ways in. which one can perform procedure
I or procedure 2 ... or pr9cedure k given by n 1 + n 2 + ... + nk (assuming
that one procedure can be performed one time or no two procedures can be
performed together).
Suppose, a group of students is planning a trip and thinking about either bus
"'\ or train to use for that. If there are 3 different routes available when using
bus for the trip and 2 different routes for train then there are 3 + 2 = 5 routes
available for that trip. '
Permutations
A permutation of n different objects -taking rat a time {O Sr Sn)
I.
P, = n. . In permutations, order of objects is important or meaningful.
~-~ .
For example, if one wants to calculate the ways in which 6 persons may be
=
seated bn a bench having a capacity of 4 seats then in this case, n 6, r"= 4
. 61
and answer is 6 P, =--- = 6 x 5 x 4 x 3 = 360.
(6-4)!
Combinations
A combination of n different objects taking rat a time {O Sr Sn)
c, =(n) = . n! . In .combinations, order of objects is not meaningful.
r r!(n-r)! .
To differentiate between the case of permutation and combination, we
coi:isider the example that out of 7 Statisticians a committee is to be formed
C/ulplel' 2: Basic Probability Theory 29
1
of 4 then in
7
c. = ( 4 ) = 4!(7-4)!
,:;J._ . -
35 different ways the committee can
.
:be formed. Now, if we specify the positions of committee members like,
one president, one secretitry, one treasurer and one speaker then this will be'
a case of permutation because here the positions (order) are meaningful and
the number of ways in which such committee can be constituted is
7
p = _ 7_!- = 840.
(7-4)!
By: Rafaqat
Chapter 2: Basic Probabllity Theory 30
Questions.-Answers
Q.1 Why do we use probability theory?

Ans. We use this theory for decision making in the situations of
uncertainty so it becomes useful in nearly every phase of the
human activity. There are uncertainties in every sphere of-life;
whenever we turn to predict the outcome of a cricke'tl hockey
match or a political election; dig for oil search for uranium
deposits; experiment with the space ships; and even when we
choose a school for our children, or buy a dress shirt. Much of
usefulness ofthe Statistics arises from the ability to measure and to
control risk and errors in the inferential and decision problem:
By: Rafaqat
This ability is embodied in the theory of the probability. Since the

probability theory is at once the language and the measure of
uncertain conditions/atmosphere. Thus, probability theory is
properly considered as the foundation of statistics. Indeed no
analytical statistics can be developed without this theory.
Q.2 Discuss the difference between the objective and subjective

interpretation of the probability.
Ans. The objective interpretation of the probability is that a probability
can be determined as a prior. A probability can also be determined
on the basis of experience. For example, "the probability of a
defective item can be determined as after observing that JO out of
sample of 100 p~rts selected at random. On the other hand, -the
subjective interpretation of the probability is one of the personal
beliefs. For example, if a manager strongly believes that the level
of demand for a white-co/or Suzuki car will be high, he might
assign the probability of0.90 to this level.
Q.3 What is the probability that a professor will meet his next class?
Is this a priori probability or posterior probability?.
Ans. This answer will depend on our previous experience that is, on the
frequency. that professor has met his previous classes. This is the
example ofposterior probability.
Q.4 . With respect to set theory, explain what is mean by ln.tersection .

and Union of two sets? . .
Ans. The Intersection of the elements of the sets A and B consists of
elements common to both A and fJ.
Cb'1ptu 2: Basic Probability Theory 31
A Quick Approach to Statisiics with Questions and Answers
The Union of the elements of the set A and set B consists of the
elements that belong to set A or set B or both A and B.
Q.5. Can two mutually exclusive events be independent?

Ans. No. Two mutually exclusive events are also dependent events so
they can .never be independent.
Q.6 What is the difference between Sample Space. and Event Space?
Ans. Sample space is the set of all possible outcomes of a random
experime~t whjle the event space is ihe set of all events dssodated
with a sa~ple space. .
't .
By: Rafaqat
Q. 7 What is ~h~)o1"t probability?_ . .

Ans. Joint probability is the probability of the occurrence of events of
interest~ c~hcurr;ently.
Q.8 Why conditional probability is different from J:omrnon

probability r .
Ans. Cf!nditiona/, probability is a probability assigned to an eveni when
a given set of Information related to. that event of interest so it is, .
obviously, differentfrom the common probability.
Q.9 Jn what sense do we say that a conditional probability is defined

on reduced sample space?
Ans. Sometimes, we are provided with the additional information
relating to the outcomes of the random experiment for this reason
sample space is cbanged and reduced The probability ofany event
derived from such a reduced sample space is ca/Jed conditional
probability.
Q.10 What is the tnajor put:pose (or advantag~) of 11,sing the Baye's
rule in decision making problem?
Ans. The advantage of this approa.ch is that it allows the initial
forecasts in the form of probabilities to be revised upward or
downward as related additional inforrrtation b~comes available.
I
Q.11 What is the weakness o/Baye's approach in .decision !"a~lng?

Ans. The weakness in this approach is that the prior pr.obabilities
represent the opinion ofp panel of experts. Anotherpanel would
probably have given a different set of initial probabilities.-That is,
there is subjective judgment involved in determining the prior
probabilities.
Cb.apter 2: Basic Probabilitj http://stat9943.blogspot.com

Tluory 32
~ A Quick Approach to Statistics with Questions and Answers
Q.11 What is the probability that ti student will pass a course in

probability? How did you arrive at your answer?
Ans: Any value between 0 and I is acceptable here. The student may use
the ratio ofthose who passed this probability course to all of those
who took the course in the last few years as estimates.
Q.13 What Is the probability. that the price of oil will higher next year
1 than this year? Why? .
f Ans. Again, any value between 0 and I is acceptable here. Judging.from
recent past experience, the probability could be q~ite high such as .
0.8, or even 1.
By: Rafaqat
( Q.14 An economist assessed the probability of continued unfavourable

balance of trade for the Pakistan as 0;80, the probability of a
rec~sion with in the next 3-year as 0.50, and the probability of
.both of these events as 0.40. Does the economist believe that
occurrence a recession is independent of an tinfavorable balance
of trade? Explain.
. Ans. Yes, because the probability ofboth events occurring is assessed as
equal to the product of the separate marginal probabilities;
P (A and B) = P (A). P (B) = 0.80 x 0.50 = 0.40.
Q.15 Weather forecaster often gives probability for various events.

Two such events are
/ (i) 1
The probability of measurable precipitation on
a day in June in your areas.
(ii) T~e probability that it will rain tomorrow in
yourarea,-
Would you expect two different forecasters to give the same or
different probabilides for (i) and (ii)? .
Ans. The probability of measurable precipitation on a day in June
would be established from historical weather records. Two
forecasters would give the same probability if they have consulted
the same historical record.
The probabilit)i of rain tom.arrow would be based on a number of
indicators'and twoforecasters might give differing weights to the
different indicators. The probability .is thus in part subjective and
two forecasters could give different probabilities.
Chapter 2: Basic Probability Tlteory

33
~Exercises
Exercise 2.1 (MCQs')
Q. I A probability of 1 represents
(a) Impossibility
(b) An improbable event
(c) A 50 - 50 chance
( d) Certainty
By: Rafaqat
Q.2 Three coins are tossed. What is probability that there will all be
heads?
(a) 1/8
(b) 114
(c) 1/3
(d) 3/2
Q.3 A cricket team captain wins the toss for three consecutive matches.
What is the probability that he will call correctly for the fourth
match?
(a) 1116
(b) 1/8
(c) 1/4
(d) 1/2
Q.4 Which one is not the characteristic of a random experiment:

a) It has at least .two outcomes .
. b) The number of all possible outcomes are not known in
advance.
c) The outcomes are not predictable in advance.
d) It can be repeated any number of times under similar
conditions.
Q.5 A and B are two mutually exclusive events. The probability of A

happening is 1/4. The probability of B happening is 1/3. The .
probability of neither A nor B happening is:?
(a) . 5/12
(b) 1/2

(c) 314
(d) 11112-
Q.6 A andB are two independent events. The probability of A is 114

and Bis 1/3. The neither probability of A nor Bis.
(a) 5112
(b) ll.3
(c) 3/4
(d) l l/12
Q.7 The probability of an event happening is l/3. The probability of it

not happening is?
(a) - 12
(b) - 0
By: Rafaqat
(c) 2/3
(d) 3
Q.8 If a letter is chosen .at random from the l 0 letters of the word
STATISTICS, what is probabiiity that it is a vowel?
(a) 0.20
(b) 0.23
-(c) 0.30
(d) 0.40
Q.9 Given the following table, what is the probability of selecting a

female university graduate student from _this group?
College graduate University graduate Total
Male 100 36 136
Female 120 44 164
Total 220 80 300
(a) 0.120
(b) 0.268
(c) 0.265
(d) 0:147
-
Q.10 Subjective probabilities are assigned to the events A arig B which
together comprise a sample space. Which of the following
probability statements is not valid?
a) P(A) = 0.7, P(B) ,;0.5, P(A and B) == 0.2
b) P(A) =0.9, P(B) =0:7, P(A and B) = 0.6
c) P(A) =0.4, P(B) "'.'0.3 , P(A and B) = 0.2
d) P(A) =0.5, P(B) =0.5 , P(A and B) = 0.0
A Quick Approacll to Statistics witll Questions and Answers
Q.11 A personal manager selects an applicant at random from a large

group for an interview. The probability of the applicant being male
is 0.60. The probability of selecting an adult -is 0.70. The
probability of selecting an adult male is 0.45. Given that a male is
selected, the probability that he is an adult is;
(a) 0.27.
(b) 0.42
(c) 0.64
(d) 0.75
Q.12 Which of the following is collection of all mutually exclusive ,

events representing a card randomly selected from a deck of
By: Rafaqat
ordinary playing cards?

(a) King, Queen, Face.
(b) . Heart, Diamond, Black, Red.
(c) JO, 7,jack.
(d) IO, red.
Q.13 Which of the following are collectively exhaustive events

representing a card randomly selected from a deck of ordinary
playing cards?
(a) Club, Spade, Heart, Diamond
(b) 7 of Spades, Spade, Club Red
(c) Face, 7, 9, lOJack
(d) Red, Spad~s
Q.14 A conditional probability might be found in which of the following

ways?
(a) Multiplying together two conditional probabilities.
(b) Dividing a joint probability by the given event's
probability.
(c) Applying the basic concepts of probability to the portion
of the event space for which the.condition holds.
(d) Finding the run frequency of times that . the. event in
question occurs out of all those times when the given
event occurs.
Q.15 Indicate in which one of the f~llowing situations the events A and
B are independent.
(a) P (A) = 0~6; P (B) = 0.3; P (AIB) = 0.6
(b) P (A) = 0.3;,I: (B) =:= 0.7; P (BIA)= 0.3
(c) P(A) = 0.3 }i>(B). = 0.3 ; P(A and B) = 0.2
(d) A and Bare mutually exclusive; P (A)= 0.1; P (B) =0.2
Q.16 Which one of the following statements is false?

. (a) P(AIB) = 0.3,_ since A and Bare independent and P(A) =
0.3
(b) P(AIB) = 0.3, since A and Bare mutually exclusive. ,
(c) P(BIA) = 0.4, since A and Bare independent and P(B) =
. 0.4
(d) P(AlB) = 0, since P(A and B)=O and P(B) = 0.7
Q.17 A marginal probability might be found by apy but which one of the
following?
(a) Adding together appropriate joint probabilities.
(b) Subtracting or the sum of several marginal probabilities
By: Rafaqat
from 1.
( c) Dividing the size of the appropriate event set by the
number of possible equally likely elementary events.
( d) Multiplying together all probabilities in the same column
or row.
Q.18 Which one of the following statement is not true?
(a) Mutually excl~sive events are statistically dependent.
(b) Complementary ev~nts have probabilities that sum to l.
(c) Opposite events, are statistically independent.
(d) An experiment's elementary events are collectively
exhaustive and mutually exclusive.
Q.19 The number of ways to select 2 persons from 6, ignoring order of

selection;
(a). 64
(b) 15
(c) 36
(d) 12
Q.20 A fair coin is tossed 50 times, the expe~ted number of heads are;
(a) 100
(b) 50
(c) 15
(d) None of these.

Exercise 2:2 (True/False)

Read the following statements carefully and indicate which statement is
"True" or "False":
I. A 50-50 chance of the rain means that the probability of the rain
is 112.
2. For a random experiment, all the outcomes are known in
advance.
3. A or B is_ an event occurring whenever A-occurs alone, B occurs
alone, or the both A and B do not occur.
By: Rafaqat
4. If A and B is impossible, then the two events A, B are mutually

exclusive.
5. If P (A and, B) = 0, then the two events; A and B must be
collectively exhaustive.
6. Suppose two events A and B are mutually exclusive and also
collectively exhaustive, this indicate that P (A) = 1- P (B).

7. Suppose that 'P (A) = 0.8, P (B) = 0.2, and P (A and B) = 0.16.
Then, A is independent of B.
8. A conditional probability is equal to the unconditional
probability for - the event A given B whenever. A and B are
mutually exclusive.
9. A marginal probability will always be equal to the sum of two or
more joint probabilities.
10. 'The event _space is the set of all possible sample points of a
sample space. )
11. The event space corresponding to the union of the events A, B

represents all elementary events of A and of B, but not events
common to both.
12. If the probability of success is 0.4 than the probability of failure

is 0.6.
13. Joint events ar~ those that are not mutually exclusive.
14. If three events, A, B, and Care independent, the probability that
A, B, and C will happen at the same time is P (A) .P (B). P(C).
15. If A and B are mutually exclusive events then the P (A or B)
equals the P (A) +-P (B)- P (An Bj.
16. . When events A and ev~nt Bare independent, P (BIA)= P (AIB)
17. The addition rule concerning probabilities appiies to independent
By: Rafaqat
events.
18. Probability is a number between 0 and I, exclusive.
19. If the two events, A and B are mutually excfusive then the
probability that eithe~ one or the other will occur is P(A).P,(B).
20. A probability can be certainty.
21. In a classical probability, we can determine a prior probability
based on a logical reasoning before any experiments t&ke place.
22. An unconditional probability is also known as a marginal .
proba.bility.
23. A subjective probability may be nothing more than an educated
guess ..
24. When using the relative frequency approach, probability figures
become less accurate for large number of observations.
25. The relative frequency approach to probability will provide
correct statistical probabilities after I 00 trials ..
26. A and B are independent events if P(A/B) = P(B).
27. Symbolically, a marginal probability is P(AB).
28. If A and Bare independent events then P(AnB) #:- P(A).P(B).
A Qui!:k Approach to Statistics with Questions and Answers
29. Using Baye's theorem, we may develop revised probabilities

based on new information; these revised probabilities are also
known as.posterior pr-0babilities.
'
30. Classical probability assumes that each of the possible outcomes
of an experiment is independent to the other.
By: Rafaqat
'.

Chapter 3
1111111
Random Variables
Random Variable . .
By: Rafaqat
A random variable is a real-valued function that associates a unique

numerical value with every outcome of a random experiment: Let n be a
sample space on which probabi.Jity function defined. Let X be a real valued
function defined on sample space, such that x transforms outcomes of the
n into points on th.e real line. Then Xis said to be random variJlble:
The outcome of an experiment need not be a number, for example, the
outcome when a coin is tossed can be 'head' or 'tail'. However, we often
want to represent outcomes as numbers.
Discrete Random Variable

A discrete random variable is one which may take on only a countable
number of distinct values such as 0, I, 2, 3, .... Discrete random variables
are usually (but not necessarily) counts. I.fa random variable can take only a
'finite number of distinct values; then it must be discrete. Examples of
discrete random variables include .the number of children in a family, the
number of patients in a doctor's surgery, the mi'"mber of defective light bulbs
in a box often.
Continuous Random Variable

A continuous random variable is a randbm variable whose possible values
from a continuous data set usually some interval of .real numbers.
Continuous random variables are usually measurements. Examples include
height, weight, the amount of sugar in an orange, the time required to run a
mile..
Chapter~: Random Variables 41
Probability Function
A probability function is a real valued function defiried on the class of all
subsets of the sample space Q; the value that is associated with a subset A is
denoted by P(A). The assignments of probability must satisfy the following
three axioms:
(i) P(Q) =I
(ii) P(A) ~ 0
(iii) Jf A; ( i = 1, 2, ... ) is a sequence of mutually exclusive events
then
co co
P(U A,)= L. P(A;)
i=I i=I
Probability Distribution
By: Rafaqat
A table listing all possible values that a random variable can take on
together with the associated probabilities is called probability distribution.
The probability distributio~ of a discrete random- variable is a list of
probabilities associated with each of its possible values. It is also sometimes
called the probability function or the probability mass function.
More formally, the probability distribution of a discrete random variable X
is a function which gives the probability P(x;) that the random variable
equals x;, for each value x,:
P(x;) = P(X=x,)
It satisfies the following conditions:
(i) 0::; P(~.)::; I
(ii) 2: P(x) =1
Probabi_lity Density Function (pdf)
The probability density .function (pdt) of a continuous random variable is a
function which can be integrated to obtain the probability that the random
variable takes a value in a given interval.
More formally, the probability density function, ft..x), of a continuous
random variable Xis the derivative of the cumulative distribution function
F(x) (defined next):
d
f(x) = dx F(x).
lfft..x) is a probability density function then it must obey two conditions:
(i) That the total probability for all possible values of the
continuous random. variable Xis 1, i.e;
I f(x)dx = i
Cllaptu J: Random Variables 42
'
(ii) That the probabiHty density function can never be negative:

fl..x) > 0 for all x.
Cumulative Distribution Function (CDF)

All random variables (discrete and continuous) have a cumulative
distribution function. It is a function giving the probability that the random
variable Xis less than or equal to x, for every value x.
Formally, the cumulative distribution function F(x) is defined to be:
F(x) =P(X :;:; x)
For a diScrete random. variable, the cumulative distribution function is found
by summing up the probabilities as;
F(x) = P(X:;:; x) = f P(X = x)
By: Rafaqat
For a continuous random variable, the cumulative distribution function is

the integral of its probability density function;
- F(x) = P(X:;:; x)_= JJ(x)dx
A cdfhas the following properties:

(i) F(+oo)= I andF(-oo)=O
(ii) F(x) is non-decreasing function of X, i.e., F(x 1) 5:. F(x2) ifx 1 $.X2
(iii) F(x) is continuous at least on the right of each x
Expected Value
The expected value (or population mean) of a random variable indicates its
average or central value. It is a useful summary value (a .number) of the
variable's distribution. Stating the expected value gives a general impression
of the behavior of some random variable without giving full details of its
probability distribution (if it is discrete) or its probability density function ..
(if it is continuous). The exp~cted value of a random variable X is
symbolized by E(X) or.
If Xis a discrete random variable with possible values Xi. x 2, x 3, , x 0 , and
P(x;) denotes P(X =x1), then the expected value of Xis defined by:
= E(X) = L;x,P(x,) ',provided that series are convergent.
If Xis a continuous random variable with probability density functionj(x),
then the expected value of Xis defined by:
J
= E(X) = xf(x)di.
It is to be noted that
Var (X) = E(X1-)-{E(X)} 2
Also
Chapter 3: Random Variables 43
A Q.ulck ApprOllCb to Statistics with Questions and Answers
E(X +C) = + C,
where C is any constant.
Independent Random Variables

Two random variables X and Y (say), are said to be independent if and only
ifthe value of Xhas no influence on the value of Yand vice versa.
Symbolically, X and Y are said to be independent if
j{x,y) = g(x).h(y),
where g(x) and h(y), are the marginal density functions of the random
variables X and Y.respectively and j{x,y) is the- joint density function of the
both variables'.
Moment-Generating Functfon (m.g.t)

By: Rafaqat
The moment-generating function of a random variable Xis

Mx (t) = E(e''), t e9l,
wherever this expectation exists. The moment-generating function generates
the mo.ments ~f the probability distribution, and thus uniquely defines the
distribution of the random variable.
Provided the moment-generating fum:tion exists in an interval art>und t = 0,
the r-th moment is given by
E(X') = M~> (0) = d', Mx (t).

dt t-0
I.
If X has a continuous probability density function j{x) then the moment
generating function is giv~n by
e
Mx(t) = Je"'f(x)dx
.. . . t2x2 .
=I (I +tx+-+
-- 2!
.. ) J<x>dx
. '
.
I
/2 ' 2
= I +fI+~+,
where 1
1
is the i-th moment about origin.
Cumulant-Generating FunCtion (c.g.t)

The cumulant-generating function is the logarithm of the moment-
generating function.
. Chapter 3: Random Variables 44 .
A Quick Approach to Statistics with. Questions and Answers
K(t) =In M 1 (t) ."'

. 2 '\. ]
t \. t
= K,t+ Ki2! + K13! +,
We have following relations between cumulants and moments:
K,=',
., ,2
Ki= 2- I = 2
. , 3 , , 2 ' 3
om K J = J - I 2 + I = 3
the
Characteristic Function
By: Rafaqat
The characteristic -function of any random variable completely defines its

probability distribution. On the real line it is ,given by the following
formula, where Xis any random variable with the distribution.f{x):
ttes
the </Jr (t) = E(e"'),
where t is a real number.
: 0, The r-th moment about origin can be found as:
E(X'). =-{ [ -rp,.

d' (i) ] .
. dt' .
, 1=0
ent Chebyshev's Inequality

Chebyshev's inequality (also known as . Tchebysheffs inequality,
.Chebyshev's theorem, or the Bienayme-Chebyshev inequality), s\ates that in
any data sample or probability distribution, nearly all the values ~re close to
the mean value, and provides a quantitative description of "nearly aW and
"close to". For example, no more than 1/4 of the values are more than 2
standard deviations away from the mean, no more than l/9 are more than 3
standard deviations away, no mqre than 1125 ~re more than 5 standard
deviations away, and so on.
Generally, let X be a random variable with expected value and fini
variance a 2 Then for any real numb.erk> 0, the probability that av~ of
X falls within k~standf!rd deviation ofthe mean is at least I - k 2 j/;~ ..

1
nt- I
Pr( - kO' < X < + ka) ~ I - -
k2
44. Chapter 3: Random Variables 45
A Qttkk Approach to Statistics with Questions and Answers
Questions-Answers
Q.J If a random variable (r.v) is a real valued function defined on
sample space, what is domain and range for r. v.?
Ans. The domain is sample space whr1e the range is real line.
Q.2 If X is a r. v then whether any function of X, say, g(X) is also a

r.v.?
Ans. Yes, a function ofa r. v is also a r. v.
Q.3 Let X denotes the number of persons in a household selected at

random, then what type of ran do in variable Xis ?
Ans. Discrete.
By: Rafaqat
Q.4 Give tire examples/or discrete and continuous random variable.

Ans. Discrete: Number o/patients, daily coming in a clinic; Number on
roadside accidents per day on a high way, etc.
Continuous: Weight, height, age, waiting time for a bus on a bus
stop, etc.
Q.5 The CDF of a continuous variable Tis definetl as follows,

0 ift<O
F(t)= t/l2 if O:St:S12
{
1 if I> 12
Determine,
a) P(T56) b) P(3 5T 54.5) c) P(-005 T 5 3. 75)
d) P(T> 9.45) e) P(T 515) j) P(T=tt)
Ans. a) P(f.56) = F(6) = 6112 = 0.5

b) P(3 5T 5 4.5) = F(4.5)-F(3) = 4.5/I2 - 3112 = 0.125
c)P(-oo5T53.75)F(3.75)-F(-oo)=3.75/12-0 = 0.3125
d) P(T> 9.45) = 1 - F(9.45) = 1- 9.45 I 12 = 0.2125
e) P (I':? 15) = 1 - F (15) = 1-1 = 0
f} P(T = tr} P{!rST5 tr}=F{tr}-F(tr}=ml 2-11112 = 0
Q.6
In a :iven business venture, a man can make profit of Rs. 1000
; a loss of Rs.500. If t!1e probability of a profit '!_f}:6,
~e'!'
. _ _ - wl;at-;.,~ tire excepted profit m tile velllure? -
Chapier 3: Random Variables 46
swers
Ans. (profit) = (1000)(0.6) + (-500)(0.4) = 400.

don Q.7 If the variance of a random .variable X is 0.8, what is the
variance of tl1e raridom variable 2X, X/2 and X + 2?..
Ans. We know that;-
Var( aX) = a1 Var(X) and Var(X) = 0. 8
rso a :. Var(2X) = 2 2 Var(X)
= 4(0.8) = 3.2
and
Var(X/2) (112)2 Var(X)
=
d at 114 (0.8) = 0.2
=
Var (X + C) = Var (.X) ifC is a constant so
Var (X + 2) = Var(X) = 0.8.
By: Rafaqat
re. Q.8 If we want to get moments about mean rath,(!r than origin then
on what change we should make in computing mg/?
Ans. We should compute mgftaking mean as origin in place ofzero i.e.,
f
, M,.1- (t) = E( e(t(X-)) =e_,,,Mx (t) '
Q.9 What are the uses of characteristic function?

Ans. Characteristic function is particularly useful for dealing with
functions of independent random variables. For example, ifXi, X 2,
... , Xn is a sequence of independent (and not necessarily identically
distributed) random variables, and
S .. =f.a.x
;:I
..
where the a; are constants, then the characteristic function for Sn is
given by
<fJ S. (t) = 'P.11 (a 1t)<fJ.12 Cai t). <fJ.r. (a.O
Because of the continuity theorem, ~haracteristic functions are
used in the most frequently seen proof bfthe central limit theorem.
Characteristic functions can also be used to find moments of
random variable. Provided that r-th moment exists, characteristic
function can be differentiated r times.
Q.10 What is the use Chebyshev's Inequality?

Ans. The inequality can be useful despite loose bounds because it
applies to random variables of any distribution, and because these
bounds can be calculated knowing no more about the. distribution
than the mean and varla~ce.
47
Exercise 3 (True/False)
"True" or "False'1:
1. Since random variable is a real-valued function so the domain.

of this function is real line.
2. If a random variable X can take on only a finite number of
values or an infinite number of values that are countable, then
we call X a discrete random variable.
By: Rafaqat
3. The value of distribution function approaches to 1 as a random

variable X approaches to -co; .
4. No. two distinct distributions have the same density functions. ~-
5. The expected value of an experiment is obtained by computing

the arithmetic average value over all possible outcomes of the
experiment. '
6. If a man looses Rs. 100 iftail comes while tossing a fair coin
otherwise he wins Rs. 100. The expected value for the person
wining is Rs. 50.
7. If a t:onstant is added to a variab.le then the variance <;hanges
by multiplying a constant with the variance of that variable.
8. The variance of difference ' of two independent random
variables is equai to the sum of the variances of the both
variables.
9. Chebyshev's inequa1ny, can only be applied to those variables
.that have finite vari!ln~:
10: The characteristic function is s~mply the.logarithm of the mgf.
Chapter.3: Random Variables 48
A Quick Approach to Statistics "'1th Questions and Answers
Chapter 4
Discrete,
Probability Distributions
Discrete Probability Distributions

Discrete probability distributions give the probability of every possible
value ofa discrete random variable (discussed in Chapter 3). In this chapter,
By: Rafaqat
we discuss some important discrete probability distributions; Binomial,

Hypergeometric, Geometric, Negative Binomial, Poisson, and Multinomial
distributions.
Bernoulli Trial
There are {Ilany practical and experimental situations where out comeof
each repeated random trail can result in just two categories, namely
' "success" and "failure", or dichotomy ofresults can be found. For example,
two possible outcomes of each exam of a student can be passing or failing
of that student, correct or wrong answer, catching or missing a bus on a
stop, hitting or missing of a target, infected or non-infected from some
disease after result of a test, etc. Such trials are called Bernoulli Trials.
Simply speaking, a trial of a random experiment whose -outcomes can be
s classified into two categories, "success" or "failure", is referred as Bernoulli
trfal. .
n Binomial Distribution
An experiment haviJlg n (say) Bernoulli trials with the following properties,
h
is called Binomial Experim~nt:
a) The results of each trial can be classified into one of two

~s
categories, say, "success" and "failure".The outcome of interest is
usually referred as "success" ..
b) The
:
probabtlity
-;; -,' . ,
of success pis the same for each trial.
'-
Chapte'r 4: Discrete Probab'ility Distributions 49
is
c) Each trial is independent of all the others.

d) The experiment can be perfonned a fixed number of times, say, n.
Distribution Summary:
Parameters: Two parameters; n and p,

n =Number of trials,
p =Probability of success
pmf: x = 0, 1, 2, , n,
where
X = Nmber of successes in n trial~.
By: Rafaqat
q = 1-p
Mean: np
Variance: npq
l-2p
Skewness:
~npq
l-6pq
Kurtosis: npq
mgf: (q+pe'Y
Char. June.:
Chapter 4: Discrete Probability D"istributions 50
r. Hypergeometric Distribution
There are many experiments in which the condition of independent trials is ..
not met and resultantly, the probability of success does not remain constant
from trial to trial. Such experiments are called hypergeometric experiments
with the following properties:

categories, say, "success" and "failure". The outcome ofinterest is
usually referred as "success".
b) The successive trials are dependent.
c) The probability of success changes from trial to trial.
d) The experiment can be perfonned a fixed number of times, say, n.
By: Rafaqat
Parameters: Three parameters, N, k and p,
N = Number ofunits in the set or population,
k =Number of successes (units of interest) in
the set or population,
p= ~ = Probability of su~cess.
pmf P(X ~ x)
(kxN-KJ
= .xn- x
. (:) '
where
X = Number of successes (units of interest)
inn (sample size of items selected,
without replacement),
x = 0, 1,2,,n, and x = 0,1,2, ,k.
k
Mean: np=n-
N
Variance: (N-n)
npq - -
N-l . N
k(1 -N'
=n- k-xN-n)
--
N-l
D . Chapter 4: Discrete Probability Distributions 51
Geometric Distribution
In many practical situations, an experimenter is interested in the first
success in the experiment. To obtain this, he repeats the experiment until he
gets the first success. For example, a researcher keeps on taking blood
samples until he gets 0-ve blood group. To handle such situation, here a
distribution presented called geometric distribution and an experiment in
which .trials are repeated until first success is obtained, is called geometric
experiment. A geometric experiment has the following pi:_operties:

categories, say, "success" and "failure". The outcome of interest is
usually referred as "success".
b) The successive trials are independent.
By: Rafaqat
c) The probability of success remains constant froin trial to trial.

d) The experiment is repeated a variable number of times until the
first success is obtained.
Distribution Summary:.
Parameters: One parameter, p,

p = Probability of success.
pmf P(X = x)= pq'; x =0, 1,2,,

where
X= Number of failures preceding first
success,
q = 1-p.
Mean:
!1.
p
I
~.
_g_
Variance: p2
_li_
mg[:
1-qe'
01apter 4: Discrete Probability Distributions 52
A Quick Approach to Statistics with Questions andAnswers
Note: Another function of geometric distribution is also used as follows:

Probability Function:
x =l,2, .. .,
where
X= Number of trials to. have first success,
p = Probability of success,
=
q 1- p,
also p is the only parameter of this distribution.
1
Mean:
p
By: Rafaqat
1
Variance: -2
p
Negative Binomial (fascal) Distribution

In geometric experiment, an experimenter is interested in the first success
and repeats the experiment until he gets the first success. But in many.
practical situations, this interest can be extended to a fixed number of
success rather than just single success. For example, a researcher keeps on
taking blood ,samples until he gets 10 donors having 0-ve blood gr9up. To
handle Sl;!Ch siri.iation, here a di.stribution presented called negative binomial
distribution and an experiment ih which trials are repeated until a fixed
number of successes~ say k. is obtai'ned, is called negative binomial having
following properties: - -
a) The results of each- trial 'Can be classified into one of two

.categories, say, "success" and "failure". The outcome of interest is
usually referred as "success''.
b) The successive trials are independent.
c) The probability of success rell)ains constant from trial to trial.
d) The experiment is repeated a variable number of times to obtain a
fixed number of successes, say k.
Chaplet 4: Discrete ProbabilitY Distributions 53
Parameters: Two parameters, p and k,

p = Probability of success.
K =Number of successes.
pmf P(X = x) = (x-1)

k-:- I pq-; x = k, k +I, k + 2,-
where
X= Number of trials to produce k successes,
q =I- p.
k !l.
By: Rafaqat
Mean: p
Variance:
mg(: . .
p ( l-qe') -
Note: Another function of negative binomial distribution is also .used as

follows:
Y=X-k,
y+k-1)
P(Y=y)= ( k-1 ~qy; y = 0, I, 2,- ,
where
Y= Number of failures preceding k successes.
The mean and variance for Ywill be changed accordingly.
Poisson Distribution
A distribution often used to compute probabilities for random variables
distributed over time and space is the Poisson distribtion. For example,
Chapter 4: Discrete Probability Distributions '54
experience has shown that Poissort is an excellent model to use for

computing probabilities associated with the number of calls coming into a
telephone switchboard during a fixed period of time. Other examples are,
the number of automobile deaths per month in a large city. the number of
bacteria in a given culture, the number of atoms disintegrating per second
from radioactive material etc. Poisson distribution is the premier probability
model and perhaps, the second most frequently used discrete distribution
after the binomial distribution. It is the limiting form of binomial
.distribution.
Following are the assumptions while using Poisson distribution:
a) Events that occur in one time interval (or region or space) are
independent of those in any other non-overlapping time interval (or
region or space).
By: Rafaqat
b) For a small time interval (or region or space), the probability that
an event occurs is proportional to the length of the time interval (or
region or space). '
c) The probability that two or more events occur in a very small time
interval (or region or space) is so small that it can be neglected.
Parameters: One parameter, A.
pmf e' A:'

P(X=x)=-- x = 0, I, 2, , oo
x!
where
X =Number of events occurring per unit time
or per unit sp!\,ce.
Mean A.
Variance:
mg{: Exp{A.(e' -1)}
Chapter 4: Discrete Probability Distributions 55
--------------
A Quick Approacll to Statistics with Questions and Answers
Multinomial Distribution
If a trait's outcomes can be classified into more than two categories, a
binomial experiment becomes multinomial experiment. For example, a
finished product may be classified as excellent, good, average; a student's
grade may be A, B, C, D, or F etc.
A multinomial experiment has the following properties:
a) The results of each trial can be classified into one of k non.-

. overlapp.ing categories, say, C., C2, , Ck.
b) The probability of ith outcome is p; which remains constant from
' t
trial to trial and L p, = I .
i=I
c) The successive trials are independent.

By: Rafaqat
d) The e)(periment can be performed a fixed number of times, say, n.
Parameters: k ( number of categories) parameters, n, Pi. p 21

.... Pk-I,
p; is the probability of success of zlh category.
pmf P(X1 = xl'X2 = xw .. ,X. = x.> .
= n! (p/l(py2 ... (pt)Xk;
x1!x2! .. x,!
x, = 0,1,2,,n,
where
X; =Number of successes in ith category,
t t
L.x. =n, LP.= I.

1=1 i=I
Mean: np; for X;
Variance: nplJ; for X;
Chapter 4: Discreie Probability Distributions 56

Questions-Answers
Q.l What is the relation between Bernoulli trials and binomial

experiment?
Ans. A binomial experiment has two or more Bernoulli trials.
--Q.2 What is. the distribution of the sum of independent binomial

variates?
Ans. That distribution will also be binomial. if we have X 1, X 2, .. , Xm
independent variates having binomial distribution with parameters
n and p then f X. has also binomi;I with parameters rnn andp.
.. 1
By: Rafaqat
Q.3 What is the reason that mean of binomial variates is always

greater than its variance? .
Ans. It is because mean ofbinomial variates is np while variance is npq.
Since q < 1, the variance will be less than np (mean).
Q.4 Criticize the following statement:

The mean of binomial distribution is 5 while standard deviation
is 3.
Ans. It is not possible. In above statement the variance is 9 which is
greater than mean so it is not possible (see also Q.3 above).
Q.5 What is the main difference beiween a binomial and a

hypergeometricf!Xperiment?
Ans. , Jn binomial distribution, the successive trials are independent and
the probability of success remains constant from trial to trial
while, on the other hand; in hypergeometric distribution, the
successive trials. are dependent and the probability of success
changes from trial to trial.
Q.6 What is the main difference between binomial experiment and

negative binomial experiment?
Ans. Jn binomial experiment, the number of trials is f rxed and the
number of successes is variable while in negative binomial
Chapter 4: Discrete Probability Distributions 57
A Quick Approach to Statistics witil Questions and Answers
experiment, the number of trials is variable but the number of

successes is fixed
Q.7 What is the relation between negative binomial and geometric
experiment?
Ans. If the number of successes is reduced to 1 to stop an experiment, .
the negative binomial experiment becomes geometric experiment.
Jn other words, if k = 1, negative binomial distribution becomes
t' geometric distribution.
[.
Q.8 Criticize tile following statement:
The mean of Poisson distribution is 3 while standard deviation is
3.
Ans. In Poisson distribution, mean and variance are equal. In the above
statement, the variance is 9 that do not equate the mean.
By: Rafaqat
Q.9 What is tile distribution of the sum of independent Poisson

variates?
Ans. That distribution will also be Poisson.
Q.10 Is _it possible to apply binomial or Poisson to the same problem?

Ans. Yes, it is possible if the probability of success is very small and the
number of trials is very large.
Q.11 Is it possible that if trial of a random experiment results in more

;,~ ' than two outcomes but we can still apply binomial distribution?
Ans. Yes, it is quite possible because a trial can result in more than two
possible outcomes but we should categorize them in just .two
categories.
Q.11 How many number ofparameters in multinomial distribution?

Ans. As many as the number of categories into which outcomes of trial
are divided
Chapt~r 4: Discrete Probability Distributions 58
Exercises
Exercise 4.1, (MCQs')
Q.l Which of the following could never be described by the binomial

distribution?
By: Rafaqat
(a) The .number of defective items produced. by an assembly

process.
(b) The amount of water used daily by a single household.
(c) The number of people in a class who can answer a
particular question correctly.
(d) All ofthese.
Q.2 Standard deviation of the binomial distribution depends upon;

(a) Probabiljty of success
(b). Number of trials
(c) both (a) & (b) above
Q.3 For a Poisson distribution with standard deviation equals to 2 then

inean of the Poisson distribution equals to;
(a) o
(b) I
(c) 2
(d) 4
Q.4 A binomial distribution with n ~ I 000 and p = 0.5 is;

(a) Symmetrical
{b) Asymmettkal
(c) Skewed to'rigltt
(d) Skewed to left
Chapter 4: Discrete Probabiiity Distributions 59
A Quick Approach to Statistics with Questio~Jand Answers
Q.S If X has binomial distribution with parameter p and n therr Xln has
the variance
(a) nl/.q
(b} npq
(c) pqln
(d) pq!n2
Q.6 If X is number of .trials for negative binomial distribution with

parameters p and k then its minimum value is
(a) _o
(b) k
(c) k+l
(d) k-1
By: Rafaqat
Q. 7 For a given binomial distribution with n fixed if p = 0.5 then

(a) Poissori distribution will provide a bad approximation.
(b) Poisson distribution will provide a good approximation.
(c) Binomial distribution will be skewed left.
(d). Binomial distribution will be skewed right.
Q.8 Which of the following is necessary conditjon for use of a Poisson

~ distribution: . . . _, _ .
(a) Probability of one arrival per second is constant.
(b) The number . of arrivals in .any OJ!e second . interval is
,.,~,.
independent of ai:rivats in other intervf!,ls. -
(c) The probability of two or more arrivals i.n _the same
second is zero. -
(d) (b) and (c) above.
Q.9 The necessary .and suffiCient .condition of the, hypergeometric

distribution is: - -
' (a)
(b)
Sampling with replacement
SampHngwi~hout replacement.
(c) 'Trials are independent
(d) Probability of success remains constant
'" -
Chapter 4: Discrete ProbQbOity Distribution$" 60
has Q.IO Which of the following is the most reasonable condition for the
binomial approximation to the hypergeometric distribution?
(a) N= 200, n = 12
(b) N= 500, n = 20
(c) N= 640, n.= 30
(d) N= 800, n = 50
vith Q.11 Suppose, we have a Poisson distribution with A/ equals to 2 then the
probability of having exactly I 0 occurrences is;
-10 e-IO
(a) . 2
10!
(b)
2'" e-2
2!
By: Rafaqat
102 e-10
(c)
10!
(d)
2'" e -2
.'JO!
son Q.12 Which of the following is a characteristic of the probability

distribution for any random variable:
(a) A probability is provided for every possible value
I is (b) The sum of all probabilities is one
(c) No given probability occurs more than once
me (d) (;i) and (b) above
Q.13 In what case would the Poisson distribution be a good

approximation of the binomial distribution:
tric (a) n = 40, p = 0.32
(b) n=40, q=0.79
(c) n = 200, q = 0.98
(d) n =; 10, p = 0.03
Q.14 The mode of the geometric distribution is:

(a) I
(b) 2
(c) 3
60 Chapter 4: DiScrete Probability Distributions 61
A Quick Approach to Statistics with Quesiion.s and Answers
(d) not possible
Q.15 A binomial dis~ibution may be approximated by a Poisson

distribution if:
(a) n is large and pis large
(b) n is small and p is large
(c) n is small and p is small
(d) none of these

By: Rafaqat

"True" or "False';:
1. Hypergeometric distribution has four parameters. ,

2. When the probability of success in a. BemoulHprocess is 0.5,
its binomial distribution 'is symmetrical.
3. T.he variance of binomial distribution can be zero if the
probability of success is one. .
4. The binomial distribution is not really necessary bec'ause its
values can always be appro:is.imated by another distribution.
5. The body temperatures of adult males can be described by
Poisson distribution.
6. Tossing an unbiased coin 5000 times can be treated as
binomial experiment.
7. Tossing a biased coin 50 times can be treated as
hypergeometric experiment.
8. For Poisson distribution, mean and variance are equal.
9. The mean of the ratio of number of successes to the.number of
trials in binomial experiment is equal to probability of success.
10. Poisson distribution is also referred as distribution of rare
events.
Chapter 4:.Discrete Probability Distribu.tions 62
111111 Chapter 5
continuous
Probability Distributions
Continuous Probability Distributions

The probability distributions for continuous random variable, discussed in
By: Rafaqat
Chapter 3, are referred as continuous probability distributions. In this

chapter, we discuss some important continuous probability distributions;
Uniform, Exponential, Normal, Gamma, Chi-Square, t-, and F-
distributions.
Uniform Distribution
A continuous rahdom variable X is said to follow a Uniform distribution
with parameters a and b, '"'.ritten X - U(a,b), if its probability density
function is constant within a finite interval [a,b], and zero outside th.is
interval (with a less than or equal to b). In other words, the values of a
uniform random variable are uniformly distributed over an interval. For
example, if buses arrive at a given bus stop every 15 minutes, and you
arrive at the bus stop at a random time, the time you wait for the next bus to
arrive could be described by a uniform distribution over the interval from 0
to 15 minutes.
One of the most ,important applications of the uniform distribution is in the
generation of random numbers. That is, almost all random number
generators generate random numbers on the (0,1) interval. For other .
distributions, some transformation is applied to the uniform random
numbers.
The following is the plots of the uniform probability density function and
cumulative distribution function:
Chapter ,5: Continuous Probability Distributions 63
... A. Quick Approach to.Statistics with Questions and Answers
.... --~
. .... . ..
I
a
r . f
! ..
I..... ...
.. ~-
I .. .. .... ...
I
, Distribution Summary:
By: Rafaqat
Parameters: a,be(-00,00)
pdf --' a<x<b

b-a
o' x < aorx>b
O; x<a
X-:-a
""' cdf --
b-a '
aSx<b
l; x~b
a+b
Mean: -- 2
Median: a+b
...' --
2
.~
Mode: any value in [a,b}
Variance: (b-a) 2
12
Skewness: 0
Chapter 5: Continuous Probability Distributions 64
6
Kvrtosis: 5
e"' -era
mg[
' t(b-a)
itb Ila
e -e
C~ar. June.:
it(b- a)
Exponential Distribution
The exponential distribution is used to model Poisson processes, which are
By: Rafaqat
situations in which an object initially in state "A can change to state B with
cons~~mt probability per unit time ..t. The time at which the state actually
changes is described by an exponential random variable with parameter ..t
The exponential distribution is also known as the waiting-time 9istribution,
describes 'the amount of time or distance between the occurrence of random
events such as the time between major earthquakes or the time between two
cons8Cutive goals ina football match or the time until you get seat in a bus
etc. This distribution is also used in connection with estimating the length of
materiallife, or the lertgth of time a process might take.
The following is the plot of the exponential probability density function and .
..,
!\
_.J\
-t~- --.,.--- - ~ - - . . . . . . . - ~- :-~,......... _,..,

11 I I
Chapter 5: Continuous Probahility.Distribut~ons 65
Parameters: A.> 0; mean time (rate) of a
process
pdf
_,,
A. e ' x~O
0, x<O.
X = time elapsed
cdf 1-e -A<
I
Mean: -
A.
By: Rafaqat
ln2
Median:
-
A.
Mode: 0
Variance: A.'
Skewness: 2
I"'
Kurtosis: 6
t. mgf: (1- ~r
Char. June.: (1-:f r

Normal Distribution
The Nonna! Distribution, also called Gaussian distribution, is an extremely
important probability distribution. in many fields. The fundamentaJ
importance of the nonnal distribution is its use as model of quantitative
phenomena in the natural and behavioral sciences: A variety of
psychological test scores and physical phenomena can be well
approximated by a normal distribution. For example, height at a given age
for a given gender in a given racial group is adequately described by a
nonnal random variable even though heights must be positive.
For both theoretical and prac.tical reasons, the normal distribution is
probably the most important distribution in Statistics .. For example, many
classical Statistical tests are based on the assumption that the data follow a
nonnitl distribution. This assumption should.be tested before applying these
tests.
By: Rafaqat
In modeling applications, such as linear and non..:linear regression, the error

term is Qften assumed to follow a normal-distribution with fixed location
and scale. The normal distribution is used to find significance levels in
many hypothesis tests and confidence intervals. The normal distribution
also arises in many areas of Statistics: for example, the sampling
distribution oftht!" mean is approximately normal, even ifthe distribution of
the t'opulation from which the sample is taken is not normal. In probability
theory, nonnal distribution arises as the limiting. distrib.ution of several
continuous and discrete families of distributions.
The following are the plots of the normal probability density function and
...
I
I'
i
ru - t
/
I O.t
.. 4 oJ _,
' ' . .. ... 4
A Quick Approac/1 to Statistics wltb Questions and Answers
Distribution Summary: s
l
Parameters: ..._ oo S s; oo; u 2 > 0 2)
d
l
pdf . &exp -(x.~}2)
I.
u 2n
(
2u_2
, . -OO<X<OO
0, otherwise s
ti
~(l+ex{:Ji ))
c
cdf
1
a
By: Rafaqat
Mean:
Median: (
1
Mode: i!
Si
Variance: cl-
a
Skewness: 0 a
ii
Kurtosis: 3 h
'
n
l
mgf: ex{ t+--
u2t2)
1
2 .
b
p
Char. June.: ex { it--
u2t2)
-
2
Ii
ti
Ii
ll
(1
'.
Standard Normal Distribution

The Standard Nonnal Distribution is the normal distribution with a mean of
zero and a standard deviation of one. It is the simplest case 'of the normal
distribution. This is written as N (0, I).
The pdf for standard Nonna) distribution is
ji;_e_z2'2 ,
/(z) -- _1 -oo<:z<oo
Since the general form of probability functions can be expressed in terms of

the standard normal distribution, all subsequent formulas in the above table,
can be obtained by plugging =0 and c/ = l .
The standard nonnal distribution has wide use, especially, in the
applications of Central Limit Theorem (CLT).
By: Rafaqat
Central Limit Theorem (CLT)

The Central Limit Theorem states that whenever a random sample of size n
is taken from any distribution with mean and varianc_e c/ , then the
sample mean X wm be approximately normally distributed with mean
and variance c/ In. The larger the value of the sample size n, the_ better the
approximation to the normal. This is very useful when it comes to
inference. For example, it allows.us (if the sample size is fairly large) to use
hypothesis tests which assume normali.ty even if our data appear non~
normal. This is because the tests use the saJl!ple mean X , which the Central
Limit Theorem tells us will be approximately normally-distributed.
The normal distribution is widely used. Part of the app_eal is that it is well
behaved and mathematic~lly tractable. ijowever, the central limit theorem
provides a theoretical basis for why it has wide applicability. The central
limit theorem basically states that as the sample size n becomes large, the
following occur:
. X-'-
Lim_ ---y-N(O,l) .
......, ulvn
In other words, as sample size increases,. the sampling distribution of the
mean becomes approximately normal regardless of the distribution of the
original variable.
'
A Qu!ck Approach to Statistics with Questions and Answers
Gamma Distribution
Gamma densities provide a fairly flexible class for modeling nonnegative
randorrt variables. The exponential distribution becomes gamma
distribution, when we 9onsider sum of independently identically distributed
exponential variates. The gamma distribution ~an be used ,to describe the
probability that a events will occur within a time period ii . Contrast it
with the exponential distribution, which describes the. probability that one
event will occur.
The following are the plots of the gamma probability density function and
By: Rafaqat
- - ...............-.. . -
...--Gll_m_...
_P_D_F-=(ea.__m_m_a_:D_.6...:;J-.., Gemma PDF (gZlmma ::11
8
~ Ii ;a.
I!! ii 0.75
.!! 4 .!!
~ 3 ~ o.s
:a :a
B ;z
~
~ 025
Cl.
\.._
o+--.,,...,......,,_.,_,.......,.-.-~~~ a-1-~,_...,:::>;......,.......,..~-.--..--1
a 1 2 :s 4 s e 1 a 9 10 0 1 2 3 4 ~ 8 7 B. 9 10
x
Ga mm a P OF fsain ma =2)
0.4r------~---
~ ~
: 0.3 ~ 0.15
l'1 ' l'1
~ 02 ~ 0.1
2l :a
B
12 0.1 la.a&
Cl. Cl.
o+""..--.....,,.....,.....,...,..........,.....,......,..~
1 2 :I 4 5 fl 7 fl 9 10 o 1 2 3 4 s fl 1 a e 10
x x
l'
A Quick Approach to Statistics with.Questions and Answers
:ive .--------------------'"'-
Gmm CDfl!Pm ..... 0.5)
--------------------------
Ga., . . CDf4'111111ms1J
1
1ma ag
1ted 0.1 l:'0.75
the !' 0.7
t it ! ~= i
0
0.5
Q: OA a:
0.3 0.25
)Oe
0.2
0.1 ........~~~~~~~-1 a,_,~....,..~~~-.-~~
o 1 2 3 4 s e 1 a 1 10
x a 1 2. 2 4 l e 1 e 10
md
~mrn11 CDf (Gemma :2] Gltmint COF (&ammo: I)
i0.76 l
JI 0.5'
I! .
I
By: Rafaqat
a.
0.25
0
o 1 2 3 4
---~-~...-~J
l e e 1 11 10
Q.
a
I
1 2 2 ' l e 1
I
e 1 10
Parameters: a > 0 shape (real)

f3 > 0 scale (real)
l a-I -x!P
pdf --.-ax e ., x2'.0
rap
Mean: a/J
Mode: (_a - l)/3 for a ;;::: I
Variance: af3
0
Chapter 5: Continuo11s Probability D~iributions 71
2
Skewness: r
-.;a
.6
Kurtosis: _
a
ingf: (I - p1r 0
fort<. l/p
Char. June.: (1-fJitr 0

By: Rafaqat
Chi-Square Distribution
In probability theory and Statistics, the Chi-square distribution '(also Chi-
squared or x' distribution) is one of the theoretical probability distributions
l most widely used in inferential statistics, i.e. in statistical significance tests.
1L
It is useful because, under reasonable assumptions, easily calculated
t
'' quantities can be proven to have distributions that approximate to the Chi-
h square distribution if the null hypothesis is true .
,,
.t
If Xi are k independent, normally !iistributed random varfables with means

t'!
~;
, and variances a:,
then the random variable
~'t z= (X,-,)z
; .. 1 U;
". is distributed according to the Chi.:square distribution.
t~
The Chi-square distribution has one parameter: k - a positive integer which
J
specifies the number of degrees of freedom (i.e.' the number of Xi). The Chi-
~' ~ square distribution is a special case of the gamma distribution.
The best-known sifuations in which the Chi-square dis'tribution is used are
\' the common Chi-square tests for goodness of fit of an observed distribution
~ to a theoretical one, and of the independence of two criteria of classification
of qualitative data. Two common examples are the Chi-square test .for
independence in an R x C contingency table and the Chi-square test- to
determine if the standard deviation of a population is equal to a pre-
specified value. However; many other statistical tests lead to a use of this
' .. ,_ distribution. One example is Friedman's analysis of variance by ranks.
,.
A Quick Approach to S~atistics with Questions and Answers.
rs
The following are the plots of the Chi-Square probability density function
and cumulative distribution function: '
..--C-hl-S__,qua_r_P_D_F~(1_dll_._~ o.s..--C_hi._S~qua_ie_P_D_F~~-dll---__,
4
i
. g.
OA
0.3
t
I u
I a.t
o~.::::;,,,,.......,._,...,...,......,.._.._,......: o~,_.., ........._,_,......~:;==....,.....,
012341871191.Q a 1_ 2 3. 4 l e ]'.. a a 10
ChHlqLll r. PDF {i di)

02....----....;_---'--'-----.
By: Rafaqat
-~
ll r O.Hi 50.075
i-
IS
i il.1 :a~ .ODli
s. ~
... 0.05 I... 0:02&
:d o+-..............-r.....-............-.,.....,,......f
i- 1 2 :a 4 s e
x
7 11 e ro a 1 2 .2 4 i e 7 a e 10
lS
t-0.75
i OJi
~
02&
o+-~~~~~~~....._. o+-~~~~~~~........-f
0 1 2 2 4 l 8 7 II 9 10 0 1 2 3 4 l 8 7 II 9 10
ChHlqusieCOF(lid1) Chl-8qLllN COF (10d1)

.os.----'-----"------.
re 0.5
~0.7&
m ~ OA
~ :a I
m JI OJi JI Q.3
h!
>r ... 02li
I!
~ 02 '1
to . 0.1
e- , 2 3 4 1e 7 e a 10
0
o t 2 2 4 l e 7 a e 10
is
A Quick Approaclt to Statistics witlt Questions and Answers
Parameters: k > O; degrees of freedom
1 !._, -~
pdf - -2 - - x z ez x::=:O
2" _f(k I 2)
0 otherwise
Mean: k
Median: 2
k--
3
By: Rafaqat
Mode: k-2 if k ::::2
Variance: 2k . .
Skewness:
J8/i.
Kurtosis: 12/ k
mgf: (1-2t)-u 2 for 2t <1
Char. June.:
..
Student's t -Distribution
The I-distribution or Student's I-distribution is a probability~distribution that
arises in the problem of estimating the mean of a normally distributed
population when the sample size .is small. It is the basis of the popular
Student;s t-test for the statistical significance of the difference between two
sample means, and for confidence interVals for the difference between two
population means. The Student's I-distribution is a special case of the
generalized hyperbolic distribution.
Cltapter 5: Continuous Probability Distributions 74
The derivation of the I-distribution was first published in 1908 by William

Sealy Gosset, while he worked at a Guinness brewery in Dublin. He was not
allowed to publish under his own name, so the paper was written under the
pseudonym Student. The I-test and the associated theory became well-
known through the work of R.A. Fisher, who called the distributidn
"Student's /-distribution".
Student's /~distribution arises when (as in nearly all practical statistical
work) the population standard deviation is unknown and has to be estimated
from the sample data. A number of Qther statistics can be shown to have t-
distribution for samples of moderate size under null hypotheses that are of
interest, so that the I-distribution forms the basis for significance tests in
other situations as well as when examining the differences between means.
For example, the distribution of Spearrnan's rank correlation coefficient p
By: Rafaqat
in the null case (zero correlation) is well approximated by the t-distribution

for sample sizes above or about 20.
Let V have a Chi-square distribution with v degrees of freedom. Further
suppose that Z is stand&rd normal variate and,. V a.nd Z are independent then
the ratio
z
.Jv Iv
has a I-distribution with v degrees of free~om.
The following are the plots of the student's t probability density function
and cumulative distribution function:
,.---
1
------------------------------------------
1 PDF
..
OAT-_ _ _ _ (1 di') _._,-':---~
OA~----~~--~
.,,. !' -
: 0.2 : 0.3
.Ji .!I
ii' 02 ~ 02-
:iii
j
~ 0.1 e. 0.1
a.
o~..--,_,..........,......,,.....,,__,.....,.;;::i ,.....,,...;;-.--i
o.P.......,co;....~..............
~.-4-3 ~ ~ i 1 2 3 4 & . ~ ~-3 ~ ~ i 1 2 3 4 &
,,. QA,-----....-----~
i 0.3
l!'
By: Rafaqat
02
2i
Ia. a.1
o+-,......,:;,..,...........,"""'"'.........;:;....,....,.
~ ~-3 ~ ~ i ~ 2 3 4 &
,,.o.75
=
I0
u
D:
026
O't-.,....,......,.....,.....,.....,,.....,,_..,_,,._. o+-..-.,...;-..--.,.....,.........-....,....--1
~ -4-3 ~ ~ i 1 2 3 4 & -6 -4 -3 ~ _, 0 ., 2 3 4 Ii
' x
1
.. t CDF(illd1 t CDf.(30d11
1
a
F
~0.76 ,,.o.7& g
ii i l
! OJi ! OJi
2 e re
a. a.
026 026 d
T
h:
Cl
v;
Chapter 5: Continuous Probability Distributioas 76 (
,
Parameters: v > 0, degree of freedom (real)
pdf r((v + 1)/2)

.{;;qv/2)(1+.x 2 1vr'" 2
-OO<X<OO
Mean: 0 for v > I, undefined fot: v = I
Median: 0
By: Rafaqat
Mode: 0
_v_ for v>2

Variance: v~2
Ske:wness: 0 forv> 3
3v-6
Kurtosis: - - for v>4
v-4
F -Distribution
The F-distribution is a continuous probability distribution. It is also known
as Snedecor's F-distribution or the Fisher-Snedecor distribution (after R.A.
Fisher and George W. Snedecor). F.stands for Sir Ronald Fisher, English
geneticist' and statistician. .
The distribution is used in the analysis of variance and is a function of the
ratio of two independent random variables each of which has a Chi-square
distribution and is divided by its number of degrees of freedom.
The F distribution is used in many cases for the critical regions for
hypothesis tests and in determining confidence intervals. Two .common
examples are the analysis of variance and the F-test to determine if the
variances of two populations are equal.
Chapter 5: Continuous Probability Distributions .. 77
A random variate of the F-distribution arises as the ratio of two Chi-squared

variates:
XJv,
X2'v2'
where X 1 and X2 are independent and have Chi-square distributions with v1
and v2 degrees of freedom respectively.
The following are the plots of the F probability density function and
FPDF(1.10dl)
4 4
By: Rafaqat
.... ....iii
..!I .
iii 3 .
8
3
~
:a
!
2
1
..
~
!
2
9
a. 2
a.
a a
a 2 3 4 Ii a 2 :I 4 I
x x
F PDF (1G, 1 di) F P~(10, 10 d1J
GJi a.a
l !' o.4 ~
:i
G.7
G.11
.! 02 I! GJj
~ ~ OA
~ 0.2 j
0.2
! !
9 2 0.2
a. G.1 a. a.1
a a
a 1 2 3 4 Ii 0 2 3 4 Ii
x x
d I
.l
'
Chapter 5: Contiiruous Probability Distributions 78

'
I http://stat9943.blogspot.com
A Quick Approach to Statistics with Quesiions and Answers
I' COF(1, 1 df) F CDl'(1,10d1)

0.8
0.7
~G.76
~0.8
...- 0.6 =
.!
0
OA i 0.6
0
~
0.3 ~
02 021
'1 a.1
a a
0 2 3 4 & 0 1 2 3 4 &
id x x
F COF (10, 1 d1j F CDF (10, 10 di)
a.7
G.8
~a.& i:'0.76
...= GA ii
0.6
.!g G.3 .!g
a. 02 a..
By: Rafaqat
02&
0.1
2 3 4 5 2 J 4 Ii
x x
Parameters: v 1 > 0, v2 > 0 degree of freedom
(v1xr1 v;2 .
pdf (v1x+vJ"1 ..2
xj{~ + i)
Mean:
Mode:
0
Chapter 5: Continuous Prob'fibility Distributions 79
78
Variance:
Skewness:
I
By: Rafaqat
1
'I
80
Chapter 5: Continuous Probability Distributions
i
j
; .
Questions-Answers
Q.1 Why is the uniform distribution also called rectangular

distribution?
Ans. The. uniform distributioh. defines equal probability over a given
range for a continuous distribution. For this reason, the shape of
its probabilitY density function is rectangular. It is parameterized
by the smallest and largest values that the uniformly-distributed
random variable can take, a and b. That is why uniform
distribution is also known as rectangular distribution.
Q.2 What are tile applications of uniform distribution?

By: Rafaqat
Ans. The uniform distribution defines equal probability over a ,given

range for a continuous distribution. For this reason, it is important
as a reference distribution.
One of the most important applications of the uniform distribution
isJn the generation ofrandom numbers. That is, almost all random
number generators generate random numbers on the (O, J) interval.
Fpr other distributions, some trqnsformation is applied to the
uniform random numbers.
Also whe.n a p-value is used as' a test statistic for a simple null
hypothesis, and the distribution ofthe test statistic is continuous,
then the .test statistic is uniformly distributed between 0 and I if the
null hypothesis is true.
Q.3 What is tile resemblance between exponential and geometric

distribution?
Ans. The exponential distribution may be viewed as a continuous
counterparf of the geometric distribution, which describes the
number of Bernoulli trials necessary for a discrete process to
change state. Jn contrast, the exponential distribution describes the
time for a continuous process to change state.
Q.4 What are the applications of expommtial distribution?

Ans. Exponential variables can be used to model situations where
certain events occur with a constant probqbi/ity per unit distance:
the" distance between mutations on a DNA strand;
8.0 Chapter 5: Conlifluous Probability Distributions 81
the distance between road- kill on a given street.

In queuing theory, -the inter-arrival times (i.e. the times between
customers entering the system) are often modeled as exponentially
distributed variables. The length of a process that can be thought
of as a sequence of several independent tasks is better modeled by
a variable following the gmnma distribution (which is a sum of
several independent exponentially. distributed .variables).
Reliability theory cmd reliability engineering also make extensive
use of the exponential distribution. Because of the memoryless
property of this distribution, it is well-suited to model the constant
hazard rate portion of the bathtub curve used in reliability theory. _
It is also very convenient because it is so easy to add failure rates
in a reliability model. The exponential distribution is however not
appropriate to model the overall lifetime of organisms or technical
By: Rafaqat
devices, because- the "failure rates" here are not cbns_tant: more
failures occur/or-very young and/or very old systems.
In physics. if one observe a gas at a fixed temperature and
pressure In a uniform gravitational field, the heights of thew1rious
molecules also follow an approximate exponential distribution.
Q.5 What is tlze importance of normal distribution in Statistics?

' Ans. The fundamental importance ofthe normal distribution is its,use as
{ model of quantitative phenomena in the natural and beh'avioral
sciences. For bot!? theoretical and practical reasons, the normal
distribution is probably the most important distri~ution in
statistics. For example, many classical statistical tests are based
on the assumption that the data follow a normal distribution. This,
'
~-
assumption should be tested before applying these tests.
In modeling applications, such as linear and non-linear
regression, the error term is often assumed to follow a normal
distribution with fixed location and scale.
The normal distribution is also used to find significance levels in
many hypothesis tests and confidence intervals.
I:
r
Q.6 Give the practical importance of the Centrl Limit Theorem
(CLT).
Ans. _ The practical importance ofthe Cl Tis that the normal distribution
can be used as an approximation to many other disthbutions. For
example a binomial distribution with parameters n and p is
t;
A Quick Approach to Statistics with Questions and. Answers
approximately normal for large n and p not too close to 1 or 0 (see

CLT above for more details).
Q. 7 What is the relation between exponential and gamma

distribution?
Ans. Gamma variate is the sum of independent exponential variates.
Q.8 What is the relation between Cid-square distribution and gamma

distribUtion?
Alts. Chi-square distribution is a special case of gamma distribution
with a = k 12 and f3 = 2 , where k is the degree offreedom of Chi-
square distribution.
Q.9 What arethe applications of Chi-Square distribution?

By: Rafaqat
Ans. The best-known situations in which the Chi-square distribution is

.used are the common Chi-square tests for goodness of fit of an
observed distribution to a theoretical one, and of the independence
of two criteria of classification of qualitative data. However, many
other statistical tests lead to a use of this distribution. One exa.mple
is Friedman's analysis o.fvariance by ran~.
Q.10 What are the applications oft-distribution?

Ans. Student's t-distribution arises when (as in nearly all practical
statistical work) the population standard deviation is unknown and
has to be estimated from the sample data. A number of other
statistics can be. shown to have t-distributions for samples of
moderattkize under null hypotheses that are of interest, so that the
t-distribution forms the basis for significance tests in other
situations as well as when examining the differences between
means. For example, the distribution of Spearman's rank
correlation coefficient p in the null case (zero correlation) is well
approximated by the t distribution for sample sizes above or about
20. .
Chpter 5: Continuous Probability Distributions 83
A Quick Approach to Statistics with Quesflons amt Answers
Q.1 I What is the shape of I-distribution?

Ans. The overall shape of the probability density function of fhe !-
distribution resembles the bell shape of a normally distributed
variable with mean 0 and variance I, except that it is a bit lower
and wider.
Q.12 What are-the uses o/F-distribution?

Ans. The F-distribution is used in many cases for the critical regions for
hypothesis tests and in determining confidence intervals. _Two
common examples are the analysis of variance (iNOVA) and the
F-test to determine if the variances of two
populations are equal.
By: Rafaqat
Chapter 5: Continuous .l''f"Ob.ability Distributions 84
Exercises
Q.1 A continuous random variable is a random variable that can:

(a) Assess only countabie values.
(b) Assess any value in.one or more intervals.
(c) Have no random sample.
(d) Assume no continuous random frequency.
a
By: Rafaqat
Q.2 For continuous random variable the area under the probability
distribution curve between any two points is always:
(a) Greater then one
(b) Less then zero
( c) Equal to one
(d) In the range zero and one
Q.3 The probability that a continuous random variable assumes a single

value is:
(a) Less than one
(b) Greater than zero
(c) Equal to zero
(d) Between zero and one
Q.4 For a continuous random ~ariable X, the total probability of the

mutually exclusive events (intervals) within which X can assume a
value is:
(a) Less then one
(b) Greater then one
(c) Equal to one
(d) Between zero and one
Chapter 5: Continuous Probability Distributi?ns 85
..
Q.5 IfXis a uniform variate U(5, 10), then the mean of Xis
(a) 5
(b) 7.5
(c) IO
. (d) I5
Q.6 If Xis a uniform .variate U (5, Io), then the variance of Xis
(a) 0.4I 7
(b) 2.08
(c) 7.5
(d) 5
Q. 7 If mean of exponential distributio~ is 2 then its variance is

(a) 2
By: Rafaqat
(b) I
(c) 6
(d) 4
Q.8 If mean of exponential distribution is 2 then sum of such I 0

independent variates will follow gamma distribution with mean
(a) 2
(b) 10
(c) 5
(d) 20
Q.9 If mean of exponeritial distribution is 2 then sum of such IO

independent variates will follow gamma distribution with variance
(a) 2
(b) IO
(c) 20
(d) 200
2
Q.10 If X - N(,a ) and a and b are re<!l numbers, then mean of
(aX + b) is
(a) a+b
(b) a+b
(c) a
(d) a+b
'.
~
t
I
Q.11 If X - N(.c/) and a ind bare real numbers, then variance of

. (aX + b) is
(a) a +b
. (b) cl-d+b
(c) cl-il
(d) a+bd
Q.12 Which of the fQllowing is not a c~;iracteristic of normal
distribution? //
(a) The total area under the curve is equal to one.
(b) The curve is symmetric about the mean..
(c) The value of the mean is always greater than the value of
By: Rafaqat
the standard deviation.

(d) The two tails of the curve extend indefinitely.
Q.13 The total area under a normal distribution curve to the left o(the
mean is always: ,
(a) 1
(b) 0
(c) 0.5
(d) 0.9
Q.14 The tails of the normal distribution:

(a) Meet the horizontal axis at z = 3.0
(b) Nevet meet or cross the horizontal' ax is
(c) Cross the horizontal axis at z = 4.0
(d) Are asymmetric
.
Q.15 For a normal distribution, the Z value for an X value is to the rig~t
of the mean is always:
(a) Equal to zero
(b) Negative
(c) Greater than one
(d) Positive.
Chapter 5: Continuous Probability Distributions
Q.16 For a normal distribution, the mean of 40 and a standard deviation

of 8. The value of Z for X = 52 is:
(a) 2.00
(b) l.50
(c) -l.75
(d) 0.80
Q.17 If mean of the Chi-square distribution is 4 then its variance is:

(a) 2
(b) 4.
(c) I
(d) 8
Q.18 IfX follows t-distribution with v d.f. then the distribution of x2 is:
By: Rafaqat
(a) Standard Normal

(b) Chi-square with v d:f.
(c) F with I and v d.f.
(d) F with v and I d.f.
Q.19 The area under the normal curve within two standard deviation of
the mean is:
. (a) 68.26%
(b) 95.44%
(c) 99.73%
(d) 99.99%
Q.20 The distribution of sample correlation is

(a) Nonna!
(b) t.
(c) Chi-square
(d) F
Chapter S: Continuous Probability Distributions 88
-~i,
A Q1,1ick Approach to Statistics with Questions and Answers

"True" or "False":
1. If Xis. a continuous uniform random variable U{a, b) then

P(X= c) = ll(b-a) where a< c < b.
2. The inflection points are the points where of a curve convexity
changes.
3. The right and left tails of the normal curve extend indefinitely,
never toching the horizontal axis.
By: Rafaqat
4. For a onnal distribution, the mean always lies between the

mode and the median.
5. If U1 , U2 , , Un are independent Chi-square variables .with one
degree of freedom, then the. di:Stribution of U1 + U2 + ... + Un is
Chi-square with n degrees of freedom;
If Z is a standard normal random variable, then the distribution
of Z 2 is Chi-square with one d~gree of freedom.
As t}le number of degrees of freedom gro\VS, the t-distribution
approaches the normal distribution with mean 0 and variance I.
The distribution of sample variance is F-distribution.
The mean of !-distribution is zero.
,The median of !-distribution is zero.
The mode of !~distribution is its degree of freedom .
. If the expected waiting time for certain service is 5 minutes then
the variance of waiting time is also 5 minutes when we consider
the waiting ti1_11eto follow exponential distribution.
A Qulc1c Approach.to Statistics with Questions and Answu.s
13. The variance of I-distribution does not exist if the degree of

freedom is less than 2. .
14. If X follows F-distribution with v 1 and v2 d.f, then l/X has
standard normal distribution.
15. If X follows F-distribution with V\ and v2 d.f. then IIX has also
P-distribution but with v2 and vi.d.f.
By: Rafaqat
Cllapter 5: Continuous Probability Distributions 90
Chapter 6
'IJlllll
. Regression & ~orrelation
Regression
By: Rafaqat
Regression Analysis
Regression analysis .refers to the methods of describing the functional
relationship between the dependent arid independent variable. It is .used to
predict value of one variable (dependent) on the basis of other (independent)
variables.
Scatter Diagram .
When we plot data in such a way that we obtain dots showing relation
betw.een dependent and independent variables, it is called scatter .diagram.
Regression Line .
A regression line is a lirie drawn through the points on a scatter plot to
summarize the relationship between the variables being studied. When it
slopes down (from top left to bottom right), this indicates a negative or
inverse relationship between the variables; when it slopes up (from bottom
right to top left), a positive or direct relationship is indicat~d. The regression
line often represents the regression equation on a scatter plot.
lf this line is a straight line then the regression is called linear regression
while, on the other hands, it is referred as non-linear regression.
Regression Equation/Model
A regression equation (model) allows us to.express the relationship between
two (or more) variables algebraically. It indicates the nature of the
. Chapter 6: Regression & Correlation 91
relationship between two (or more) variables. In particular, it indicates the

extent to which you .can predict some variables by knowing others, or the
extent to which some ate associated with others.
A simple linear regression equation is usually written;
r =Po+ /3, x + e
where Y is the dependent variable, /J 0 is tile intercept; the average value of Y
when there is no contribution of X, {31 . is the slope or regression coefficient;
rate of change in Y when unit change in X, Xis the independent variable (or
covariate), & is the random error term.
The. equation will specify the average magnitude of the expected change in
Y gfven a change in X. The regression equation is. often represented on a
scatter plot by a regression line .
By: Rafaqat
.Assumptions of Linear Regression Model

Some important assumptions are: . .
(i) &;'s are normally distributed with zero mean and a constant variance
d.
(ii) (&; ; ) = 0 for all i f:.I i.e., error terms are independent of each other.
(iii) E(X&;) = 0 for all i, i.e., X and & are independent of each other.
Simple Regression .
Simple. regression investigates the effect of one independent variable on the
:;. dependent variable.
~
Multiple Regression
Multiple regression investigates the effect of tWo or more independent
variables on the dependent variable.
Method of Least Squares .

The method of least squares is a criterion for fitting a specified model to
observed data such that the sum of squares of the residuals (difference
between observed and estimated value) is minimized. '
The estimated model can be written as
y=pA +pA x.
0 I
Alsu y - y =e = residual.
\ -
Standard ~rror of Estimate
A measure of the scatter' of actual values of the dependent variable from
their estimated values. It can be used to set up confidence intervals for the
actual values of the dependent variables.
Total Variation
A measure of the variation of the actual values of the dependent variable
from their mean.
Symbolically,
Total Sum of Squares= TSS = L (y - ji) 2
Total Variation= Explained Variation+ Unexplained Variation.
By: Rafaqat
Explained Variation
A measure of the variation of the estimated values ofthe dependent variable
from the mean of actual values.
Symbolically,
. Explained S.Um of Squares = ESS =L (ji- ji) 2
Unexplained Variation
A measure of the variation of the actual values of the dependent variable
from the estimated values of that variable.
Symbolically,
Residual Sum ofSquares = RSS = L (y- ji) 2
Coefficient of Determination
It measures the relative amount of variation in the dependent variable that
has .been explained by variation in the independent variable. It is the
measure of strength of association that exists between variables.
It is denoted by R!-.
Correlation
Correlation Analysis
Correlation analysis refers to the methodology of measuring the
interdependence between two or more variables. For this measure,. we
compute coefficient of correlation whose value ranges from -1 to + 1. The
negative correlation shows that both the series under consideration are
moving in different directions while the positive correlation shO\~S that both
the series jointly move in the same direction i.e., either going to increase or
going to decrease. The zero correlation shows the statistical independence
of the variables. Negative correlation shows indirect relation among the
series while positive correlation shows a direct relation. Furthermore, -1 or
By: Rafaqat
+ 1 correlation shows a perfect relation among the series.

It is also to be noted that correlation coefficient is the square root of the
coefficient of determination
Multiple Regression Correlation Coefficient

The multiple regression correlation coefficient, R2 , is a measure of the
proportion of variability explained by, or due to 'the regression (linear
relationship) in a sample of paired data. It is a number between zero and one
and a value close to zero suggests a poor model. A very high value of R2 can
arise even though the relationship between variables is non-linear. The fit of
a model should never simply be judged from t\le R2 value.
Partial Correlation
A partial correla~ion is used to measure the degree of liner relationship
between any two variables in a multivariable problem by removing any
common relationship or influence with all other variables.
Transformation to Linearity . .
Transformations allow us to change all the values of a variable by using
some mathematical operation, for example, we can change a number, group
of numbers, or an equation by multiplying or dividing. by a constant or
taking the square root. A transformation to linearity is a transformation of a
response variable, or independent variirble, or both, which produces an
approximate linear relationship between the variables.
A Quick Approac/1 to Statistics with Questions and Answers
Questions-Answers
Q.l How would yo,u decide to choose regression or correlation
analysis for study of relationship?
Ans. If we want to measure of dependence of one variable on one or
more variables, we use regression analysis. On the other hand, if
we want. ,to measure the interdependence among two or more
variables, we use correlation analysis.
Q.2 ' Defme Pearson's Product Moment Correlation Coefficient

Ans. Pearson's product moment correlation coefficient, usually denoted
by r, is one example of a correlation coefficient. It is a measure of
the linear association between two variables that have been
By: Rafaqat
/neasured on interval or ratio scales, such as the relationship

between height in inches and weight in pounds. However, it can be
misleadingly small when there is a relationship between the
variables but it is a non-linear .one.
Q.3 .Defme Spearman 's Rank Correlation CoefjlclenL

Ans. The Spearman rank correlation coefficient is usually calculated on
occasions when ii' is not convenient, econQmic, or even possible to
give actual values to variables, but only to assign a rank order to
instances of each variable. It may also be a better indicator that a
relationship exists between two variables when the rela~ionship is
non-linear.
Q.4 Before correlating two series, how can you determine which will
be the dependent variable?
Ani. The dependent variable is the one for which estimates and
forecasts will be ma_de.
j
Q.5 Is the computation of coefficient of correlation is part of
!
regression analysis?
/fns. No, although a correlation analysis usually makes a regression
analysis more meaningful and useful.
Q.6 Can a coefficient of correlation be computed without computing
a regression equation?
1
Ans. Yes.
Q. 7 ls a coefficient of correlation, in any way, related to a regression

equation even though a regression equation has not been
computed?
Ans. A coefficient of correlation is always related to some regression
equation. Closer the dots on the scatt~ diagram cluster .about a
.
regression line the higher the coefficient of correlation.
Q.8 Is it important to construct a scatter diagram before computing a

regression line or coefficient of correlation, why or why not?
Ans. Yes, a scatter diagram is necessary in order to determine what type
of regression line and coefficient of correlation to compute. The
shape of the scatter of the dots on the diagram will be the
d~termine factor.
By: Rafaqat
Q.9 Under what conditions, caf! Spearman's rank correlation

coefficient be used effectively'!
Ans. Th.e Spearman's rank correlation coefficient is used for measuring
the closeness of the .relationship offactors which can be ranked in
order of importance or magnitude but for which exact
measurement cannot be made. Also data for very small samples
may be ranked and analyzed A regular coefficient of cprrelation is
not a very reliable measure for small samples because of sampling
errors.
Q.10 Why do r (Coefficient of Correlation) and P (Coefficient of

Regression) always have the same sign?
rL Ans. We know that: ,
Cov(X,Y)
r.\T =rn =
ax O'r
and
=Cov(X,Y)
/3 XY ar
2 '
= Cov(X,Y)
PYX ax
2
.,. http://stat9943.blogspot.com
A Quick Approach to Stt#istics wiili Questions and Answers
Since V(X) and V(Y) both have positive values and only the
quantily Cov(X. Y) can have a posifive or negative sign. According
to the above equations if Cov(X, Y) is positive then r an<l p are
positive and vice versa. So ii is clear that due to common
numerator i.e. Cov (X. Y) both r and fJ have the same sign.
Q.11 Define multiple regressions and multiplec;orrelations.

Ans. Multiple regression is the term used concerning the combined"
effect of two or more independent variables on one dependent
variable. Multiple correlation is the term us.ed concerning the
measurement the degree . of closeness of the combined
movement of two or more independent variables and the
movement of the ane dependent variable.
By: Rafaqat
Q.12 What are the net regression coefficients?

Ans. A net regression coefficient measures the effect of one independent
variable on the .dependent variable when the effects of other
independent variables are held constant.
Q.13 What does a coefficient of multiple determination measure?

Ans. A coefficient of multiple determination measures the proportion of
the total variation in the dependent variable that has been
explained by the combined movements of the several independent
variables.
Q.14 What does a coefficient of multiple correlation measure?

Ans. The coefficient of nJU/tiple correlation is the square root of the
coefficient of multiple determinations. It measures the degree to
which the movement of the several independent variables is
synchronized at the same time with the movement of the dependent
variable.
Q.15 What does a coefficient ofpartial correlation measures? .

Ans. A partial coefficient correlation measures the degree of the net
influence of the movements of an independent variable on the
movements of the dependent variable.
Chapter 6: Regression & Correlation 97.
A Quick Approach to Statistic~ with Questions and Answers
Q.16 Of what use is a standard error of estimate in a multiple

regression analysis?
Ans. The standard error of estimate indicates the reliability of the
estimates made with the estimating equations. It can be used to
calculate prediction intervals for actual values of the dependent
variable.
Q.17 ffnder what condition will tire net regression coefficient be

smaller titan their corresponding simple regression coefficients?
Ans. _The net regression coefficients will tend to be smaller when there is
a positive simple correlation between the two independent variable
in a three variable multiple regression analysis.
Q.18 How is'the sign (+pr-) determines for a coefficient of partiI

By: Rafaqat
)
correlation?
Ans. The partial correlation coefficient receives the same sign as its
corresponding net regression coefficient.
Q.19 A sign should not be afftxed to a coefficient of multiple

correlation. Why? .
Ans. A sign should not be a.ffvced because some independent variables
may be positively correlated with the dependent variables while
other are negatively correlated.
Q.20 Wiry are two vari{lbles are sometimes closely related?

Ans. The two variables may be closely related because changes in one
causes in changes in another or because both variables are
affected similarly by other forces. A close correlation can also
occur by chance.
Q.21 If twQ variables are closely correlated, then tire movements in one
variable cause the movements in another. Why or why not?
Ans. No, there is no proof of causation in correlation theory. A
correlation analysis. never proves or disproves that there is
relationship between two variables. All it does is measure the.
closeness of the relationship.
Q.22 What is tire interpretation if coeffteient of determination R1 =

0.7?
Chapter 6:,Regression & Correla~'9;,n 98
Ans. It means that 70% variation of the dependent variable can be

explained by the regression. Hence its shows relatively good fit of
regression model.
Q.23 What values does r assume if all the sample points fall on the
. same straight line and if the line has:
(a) a positive slope (b) a negative slope.
Ans. (a) +I (b) -1.
Q.24 A police research cell has shown that the crime rate is cocrelated
with the number of unemployed people in Pakistan. Would you
expect the correlation to be positive or negative?
Ans. Positive, because as number of unemployed people increases, the
crime will also increase.
By: Rafaqat
Q.15 Is the demand/or a product correlated with price? If it is, would

you expect the correlation to be positive or negative?
Ans. Yes; Negative.
Q.16 Do you .believe that an exact relationship exists between two

variables?
Ans. In reality, the answer is very definite no,
Q.27 What is the general form of a probabilistic model?

Ans. =
Y Deterministic component + Random error. The mean value of
the random error term equals zero. This is equivalent to assuming
that the mean value of Y. i.e. E(Y) = Dete~ministic component.
Q.28 Give any example for the use ofpartial correlation.

Ans. Correlation between amount of additive oil and mi/eag covered
by vehicle while removing the effect of age ofthe engine.''..
Q.19 Define the straight line probabilistic model

Ans. The equation of the straight linecrs:
Y =a+ bX + e,
Where Y = Variable to be predicted, called dependent variable and
X = independent variable. . .
E(Y) =a+ bX. is the deterministic porti'on of the model.
A Quick Approach to Statistics with Q~estions and Answers
a = Y- intercept of the line and b = siope of the line, i.e. rate of

change of in the deterministic component of Y for every I unit
increase in X
Q.30 Show that the coefficient of correlation is the geometric mean of

the coefficient of regression.
Ans. Regression coefficient ofYonXis: byx = ra.
. O"y
ra
Regression coefficient ofX on Y is: hxy = __
Y
. . a.
ra ra
G.M of Regression coefficie~ts:.. ~b,,.b.,, = --y .-- =r
a. ay
By: Rafaqat
Q.31 ls this statement correct? Give reasons, "The regression

coefficients of X on Y is 3.2 and that of Yon Xis 0. 8.
Ans. No, it is not correct.
Here byx = 0.8 and bxy = 3._2
r_=M
= ./0.8x3.2
= .J2.s6
= 1.6
Since - I S r S I . Therefore lrl I. 6 is i~possib/e, hence the
given statement is wrong.
. Q.32 Whether it is true that if one of the regression coefficient is

greater than unity, the other must be less than unity?.
Ans... Yes, it is true. Suppose that one of the regression coefficients say
byx > 1 then JI byx < /, also? S 1 => byx bxy :s; I => byx S /.
Q.33 If two regression coefficients are 0.8 and 0.2. What would be the
value of coefficient of correlation.
Ans. r1 = byx. bxy = (0.8) (0.2)~ 0.16,
r = 0.4, positive since byx and bxy both will be positive.
A Quick Approaclr to Statistics with Questions and Answers
Exercises
Exercise 6 (MCQs')
Q.1 The major difference between regression analy~is and correlation

analysis is that in regression analysis:
(a) The independent variable is known without error.
(b). Both variables are random variables.
(c) The I-distribution is used for hypothesis testing.
By: Rafaqat
(d) Only the slope is important.
Q.2 The term homoscedasticity~fers to:

(a) Equal variances fot X and Y.
(b) A normal bivariate distribution only.
(c) Equal variance of the dependent variable, Y, for any value
of independent variable, X
(d) Uniform variance of the independent variable.
Q.3 The sum of squares of which type of deviations is minimized by

the least square regression:
(a) Deviation perpendicular to the line.
(b) Deviation parallel to the line.
(c) Deviations of the values of the dependent variable from
the line.
(d) Deviations of values of the independent variable from the
line.
Q4 A random sample of paired observations has been selected and the

sample correlation coefficient has been found to be -1. From this
result we know that:
(a) At least one sample observation does not lie on the
sample regression line.
(b) . All sample observations lie on the sample regression line
Chapter 6: Regressi<fn &: Correlation 101
. A QuiC:: Approach to Statistics with Questions and Answers
(c) At least one sample observation lies on the population

Regression.
(d) None of above
,Q.5 The output of a certain chemical-processing machine is linearly

related to temperature. At -l0Cthe processor output is 200 Kgs.
per hour and at 40C the output is 220 Kgs. per hour. Calculate the
linear equat,ion for Kgs. per hour of output ( }') as a" function of
temperatUre in degree Celsius (X):
(a) Y= 193.3 + 0.61X (b) Y= 204.0 + 0.40X
(c) Y,;,, 29.0.0 +l.50X (d) Y= -510.0 + 2.50X
Q.6 What iqformation is given by a value of the coefficient of .

determination?
By: Rafaqat
(a) Strength ofrelationship

(b) Both strength and direction of relationship
(c) Neither strength nor direction of relationship
( d) Direction of relationship only
Q.7 For the regression equation Y=IO + 2X, the Y intercept is:
(a) 10 (b) 2 (c) 0 (d) -2
Q.8 The correlation coefficient:

(a) Can be calculated only after regression analysis . is
performed.
ii;. (b) May be smaller than -1 only when X and Y are inversely
related.
(c) Equals the positive or negative square root of the
coefficient of determination.
(d) Equals the standard error of the estimate divided by the
. square root of the sample size _
Q.9 If the coefficient of determin.ation is 0.49; the correlation
coefficient may be:
(a). 0:51 (b) . 0.49 (C) 0.24 (d) 0.70
Q.10 The estimated regression .line relating the market value of a

person's stock portfolio to his annual income is Y = 5,000 +O.IOX.
This means that each additional rupee of income will increase the
. stock portfolio by:
(a) Rs.0.50 (b) Rs.1.00

(c) Rs.0.10 (d) Rs.10.00
Q.11 Which one of the following situfltions is inconsistent?

(a) Y == 500 + O.Ql X, arid r = 0.75
(b) Y -200 + 0.9X, and .r = -0.86
(c) Y 10 + 2 X, and r = 0.50 '
_(d) Y -8 3 X , and r ,;,, -0.9S
Q. 12 Which one of the following statement is true?

(a) The estimated and the true regression lines are always the
same.
(b) The rs must be normally distributed about true regression
line before the sample coefficient of determination can be
By: Rafaqat
calculated.
(c) The units in which X and Y are measured will hot affect
the value of r.
(d) The correlation coefficient can be calculated only after the
estimated regression line has been found.
Q.13 The true correlation coefficient p will be zero only if:

(a) The Y intercept of the true regression line is equal to zero
(b) The slope of the true regression line is equal to zero
(c) r 0
(d) b = 0
Q.14 _Whenever predictions are made from the estimated regression line,
the relation between X and Y is assumed to be:
(a) Direct (b) Inverse
(c) Linear (d) Perfect
Q.15 . The estimated coefficient of determination- is equal to all except

which one of the following?
{a) The square of the correlation coefficient.
{b) The proportion of variation in Yexplained by regression.
(c) .I min1;1s the proportion of variation in Y unexplained by
regression. .
(d) The slope of the estimated regression line.
Chapter 6; Regression & Correlation 103
Q.16 The coefficient of partial determination differs from the coefficient

of multiple determination in that:
(a) One provides. the slope of the regression plane and the
other does not.
(b) One may be pegative, but the other is always positive.
(c) Both of the above.
(d) None of the above.
Q.17 The coefficient of multiple determination is 0.81. Thus, the

multiple correlation coefficient is:
(a) O.I.9 (b) 0.9 (c) 0.6561 (d) 0.1
Q.18 A larger sample size can be expected to achieve all qut which one
By: Rafaqat
not of the following?

(a) A smaller value for standard error ofregression.
(b) Increase degrees of freedom.
(c) An estimated regression plane that is .closer to the true
regression plane.
(d) Increase in the value of coefficient of determination.
Q.19 In multiple linear regression analysis, the square root of Mean

Squared Error (MSE) is called the:
(a) Multiple correlation coefficient
(b) Standard error of estimate
~ ( c) Coefficient of determination
(d) Coefficient of non-determination
Q.20 Iii multiple regression analysis, the purpose of solving the normal
equations is to find:
(a) The standard error of estimate.
(b) The constant and coefficients in the least squares
relationship.
(c) The number of independent variables in the least squares
relationship.
(d) The variance around the least squares relationship.
Chapter 7
Sampling
Basics of Sampling
By: Rafaqat
Population _
A group of individuals or entities about which you wish to know something
and from which a sample will be taken, is called population.
Sampling Unit
This is a single member of a population; e.g., if the population is defined to
be l 00 trees on a lot, then the sampling unit is a single tree.
Sample . .
A sample is a sub-collection of elements drawn from a population; a subset
of a population (a collection of Sampling Units), with the assumption that it
represents the whole population. -
Sampling _
The procedure by which a few subjects are chosen from the universe
(population) to be studied in such as way that the sample can be used to
estimate the same characteristics in the total is referred to as sampling.
Sampling Frame
A iist of all the sampling units. It includes lists that are available or that are
constructed from different sources specifically for the study. Directories,
membership or customer lists, even invoices or credit card receipts can
serve as a sampling frame. ' .
Sample Design
A program layout including all the procedures to take a sample.
Parameter
A summary, typically numerical, of a variable (or variables) over the entire
population.
. I
Statistic
A summary, typically numerical, ofa variable (or variables) over a sample .
.Statistical Inference
The process of making a statement about a population on the basis of
sample infol1!1ation.
By: Rafaqat
~xample:
In an example to find out the average number of cups of tea that office
worker take daily in a. particular city:
Population all the office workers in that city

Sa1J1pling Unit an individual worker
Sampling frame list of all the office workers obtained from all
the offices of the said city
'Design probability sampling (discussed later)
Sample 200 (say) office workers
Data gathered . the daily tea consumption (number of cups). of
each of the 200 workers selected in the sample
Statistic the average number of tea cups/day of the 200
workers selected in the sample
Parameter the average number of tea consumption
(cups/day) the office workers obtained from all
the offices of the said city
Census
the method that collects data from all members of the population, rather
than from a selected subset ~fthe population.
A Quick Approach to Statistics with Questions aatl ..4JmNls
Sampling Technlques
Non-~robability Sampling Techniques

In non-probability sampling, the sample is selected in such a way that'the
chance of being selected of each unit within the population or univen;e is
unknown. Indeed, the selectiori of the subjects is arbitrary or subjective,
sine~ the researcher relie's on his/her experience and judgment.
By: Rafaqat
There are five main types -0f non-probability sampling that we will review
more closely:
Purposive Sampling
.Convenience Sampling
Quota Sampling
Judgment Sampling
Snowball Sampling
Purposive Sampling
In purposive sampling, the researcher selects the units with some.purpose in
mind, for example, students who live in dorms on campus, or experts on
urban development. This is where. the researcher targets a group of people
believed to be typical or average; or a group of people specialty picked for
some unique pqrpose. The researcher never. knows if the sample is
representative of the population, and this method is largeiy limited to
exploratory research. Sampling of cricket players, sampling of dresses in a
market, and sampling of Jewelry etc., are few examples of puq,osive
sampling.
Convenie~ce Sampling .
A convenience sample is used when a researcher simply stop anybody in
the street who is prepared to stop, or when a researcher wander round a
business, a.shop, a restaurant, etc. and asking people he meet whether they
_Cliapter 7: Sampling 107
will answer his questions. In other words, the sample comprises subjects
who are simply available in a convenient way to the researcher. There is no
randomness and the likelihood of bias is _high. One can't draw any
me<lningful conclusions from the results h~ obtains.
Quota Sampling
In quota sampling, the researcher constructs quotas for different types of
units. For example, to interview a fixed number of shoppers at a mall, half
of whom are male and halfofwhom are female. It is widely used in opinion
polling and market research. Interviewers are each given' a quota of subjects
of speci~ed type to attempt to recruit e.g., an interviewer '!flight be told to
go out and select 20 working men and 20 working women, 10 school girls
and l q school boys so that they could ,interview them al?out their television
viewing.
By: Rafaqat
Judgement Sampling
In judgement sampling, the researcher or some other "expert" uses his/her
judgement in selecting the units from the population for study based on the
population parameters.
This type of sampling technique might be the most appropriate if the
population to be studied is difficult .to locate or if some members are
thought to be better (more knowledgeable, more willing, etc.) than others to
interview.
Snowball Sampling
It is also called network, chain, or reputational sampling. This method
begins with a few people or cases and then gradually increases the sample
size as new contacts are mentioned by the people you started out with.
A Quick Approach to Statistics with Questions and Answas
Probability Sampling Techniques

In probability sampling, the sample is selected in such a way that each unit
within.the population or universe has a known chance of being selected. It
is this concept of "known ch~nce" that allows for the statistical projection of
characteristics based on the sample to the population. .
Following are the m~in types of probability or random sampling:
Simple Random Sampling

Stratified Random Sampling
Systematic Random Sampling
Cluster (Area) Random Sampling
Simple Random Sampling

By: Rafaqat
A sampling procedure that assures that each element in the population has
an equal chance of being selected is referred to as simple random sampling.
Let us assume you had a school with a IoOO students, divided equally int()
boys and girls, and you wanted to select 100 of them for further study. You
might put all their names in a drum and then pull l 00 names out. Not only
does each person have .an equal chance of being selected, we can also easily
calculate the probability of a given person being chosen, since we know the
sample size (n) and the population size (N) and it becomes a simple matter
of division:
n/N x 100 or 100/1000 x 100 = 10%.
This means that ever.y student in the school as a 10% or 1 in 10 chance of
being selected using this. method.
Stratified Random Sampling

Stratified sampling techniques are generally used when the population is
heterogeneous, or dissimilar,. where certain homogeneous, or ~imilar, sub-
populations can be isolated (strata). In this random sampling technique, the
whole population is divided first into m~tuafly exclusive subgroups or strata _ -i
and then units are selected randomly from each stratum. The segments are
based on some predetermined criteria such as geographic location, site or
demographic characteristic. This method is appropriate when you're
interested in correcting for gender, race, or age disparities in your
population. It is important that the segments be as heterogeneous as
possible.
Systematic Random Sampling _

In such sampling, each unit in the population is identified, and each unit has
an equal chance of being selected in the sample. For example, to select a
sample of 25 shops in a market, make a list of all theshops. Say there are
1OQ shops, number these shops form I - I 00. Divide the total number of
shops (100) by the number of shops you want in the sample (25). The
answer is 4.
N= 100
n=25
K = NI n = 100/25 = 4 (Sampling Interval). .
This means that you are going to select every fourth shop frotfftthe list. But
you must first consult a table of random digits. Pick any point on the table,
By: Rafaqat
and read across or down until you come to a digit between 1 and 4. This is
your random starting point. Say your random starting point is "3". This
means you select shop 3 as your first shop, and then every fourth shop down
the list (3, 7, 11, 15, 19, etc.) until you have 25 shop selected.
Cluster Sampling
Cluster sampling. is a sampling technique where .the entire population is
divided into a number of heterogeneous groups, or clusters, and a random
sample of these clusters is selected. All observations in the selected clusters
are included in the sample.
Contrary to simple random sampling and stratified sampling, where single
subjects are selected from the population, in cluster sampling the subjects
are selected in groups. or cJusters. Suppose, you have a population that is
dispersed across a wide geographic . region. This method allows you to
divide this population into clusters (usually counties, census tracts, or other
boundaries) and then randomly sample everyone in those clusters. This
approach allows to overcome the constraints of costs and time associated
with a very dispersed population .. Cluster sampling views the units in a
population as not only being members of the total population but as
members also of naturally-occurring in clusters within the population. For
example, city residents are also residents of neighborhoods, blocks, and
housing structures.
Chapter 7: Sampling . 110
Bias and Errors
Bias and Sampling Bias

Bias is a term that refers to how far the average statistic lies from the
parameter it is estimating, that is, the error that arises when ~stimati~g a
quantity stated differently it is the difference between the expected value of
an estimator and the true population value being estimated. It is also known
. as the systematic component of error. Errors from chance will. cancel each
other out in the long run, those from bias will not.
For a given estimator, the difference between the expected value of the
estimator based on a.sample and the estimate that would result if the sample
By: Rafaqat
were to include the entire population. i.e.,

If iJ is an estirn!ltor of () then
Bias= E(B)-0.
Standard Error
The standard deviation of the sampling distribution tells us something about
how different samples would be distributed. It is referred to as the standard
error.
e Sampling Error
s For a given estimator, the difference between an estimate based on a sample
s and population parameter.
()
:r
is
Non-Sampling Error
d For a given estimator, the difference between the estimate that would result
a if the sample were to include the entire population and the true population
value being estimated.
IS
>r
ld Coverage Error
Error due to omissions, erroneous inclusions, and duplications of units in
the frame us.ed to conduct the survey; also, for housepold surveys, any
omissions or duplicates wi~hin the householcJs.
.chapter 7: Spmpling 111
Non-Response Error
Error caused by survey failure to get a response to one or possibly all of the
questions. Indirect measures include the detail disposition rates . (un-
weighted and weighted) of all the selected sample c~ses during data
collection. Direct measures may require non-response follow-up.
'
Measurement Error.
Error when the response received differs from the. "true" value due to the
respondent, the interviewer, the questionnaire, the mode of collecti9n, or the
respondent's record-keeping system(s):
Processing Error
By: Rafaqat
Error during data editing, coding, capture (keying and scanning),

imputation, and tabulation.
.Estimation Error
For a given estimator, the difference between the value of the estimate anq
the true population value being estimated. Includes both sampling and non-
sampling error.
Chapter 7: Sampling , 112
!TS A Quick Approach to Statistics with QuestiOns and Answers
Questions-Answers
he
1n-
1ta Q.J . What is the representative sample? ,
Ans. A sample is "representative" if the distribution of the- sample's
characteristics is the same, on average, as the distribution of the
characteristics of the population. The size of the sample and the
he type of sample (random er purposeful) are key decisions if you
he wan_t to say that your sample represents the population.
Q.2 Why are samples studied?

Ans. Samples are studied to:
(a) Save time and money
By: Rafaqat
~).
(/?) Save items that must be injured or destroyed in the
process ofstudying their characteristics.
(c) Samples are also studied because it is also the only
practical way t() ~btain information about a universe
nd because of its large size. '
1n-
Q;J Explain tire difference between a sample and a census.
Ans. A census is a survey that attempts to include every elemeni in the
population while a sample is a partial enumeration ofpopulation.
Q.4 What is target population?

Ans. The target population is the entire group, a researcber is
. , interested in; the group about which the researcher wishes to draw
conclusioflS.
Q.5 JJlli~t is sampled population?

Ans. A conceptual population that is a subset of the intersection of the
Target Population and the Sample Frame. A population from
which the sample is chosen.
Q. 6 Define sampling frame.

Ans. The list of people from which the sample is taken. It should be
comprehensive, complete and up-to-date. Examples of sampling
frame are electoral register, postcode address file, telephone
12 Chapter 7: Sampling 113
directory etc. In other words the sampling frame is the list of

ultimate sampling entities. .
Q.7 Differentiate sampling with replacement and sampling without

replacement.
Ans. Sampling with replacement. is a method of sample selection in
which a sample is obtained by first selecting one sampling unit
from the sampling frame, replacing it, then making a second
selection and replacing it before making a third selection, etc.,
until n selections have been made, A particular unit could be
included more than once in the sample and possibly. up ton times.
Sampling without replacement is a method of sample selection in
which a sample is obtained by selectil:ig one sampling'unit from the
sampling frame and, without replacing it, selecting one of the
By: Rafaqat
remaining sampling units; then continuing this process until n

different selections have been made: A unit can be included only
once in any sample.
Q.8 ls there any difference between the terms; Sampling. Technique,

Sampling Design and Sampling Model?
Ans. There is no difference between all these terms. All these are
conveying the idea of sampling strategy .we are adopting. These
techniques include simple random sampling, stratified sampling
and all those discussed above.
Q.9 What is sample size? .

Ans. A size of sample determined in relation to' the required precision
and available budget for observing the selected units.
Q.10 Define an error. }

Ans.. An error is the amount by which an observatior:z differs from its
expected value,
Q.11 Distinguish clearly the difference between systematic error,s and

random errors. Explain which type of the error will decrease with
the larger sample size, which will not, and why. Which error can
and should /Je eliminated?
Ans. Random error is attributable to chance fluctuations of sampling
and tend to decrease, as sample size increases. On the other hand,
Chapter 7: Sampling II4
A Quick Approach to Statistics with Questlonsoand Answers
r systematic errors are primqrily attributable to faulty experimental

design and' measurement. Since systematic errors are not
associated with the number of the elements sampled, they don't
decl'ease as the sample size increase. Systematic error can and
should be eliminated
Q.12 Define sampling variability?

Ans. Sampling variability refers to the different values, which a given
function of the data takes when it is computed for two or more
.samples dri:Jwnfrom the same population.
Q.13 Establish the idea ofprecision. .

Ans. Precision is a measure afhow elose an estimator is expecled to be
to the true value of a parameter. Precision is usually expressed in
By: Rafaqat
terms of imprecision and related to ihe standard error of the

estimator. Less precision is reflected by a larger standard error.
Q.14 In what situations matched samples arise?

Ans. Matched samples can arise in the following situations:
(a) Two samples in which the members are Clearly paired, or
are matched explicitly by the researcher. For example, IQ
measurements on pairs of identical twins.
(b) Those samples in which. the same attribute, or variable, is
measured twice on eac,h subject, under different
circumstances. Commonly called repeated measures.
Examples include the times of a group of athletes for
I 500m before and after a week of special training; or the
milk yields of cows before and-after being fed a particular
diet.
Q.15 Define sampling distribution?

Aris. The distribution of an infinite number of samples of the same size
as the sample in your study is known as the sampling distribution.
Q.16 What do you know about independent sampling?

Ans. Independent samples are those samples selected from the same
population, or different populations, which have no effect on one
another. That is, no correlation exists among the samples.
Chapter 7: Sampling '115
Q.17 What are different types have_ non-probability sampling?

Ans. Purposive sampling, - quota sampling, convenience sampling,
snowball sampling, etc. are the main types of non-probability
sampling.
Q.18 What are the advantages of non-probability sampling?

Ans. The main advantages ofnon-probability sampling are;
Cheaper
Used when sa'mplingframe is not available
~-
Useful when population is so widely dispersed that cluster
I -
sampling would not be.efficient
t Often used in exploratory studies, e.g. for hypothesis
By: Rafaqat
generation
Some research not interested in working out what proportion -
. of population gives a particular response but rather - in
obtaining an idea of the range of responses on ideas that
people have._
Q.19 . What is the major disadvantage of convenience sampling?

Ans. The major disadvantage of this technique is that we have no idea
how representative the information collected about the sample is to
the population as a whole. But the Information could still provide
some fairly significant insights, and be a good source of data in
exploratory research.
Q.20 What is the main disadvantage of quota sampling?

Ans. The problem with it is that bias introduces on the sampling frame.
Once the researcher identifies the people to be studied, th'ey
have to resort to haphazard or accidental sampling because no
effort is usually made to contact people who are difficult to reach .
in the quota.
Q.21 What are different types have random or probability sampling?

Ans. There are some types of probability sample. The choic.e of these
depends on nature of research problem, the availability of a good
sampling frame, money, time, desired level of accuracy in the
sample and data collection methods. Each has its advantages and
CIUipter 7: Sampling 1i6
A Quick Approach to Statislics with Questions and Answers
disadvantages. They are: simple random sampling, stratified

. random sampling, systematic sampling, cluster sampling etc.
Q.22 Why random samples are always preferred?.

Ans. Random samples .are always strongly preferred because only
random samples permit statistical inference. That is, there is no
way to assess the validity ofresults of non-random samples.
Q.23 Which would be the better sample of universe; a nonrandom

sample of 50% "of the universe or a random sample of 1% of the
universe? Discuss and justify your answer.
Ans. A random sample of 1% of the universe would usually be a more
representative sample. The more heterogeneous the universe, the
more likely that this would be true. If a universe is quite
By: Rafaqat
hpmogeneous a 50% nonrandom sample may give representative

results.
Q.24 What is the difference between the random selectionand random

assignment?
Ans. Random selection is how you can draw the sample of people for
your study from a population. Random assignment is how you can
assign the sample. that you draw to dif/erent groups or treatments
.in your study.
Q.25 In what situations, there is tlie main application of simple .,

random sampling? ;;
,.~
Ans. Simple random sdmp/ing is most appropriate when the entire (
population from which the sample is taken is homogeneous.
Q.~ Define stratification.

Ans. A division of the sampling frames into subsets (called strata)
before the selection of a sample within each of the subsets; for
statistical efficiency, for production of estimates by stratum, or for
convenience. Stratification is done such that each stratum contains
Units that, are relatively homogeneous with respect to variables
that are belfev.ed to be highlY, correlated with the ,information
requested in ihe survey.
Chapter 7:'1aml!ling 117
..
A Quick tfpproach to SJatistlcs with Questions and Answers
Q.17, What are stratifying variables? ' .

Ans. The variables whose joint values are u~ed to classify a sampling
frame into several classes (called strata) from each one of which a
sample is drawn independently. Both numerical and non-numerical
variables may be used for creating strata. Variables from outside
sources, such as administrative records, censuses, etc., may be
used as stratifying variables.
Q.18 What are the main reasons for using stratified sampling over
simple random sampling? ..
Ans. Some reasons for using stratified sampling over simple random
sampling are:
(a) The cost per observation in the survey may be ~educed
(b) Estimates of the population parameters may be wanted
for each sub-population; increased accuracy at given .
By: Rafaqat
cost.
Q.19 What is sample size allocation?

Ans. The method in determining, how the sample should.. be
distributed. In stratified sampling, it usually refers to the
determination of the Uf!its selected from each stratum. In cluster
sampling, it refers to the decision as to the number ofclusters to be
selected and the size of the sample in each cluster.
Q.30 What are different methods used for the allocaticn of sample
size?
Ans. Proportional allocation, optimum allocation and Neyman
allocation are usedfor sample size selection.
Q.31 Define proportional allocatio11, optimum allocation and Neyman

allocation
Ans. Proportional All~cation:
Allocation in which the ratio of the number of sampled sampling
units to the total number of sampling units is the same for each
stratum.
Qptimum Allocation:
In stratified; sampling with a /in.ear cost function, allocation in
which the variance of the estimatea mean or total is minimized for
a specified cost or the cost is minimized for a specified variance of
ihe estimated mean or total. For a given stratum, the sample ~ize is
A QuickApproach to Statistics with Questions and Answers
proportional to the product of the . stratum's total number of

sampling units and standard deviation, and it is i'nversely
proportional to the square root of the cost per unit for the stratum.
Neyman Allocation: .
A special case of optimum allocation in which the var.iance of an
estimated mean or total is minimized/or. a specified total sample
size.For a given stratum, the amount of the total sample size that
is allocated depends on the relative size of ih._e product of the
stratum's total number ofsampling units and standard devfatibn.
Q.32 In what situations, there is the main application of systematic

random sampling?
Ans. This method is useful for selecting large samples, say J00 or more.
It is less cumbersome than a 'simple random sample using either a
By: Rafaqat
table of random numbers or a /ott~ry method For example, you

might have to sample files in a large filing cabinet. It is easier to
select every 17th file than to pull out all the files and number them,
etc.
Q.33' In what situations, there is the main application of cluster

sampling?
Ans. Cluster sampling is typically used when the researcher cannot get
a complete list of the members of a population they wish to study
but ciilt get a complete /(st ofgroups or 'clusters' ofthe population..
Q.34 Why you might use a cluster sample of the households in an area
rather than a simple random sample drawn from a directory
giving addresses of houseliolds?
Ans. A cluster sample of household might be used because directories
with the address of the hous~ho/d are never 100% accurate.. By t!u;..
time the directory information is obtained and published, people
have J1ZOved in an out of the area, which results in the population
different from that listed in the directory.
Q.35 Why you might use a cluS,er sample of the households in an

area t:atherthan. a simplerandom sample drawn from a directory
giving addresses of household?
Ans. A cluster sample of household might be used because directories
with the address of the household are never 100% accurate. By the
time the directory information is obtained and published, people

have moved in and out of the area, 'which results in the population
in
different from that listed the directory.
Q.36 How you cal! proceed while data collection using multistage
sampling?
. Ans. In multistage sampling the researcher divides the population into
groups, samples the groups, then stratifies the samples, and then
resample, repeating the process until the 'ultimate sampling units
are selected at the last of the hierarchical levels. When the strata
are geographic units, this method is sometimes called area
sampling. For instance, at the top level, states may be sampled
(with sampling proportionate to state population size); then cities
may be sampled; then schools; then classes; andfinally students. A
sample of clusters is selected and then a sub sample of units is
By: Rafaqat
selected within each sample cluster. If the sub sample bf units is

the last stage ofsample selection, it is called a two-stage design. If
the sub sample is also a cluster from which units are again
selected, it is called a three-stage design, etc.
Q.37 Differentiate cluster sampling and multistage sampling.

Ans. Cluster sampling is where all subjects at the lowest hierarchical
level (e.g. all students in. a school) are sampled for each primary
sampling unit (PSU), which are the second-lowest hierarehical
level, such as. schools or census blocks), whereas multistage
5ampling is where only a random sample of lowest .hierarchical
level subjects are selected The greater the heterogeneity of the
group and the finer the grouping (that is, th.e smaller the clusters
involved) depending on the objectivity of study, the more the
precision of the rest.tits. For instance, grouping by gender at the
highest level might well introduce bias in measuring opinions
about an item known lo be gender-related, whereas grouping by
state would be less likely to introduce a bias since there are more
categories (more states than genders) and there is less likely to be
a correlation with the opinion item.
Q.38 Define primary sampling units.

Ans. Units that are selected first. Which are clusters of reporting units
from which there is sub sampling to obtain reporting units in a
multistage sample.
i '
Q.39 Define self-weighting sample.

Ans. A sample in which every sampling unit on the sampling frame has
the same chance of selection, although unequal prebabilitie'S may
have been used at various stages ofsampling.
Q.40 Define double sampling.

Ans. A method of sample selection in which a sample is obtained b;
selecting a large sample in the first phase and then a sub sample of
"the first-phase is selected in the second phase. Information needed
for sample design or estimation is collected from the large firs/-
phase sample and then used in the design of the second-phase
sample or in the final estimation.
By: Rafaqat
Q.41 What do you know about spatial sampling?

Ans. A sampling technique concerned with sampling in two (or more)
dimensiorz.s. For example, sampling offields or other planar areas.
Q.42 On Friday 18th April, a personal interview oj 1000 men,

shopping in a mall during the hours at 5.00 p.m. to 9.00 p.m.
showed that the 65% of these men thought there was too much
news bulletins on the television. Do you think that the both
sampling and systematic errors are present in the results of this
study? Why or why not.
Ans. The systemr;itic error is present because most the news bulletins are
telecasted on Friday between 5.00 p.m. and 9.00 p.m. Therefore
there is probably an over representation of the men who are not very much
interested in the news bulletin in the survey. The random error is
attributable to chance fluctuations in the men interviewed
Q.43 Explain and criticiie each part of the following statement: The
frequency distributions of the family income, size of the business,
and salaries of the skilled employees all tends to be skewed to the
right. .
Ans. Agree, since in al/. three cases it is quite logical to assume that
there would be extreme values at the upper end of the scale that
would tend to make the ar(thmetid mean larger than the median.
Q.44 Whalis survey design?

Ans. The process of selecting sites at which a response will be
determined Includes a probability model for inference based on
the randomized selection process.
Q.45 Outline the steps involved in planning a sample survey.
Ans. The steps are:
{j) Define and state the purpose ofthe study

. (ii) Determine the conditions under which the survey will be
made ,
(iii) Find out if any or part of the desired information is
already available
(iv) Chose the size and the type of the sample to be used and
the method ofselecting sample items and collecting data.
By: Rafaqat
(v) Prepare the questionnaire

(vi) Collect the data
(vii} . Edit the returned questionnaire
(viii) Tabulate and classify the information gathered
(xi) Analyze the information
(x) Interpret the findings
Q.46 What is meant by data collection?

Ans. Any process whose purpose is to acquire or assist in the
acquisition of data. Collection of data is achieved by requesting
and obtaining pertinent data from individuals or organizations via
an appropriate vehicle.
Q.47 Explain the difference between an observational study and a

controlled study. -
Ans. An observational study essentially involves examination of the
historical relationships that exist among' the variables of interest; a
controlled study involves controlling or fixing factors other than
the variable of interest
Q.48 What is Questionnaire?
Ans. A. set of questions designed to collect information from a
respondent. A questionnaire may be interviewer-administered or
respondent-completed, using paper-and-pencil methods for data
collection or computer-assisted modes of completion.
Q.49 'list ten rules that are useful/or making out questi~nnaire.
Ans. (i) Use items that can be easily understandable
(ii) Avoid ambiguous questions
(iii) Make sure that the questions asked can b(! accurately
answered
(iv) Avoid double questions
(v) Avoid the direct embarrassing questions
(vi) Word the questions so that the answer can be easily
tabulated and easily classified
(vii) Avoid the leading questions
(viii) list the question in a logical sequence
(ix) Make the questionnaire short and attractive
(x) Place the research organization's name and address on
By: Rafaqat
each questionnaire.
Q.50 If a .questionnaire asks questions that. are understood !Jy the

respondents. The answers to the questions are considered reliable
and useful? Explain.
Ans. Not necessary, some people may fail to givecwrect answers even
though they understand the questions. Also, the summarized
answer may be of little value if the respondents are not from a
probability sample ofa well- defined universe.
Q.51 What is response rate?

Ans. The percentage of the intended sample from which one may
. actually able to collect data. For example, if JOO health
professionals are sampled and sent them all, questionnaires, but
only 40 returned completed questionnaires, the response rate is.
40%.
Q.52 Define non-response. .

Ans. The failure to obtain responses or measurements for all sample
elements.
Q.53 What are different types of non-response?

Ans. Item .non-response, partial non-response and unit non-response
are the main types ofnon-response.
Q.ff Define item non-response, partial non-response and unit non-

response.
Ans. Item Non-response:
It .occurs when a respondent provides some, but not all, of the
requested information, or if th.e reported information is not
useable.
Partial Non-response:
A partial intervtew is when some but not all items have
responses. A partial interview is treated as a "unit response" when
a sufficiently accurate response is obtained for only some of the
data items required from a respondent and meets some minimum
threshold level. A partial interview is treated as a "unit non-
response" when this threshold is not met.
By: Rafaqat
Unit Non-response:
It occurs when the sampled unit response does not meet a
minimum threshold and is classified as not having responded at
all; failure IQ make measurel!lents or obtain observati<Jns on a
listing unit selected/or inclusion in a sample.
Q.55 What do you know about sub sampling non-respondents?

Ans. An economical method for reducing non response bias in which
new attempts are made to obtain responses from a sub sample of
sampling units that did noi provid~ responses to thefirst attempt.
Q.56 Define coverage, over coverage and under coverage.

Ans. Coverage:
The extent to which a frame includes all the elements of the
sampled population.
Over coverage:
The extent to which a frame includes more element$ than the
sampled population; including duplicate elements.
Under coverage:
The extent to which a frame irlcludes fewer elements than the
sampled population.
Exercises
Read the following statements carefully and !ndicate which statement is
"True" or "False":
1. There are many practical situations where sampling is

By: Rafaqat
inevitable that is there is no other way of data collection other

than sampling.
2. There is no differ~nce between random error and sampling
bias.
3. When the items included in a sample are baSed on the
judgment of the individual conducting the sample, the saniple
is said to be nonrandom sample.
4. A statistic is a characteristic of population.
5. A sampling plan that selects members form a population at
uniform intervals in time, order, or space, is called stratified
sampling.
6. As a general rule, it is not necessary to include a finite
population multiplier in a comptation for standarp error of
th~ mean when the size of the sample is greater than 50.
7. The probability distribution of all the possible means of
samples is known as the sampling distribution of.the mean. I
..
~
j
125
I
Chapter 7: Sampling
A Quick Approach to Stqtistics with Questions and Answers
8. The principles of simple random sampling are the theoretical

foundations for statistical inference.
9. l"he .standard error of the mean is as the standard deviation of
the distribution of sample means.
10. A sampling plan that divides. the population into well defined
groups from which random samples are drawn is known as
cluster sampling.
11. With increasing sample size, the sampling distribution of the
I,
mean approaches. normality, regardless of the distribution of

By: Rafaqat
the population.
12. To perform a complete enumeration, one would need to
examine every item in a population.
13. In everyday life, we see many examples of infinite populations
of physical objects.
14. Large samples are always a good idea because they decrease
the standard error.
15. The main problem of data collecti.on through mail
questionnaire is bias.
A Quick Approach to Statistics with Questions (llld Answers
-
'S
>f Chapter 8
:d
iS Statistical Inference
1e
of
Statistical Inference
By: Rafaqat
Statistical Inference makes use of information from a sample to draw

to conclusions (inferences) about the population from which the sample was
taken. It has two branches; .Estimation and Testing of Hypothesis.
ns Estimator
An estimator is any quantity calculated from the sample data, which is used
to give information about an unknown quantity in the population. In other
se words, any statistic that is used to estimate a population parameter is called
estimator. For example, the sample mean is an estimator of the population
mean.
ail
Estimate
An estimate is a specific value or range of values used for indication of the
value of an unknown quantity based on observed data. More formally, an
estimate is the particular value of an estimator that is obtained from a
particular sample of data and used to indicate the value of a parameter.
Estimation
Estimation i; the process by which sample data are used to indicate the
value of an unknown quantity in a population.
Results of estimation can be expressed as a single value, known as a Point
Estimate, or a range of values, known as an Interval Estimate ..
26 Chapter 8: Statistical Inference 127
Properties of a Good Point Estimator

Unbiasedness:
An~stimator B is said to be unbiased estimator of parameter () if the mean
of sampling distribution of the values of B is equal to ()i.e.
E(B)=().
Efficiency:
EffiCiency refers to the size of standard error of the statistic. If we COl1'\l)are
two statistics from a sample of the same size then the estimator with smaller
standard error is said to be more efficient.
. Particularly, an estimator fJ 1 is said to be efficient estimator than iJ 1 if
Var ( fJ 1 ) < Var( fJ 2 ). The variane is calculated if the estimators are
..
By: Rafaqat
unbiased. If estimators are biased then mean squared errors will be

compared.
Consistency:
An estimator is.said to be consistent estimator of a population parameter if
as the sample size increases, it becomes almost certain that the value of the
statistic comes very close to the value of population parameter.
.
In other sense an estimator fJ (as n
. is the sample size) is a consistent
. estimator for parameter () if and only if, forall & > 0,110 matter how small,
we have;
P( I Bn - B I< & ) = I ' when n --+ oo.
Sufficiency: .
An estimator is called sufficient estimator if it makes so much use of the
sample information that no other estimator could extract form the sample
additional information about the population parameter being estimated.
Hypothesis
It is a supposition or assumption, which acts as a foundation or as a starting
point in an investigation, irrespective of its probable truth or falsity. For
example, average body temperature of adults is 98.6F, procedute A of
cultivation is better than that of B, etc. .
C!rapter 8: Statistical Inference . 128
l
,
Statistical Hypothesis .
A statistical hypothesis is a statement about parameter(s) of population(s).
Fo~ example, average body temperature. of adults is 98.6F, more than 10%
voters are in favour of a particular party, etc.
Testing of Hypothesis
Hypothesis testing begins with an assumption, called a hypothesis that we
make about a population parameter. Then we collect sample data, produce
sample statistic, and use this as information to decide how likely it is that .
our hypothesized population parameter is correct. The purpose of this type .
of inference to determine whether enough statistical evidence exists to
enable us to conclude that a belief or. hypothesis about a parameter is
supported by the data.
By: Rafaqat
Null Hypothesis
A hypothesis to be tested for possible rejection under the assumption that it
is true, is called mill hypothesis and is denoted by H0 For example, in a
clinical trial of a new drug, the null hypothesis might be that the _new drug is
no better, on average, than the current drug. We would write
H 0 : there is no difference between the two drugs on a~erage;
We give special consideration to the null hypothesis. This is due to the fact
that the null hypothesis relates to the statement being tested.
Altern~tive Hypothesis
The alternative hypothesis, denoted by H., is to be considered as an
alternate to the null hypothesis. It is also known as Research Hypothesis.
For the above example, we :would write
H 1: the two drugs have different effects, on average.
l'h;e alternative hypothesis might also be that the new drug is better, on .
average, than the current drug. In this case we would write
H 1: the new drug is better than the current drug, on average.
Simple ~ypothesis
A simple hypothesis is a hypothes;:,, which specifies the population
distribution completely.
For example, ,
1. Ho: p = 0.5, i.e., p is specified
2. Ho: X - N(S, 20), i.e., and u 2 are specified
Lnapter 8: Statistical Inference 129
I
Composite Hypothesis
A composite hypothesis is a hypothesis, which does not specify the
population distribution completely.
For example,
I. H 1: p > 0.5, i.e., p is not completely specified
2. H1: X - N(5, u 2 ), te., c;2 is not completely specified
Type-I Error
In a hypothesis test, a type-1 error occurs when the null hypothesis is
rejected when !t is in fact true; that is, Ho is wrongly rejected. The
probability of committing type-I error is denoted by a.
A type-I error is often considered to be more serious, and therefore more
important to avoid, than a type II error. The hypothesis test procedure is
therefore adjusted so that there is a guaranteed 'low' probability of rejecting
By: Rafaqat
the null hypothesis wrongly; this probability is never zero.
Type.JI Error
In a hypothesis test, .a type-II ~rror occurs when the null hypothesis H0 , is
not rejected when it is in fact false, Ho is wrongly accepted. The probability
of committing type-II error is denoted by f3.
A type-II error would occur if it was concluded that the two drugs produced
the same effect, i.e. there is no difference between the , two drugs on
average, when in fact they produced different ones. A type-II error is
frequently due to sample sizes b~ing too small.
Significance Level
The significance level of a statistical hypothesis test is a fixed probability. of
wrongly rejecting the null hypothesis Ho, if it is in fact true. It is the
probability of a type I error and is set by the investigator in relation to the
consequences of such an error. That is, we want to make the significance
level as small as possible in order to protect the null hypothesis and to
prevent, as far as possible, the investigator from inadvertently making false
claims. The significance level is usually denoted by a.
Test Statistic
A test statistic is a quantity calculated from the sample of data. Its value is
used to decide whether or not the null hypothesis should be rejected in our
(;_hapter 8: Statistical Inference 130
A Quick Approac/1 to Statistics with Questions and Answers
hypothesis test. The choice of a te~ statistic will depend on theassumed

probability model and the hypotheses under question.
Critical Value
The critical value for a hypothesis te~t is a threshold to which the value of
the test statistic in a sample is compared to detennine whether or not the
null hypothesis is rejected.
The crltiCal value for any hypothesis test depends on the significance level
at which the test is carried out, and whether the test is one-sided or two-
sided (described below).
Critical Region
The critical region (CR), or rejection region (RR), is a set of values of the
By: Rafaqat
test statistic for which the null hypothesis is rejected in a hypothesis test.
That is, the sample space for the test statistic is partitioned into two regions;
one region (the critical region) will lead us to reject the null hypothesis H0 ,
the other will not. So, if the observed value of the test statistic is a member
of the critical region, we conclude "Reject H0 "; if it is not a member of the
critical region then we conclude, "Do not reject H0".
One-Sided (One-Tailed) Test

The tails in a distribution are the extreme regions bounded by critical
values. A test is said to be one-sided or one-tailed test if its entire critical
. region lies on just one (right or left) tail of the distribution under H 1
In other words, the critical region for a one-sided test is the set of values
less than the critical value of the test, or the set of values greater than the
critical value of the test.
Example:
If = average body temperature of adults;
Ho: =98.6,
Against;
H 1: < 98.6 or H 1: > 98.6.
Two-Sided Test
A test is said to be two-sided or two-tailed test if its entire critical region
lies on both tails of the distribution under H 1 In other words, the critical
region for a two-sided test is the set of values less than a first critical value
of the test and the set of values greater than. a second critical value of the
test.
A Quick Approach to Statistics with. Questions and Answers
Example:
Suppose, we want to test a manufacturers claim that there are, on average,
50 sticks in a match-box. We could set up the following hypothesis
Ho:= 50,
. Against;
H1: * 50
The choice between a one-sided and a two-sided test is determined by the
purpose of the investigation or prior reasons for ~sing a one-sided test.
P-Value .
The probability value (p-value) of a statistical hypothesis test is the
probability of getting a value of the sample test statistic that is at least as
extreme as the one found from the sample data assuming that the null.
_hypothesis is true.
By: Rafaqat
It is the probability of wrongly rejecting the null hypothesis if it is in fact

true. It is, actually', observed significance level of the tes.t for which. we
would only just reject the null hypothesis. The p-value is compared with the
presumed significance levei of our test. and, if it is smaller, the result is
significant. That is, if the null hypothesis were to be rejected at the 5%
significance level, this would be reported as "p < 0.05".
Small p-values suggest that the null hypothesis is unlikely to be true. The
smaller it is the more convincing is the rejection of the null hypothesis.
Power of.the Test

The power of a statistical hypothesis test measures the test's ability to reject
the null hypothesis when it is actually false, that is, to make a correct
decision.
In other words, the power ofa hypothesis test is the probability of not
committing a type- II error. It is calculated by subtracting the probability of
a type- n error from I, usually expressed as:
Power = I - P (type II error) = I - f3
The maximum power, a test can have is land the minimum is O. Ideally, we
want a test to have high power, close to l. '

Questions-Answers
Q.l What are the major parts of the statistical inference?

Ans. Estimation of parameters and testing of hypothesis are the two
major parts ofthe statistical iriference.
Q.1 What Is inductive inference?

Ans. The research worker performs an experiment and obtains some
data. The conclusions usually go beyond the materials and
operations of the particular experiment. This sort of the extension
By: Rafaqat
from the particular to the general is called inductive inference.
Q.3 What is deductive inference?

Ans. The conclusions based on "general to particular" in known as
deductive iriference. In other words when an argument claims that
the truth-of its premises guarantees the truth of its conclusion, it is
said to involve a deductive inference.
Q.4 Differentiate positively and negatively biased estimator?

Ans. When the expected value of the estimator is greater than the
corresponding parameter value, the estimator is positively biased.
If the expected value of the estimator is smaller than the
corresponding parameter value, the estimator is negatively biased.
Q.5 What are the methods used/or parameter estimation?

Ans. Method of least square, method of maximum likelihood, method of
moments etc. are usedfor param?ter estimation.
Q.6 Define likelihood/unction.

~s. A likelihoodfanction L(B) is the probability or probability density
for the occurrence of a sample configuration Xi. Xi. :.. , Xn given
that the probability density function f(x; BJ with parameter 0 is
known. Symbolically it can be written as;
L(O) = /(xt; O).f(x 2 ; 0) ... f(x.; 0).
Chapter 8: Statistical Inference 133 '
Q. 7 What is the principle of Maximum Likelihood (ML)?

Ans. The M/,, principle directs us to take as our estimator of () that
value within the admissible range of() which make the likelihood
function as large as possible. i.e. we choose B so that for any
admissible value of();
L(x/ B)~L(xl B)
Q.8 What is the method of least squares?

Ans. Method of least squares 'is a mathematical optimization technique
that attempts to find a 'best fit' to a set of data by attempting to
minimize the sum of the squares of the residuals between the fitted
function and the c/ata.
By: Rafaqat
Q.9 What is the method of moments?

Ans. The method of moments is a method ofestimation ofparameters by
equating sample moments with unobservable population moments
and then solving those equations for the quantities to be estimated
Q.10 What is confidence coefficient? How it can be interpreted?

Ans. The degree of coef,dence is the probability (I - o.) that the
population parameter is contained in the confidence interval. This
probability is often expressed as the equivalent percentage value.
The degree of confidence is also referred to as the level of
confidence or the confidence coefficient. .
If we suppose o. = 0.05 then 100(1 - o.)% = 95% confidence
coefficient means that we are 95% confident that the true value of
population parameter will lie between the computed 95% interval
estimate.
Q.11 Can every hypothesis be treated as statistical hypothesis? If not,

then specify the difference between statistical and non-statistical
hypothesis?
Ans. No, every hypothesis cannot be treated as statistical hypothesis.
The kind of hypotheses, which we test in 'Statistics, is more
restricted than the general scientific hypotheses. It is a scientific
hypothesis that every particle of matter, in the universe, attracts
every other particle, or the life exists on Mf}rs; but these are not
'hypotheses such as arise for testing from the statistical view point.
Statistical hypothesis concerns the behavior of observable random
variable. Jn other words statistical hypotheses are testable claim.s
or assertions about one or more parameter of empirical
distributions.
Q.12 At what point in the research procedure should a hypothesis be

stated?
Ans. Ideally, the hypothesis should be stated before the sample data
are collected and analyzed
Q.13 How does an analyst know which significance level is to be used

in a test?
Ans. He must determine what risk he is willing to take in making a type-
By: Rafaqat
/ error. Jn doing this, consideration is given to the consequences of

making such error. If thf!/ decision is to be made about much
critical situation the level of significance will be small, conversely
if the decision to be made about less critical.situation the level of
significance will be large.
Q.14 Which type of testing error is more serious?

Ans. A type-I error is often considered to be more serious, and therefore
more important to avoid, than a type-II error. The hypQthesis test
procedure is therefore adjusted so that there is a guaranteed 'low' .
probability of rejecting the null hypothesis wrongly.
Q. I 5 Define degree offreedom.

Ans. The number of independent groups or sub-categories into which
. sample or population may be divided is called degree offreedom.
For example, in a sample of constant size n, grouped into k
intervals there are 'k-1' degrees of freedom .because if 'k-1'
frequencies are specified the other is determined by the total size.
Q.16 Can 'the probability of making a type II error is determined in a

life sampling problem?
Ans. No, because one does not know how false the hypothesis is.
Q.17 How we minimize both type I error and type II error

simultaneously?
Ans. If we increase the sample size both errors will be minimized.
Q.18 Criticize the following statements.

(a) It is impossible to control both type-I and type-II error.
Since decreasing one will increase the other.
(b) A one sided alternative mean to test H 0 /or only type-II
error.
(c) If a person has a choice of applying lwo different tests,
to decide between ti certain set of null and alternative
hypotheses, assuming both have the same significance
level Then the person should use the test which has
snialler probability-of committing type- II error
By: Rafaqat
(d) It is important to select the .significance level before

choosing a. decision rule.
Ans. (a) This would be true only for a frxed sample size. By
increasing the sample size we can decrease the
probabilities ofboth types of error.
(b) Disagree, a one sided alternative simply specifics the
location of critical region an either side. It does not te,st
null hypothesis for only type-JI errot, one may commit a
type- I error.
(c) Agree.
(d) Agree.
Q.19 How we establish the conclusion of the hypothesis test?

Ans. The final conclusion once the test has been cqrried out is always
given in terms of the null hypothesis. We either "Reject Ho In
failour of H/' or "Do not reject H0 ". We never conclude "Reject
H/', or even "Accept H/''.
Ifwe conclude "Do not reject H0 ", this does not necessarily mean
that the null hypothesis is true; it only suggests that there is not
sufficient evidence against H0 in favor of H 1. Rejecting the null
hypothesis then, suggests that the alternative hypothesis may be
true.
A Qick Approach to Statistics with Qu{!stions and Answers
Q.10 Discuss the importance of normal distribution in testing of

hypothesis.
Ans. One reason the normal distribution is importaYJt is that many
variables (e.g. psychplogical, educational etc.) are distributed
approximately normal. Measures of reading ability, introversion,
job satisfaction, and memory are among the many variables
approxima_tely normally distributed Although the distributions are
only approximately normal, they are usually quite close. A second
reason the normal distribution is so important is that it is easy for
mathematical statisticians to work with. This means that many
kinds of statistical tests can be derived for normal distributions.
Almost all statistical tests assume normal distribution.
Fortunately, these tests work very well even if the distribution is
vnly approximately normally distrib~ted Some tests work well
By: Rafaqat
even with very wide deviations from normality. Finally, if the mean
and standard deviation of a normal distribution are known, it is
easy to convert back and forth from raw scores to percentile&
,.
Q.21 Discuss the role of Chi-square distribution in testing of
hypothesis and confidence interval estimation.
Ans. Let Z -N(O,J) be a standard normal variable. Jf n random values
Z1., Z2, .. , Zn are drawn from this distribution, squared, and
summed, the resultant statistic is said to have a I distribution with
n degrees offreedom. It is right skewed distribution rangingfrom 0
to oo. Chi-square distribution fzas a wide range of application. In
.tesiing of hypothesis it is being used to test the variance, equality
of variances. oreover it is used to test the goodness offit and
also for testing association of attributes. With reference to
confidence interval it is being used to. construct the confidence
interval for population variance.
Q.22 What are the main uses of I-distribution In testing of hypoJ/fesis

and construction of confidence intervals?
Ans. The I-distribution may be defined in terms of normal and
independent I variables.
Let Z - N(O, I) and Y- / (n) where Z and .Y are independently
distributed then;
..~
A Quick Approach to Statistics with Questions .and Answers

It is symmetrical about its mean point, zero, and tends
asymptotically to the standard normai distribution. The standard
normal distribution is replaced by t-distribution when population
variance is estimated from sample data and central limit theorem
is not applicable. I-distribution may be used for testing the means,
equality of.means (two), proportion and difference of proportions
(two). All this. will be done when the population variance is
unknown.
The distribution is also applicable to construct the confidence
interval for:
By: Rafaqat
i) Population mean and population proportion

ii) Difference between two population means and
population proportions either samples are independent or
related
. Q.23 What type of hypotheses can be tested by using F-distribution?
Ans. The F-distribution is defined in terms of two independent l
variables. Let U and V are independently. distributed l variable
with n 1 and n2 degrees offreedom, resp.ectively. Then the statistic;
F=Uln 1 ;
VI nz
is distributed as F with n1 and n2 degrees offreedom.
It is a continuous probability distribution, which has two
parameters, n 1 and n2 which are positive whole numbers. Like the
Chi-square distribution it is right skewed rar,iging from 0 to oo. In
testing of hypothesis it is being used to test the equality of two
population variances, equality of more than two means (i.e. in
ANO VA) and also to test the significance ofoverall regression~
Q.24 Define power and operation characteristic (OC)functions.

Ans. The function, which gives the probabilities of rejecting the .null
hypothesis for different values of alternative hypothesis, then, the
function will be known as power function. Similarly the function,
which gives probabilities of accepting the null hypothesis for
different values of alternative hypothesis, is known as OC function.
. The complement of the power function is the operating

characteristic function.
Q.25 What are power and OC curves?

Ans. If the probabilities of rejecting the null hypothesis plotted for
various alter'!ative hypotheses then the obtained curve is power
curve. Similarly if the probabilities of accepting null hypothesis
are plotted for various alternative hypotheses then the operating
characteristic curve is obtained
Q.26 Describe the general procedure Jot testing of hypothesis.

Ans. Following are the desirable steps comprising the procedure for
testing of hypothesis:
i) Formulation ofnull and alternative hypothesis
By: Rafaqat
ii) Setting oflevel ofsignificance

iii) $pecification oftest statistic
iv) Determination ofcritica/,region
v) Computation
vi) Results and conclusions.
Q.27 Define Best CriJical Region (BCR).

Ans. If. among critical regions offzxed size there is one which minimizes
the probability of type-I~ error, then it is known as the BCR.
Q.18 Define and explain sequential analysis.

Ans. In mcmy circumstances, the observaiions to. be used in evaluating'
models an:ive sequentially rather than all at once. Sequentiti{
analysis is the procedure in which sample size is not fzxed in
advance. The decision is to be made after taking an observation If
the observation have a potency to make a decision about the truth
or /<;1/sity of the hypothesis, than we stop, taking observations. If
not so, then we take another observation and test for acceptance. or
rejection of hypothesis and so on until we reach at the result. In
this type of testing we do not specify the sample size but we fzx
probability oftype-I error (a) and probability oftype-II error (/J).
Chapter 8: Statistical Inference 139.
Exercises
Read the following statements carefully and indicate which statem.~nt is
"True" or "False":
1. One estimator is more efficient than another if a large sample is

By: Rafaqat
used with the first.

2. A statistic will always be an unbiased estimator if the sample
itself is chosen without bias.
3. A consistent estimator is one that tends to be more accurate with
large samples than with small ones.
4. All efficient estimators are automa\ically unbiased.
s. The sample variarlce s2 is calculated somewhat differently than
.), the population variance d in order to make s2 an unbiased

estimator o(d.
6. Confidence may always be increased by reducing the precision
of the interval estimate.
7. The 95% confid_ence interval for the mean gasoline mileage for a
certain make of a car is 15 ~ ~ 17. This means that there is a
probability of 0.95 that actually falls between l 5 and 17.
8. A 99% confidence interval will be wider than a 95% confidence
interval constructed from the same data.
Chapter 8: Statisticai Inference 140 CJ
9. The student's !-distribution applies when X is used to estimate

and when u is an unknown value.
1O. The more disperse of two populations tends. to have a narrower
confidence interval at the 95%.
11. Null hypothesis (Ho) always indicates no change.
12. Hypothesis testing models assume either that Ho must be true or
that it qiust be false.
13. The type-II error is equivalent to rejecting the alternative
hypothesis when it is .true.
By: Rafaqat
14. The power curve relates the probability ofrejecting H0 to various

possible population parameter values.
15. For an upper-tailed test, a reduction of the significance level
from 0.05 t6 O.ol implies that there must be a corresponding
increase in the critical value.
16. An upper-tailed test occurs when H 0 : _?:. 0 and H, : < 0
17. Whenever the Type-I error .does not occur, the Type-II error
must occur.
18. H 0 is that ::;; I 0 . )'he critical value is X= 12, and the observed
sample mean is X= 15. H0 should therefore be accepted.

19. If a is the desired probability for incorrectly rejecting H0 when
in fact 2: 110, !hen the null hypothesis is H0 : 2: 110.
20. Hypothesis testing is a tool that statisticians use so that they
never make incorrect decisions.
21. With a two sided hypothesis testing procedure, there are two
ways in which the type~I error may occur.
22. In an upper tailed test, large values of the test statistic result in
rejecting the null hypothesis.
23. - For a prescribed decision rule, a larger sample size will result in
a larger. type-I error probability.
24. A contingency table can have only two rows and two columns.
25. The Chi-square test can be used to determine if there is a
significant difference between two sample percentages.
26. In a .Chi-square analysis tile- number of rows in the contingency
table should equal the number of columns.
By: Rafaqat
27. A Chi-square analysis should not ordinarily be .made when there

ate fewer than five observations in one or more of the cells.
28. There are two main type of Chi-square tests i.e. goodness of fit
and badness of fit.
29. A Chi-square . test is actuaUy a test to determine if tlie
distribution of sample data conforms to some expected or actual
distr-ibution.
30. In a <::hi-square test, it is assumed that the sample observations
were selected from a normally distributed universe.
Chapter 8: Statistical Infetence 142
1111111 Chapter 9
Design and
Analysis of Experiments
Experiment
An experiment is any process or study, which results in the collection of
data, the outcome of which is unknown. In statistics, .the term is usually
.restricted to situations in which the researcher has control over some of the
By: Rafaqat
conditions under which the experiment takes place ..

;
Design of Experiment (DOE)

t We are concemed with the analysis of data generated from an experiment. It
is wise to take time and effort to organize the experiment properly to ensure
that the right type of data, and enough of it, is available to answer the
e . questions of interest as clearly and effiCiently as possible. This process is
called DOE. In other words a DOE is a structured, organized methbd for
LI
determining the relationship between factors (X's) affecting a process and
the output of that process (Y). DOE refers to experimental methods used to
quantify indeterminate measurements .of factors and interactions between
lS
facfors statistically through observance of forced changes made
methodically as directed by mathematically systematic tables. J
.~
Experimental Unit
Experimental units is the basic object upon which the study or experiment is l
carried out. The entity to which a specific treatment combination is applied.
An.experimental unit can be a individual agricultural plant, plot of land, PC
board, silicon wafer, tray of components simultaneously treated, automotive
transmisl!ions etc.
Basic Principles of Experimental Design

Randomization, replication and local control are the basic principles of
experimental design. .I
--
12
Chapter 9: Design Rnd Analysis of Experiments 143
j
J
A Quick Approach to Statistics ivith Questions and Answers
Randomization
Randomization is a schedule for allocating treatment material and for
conducting treatment combinati,ons in a DOE such that the conditions in one
run neither depend on the conditions of the previous run nor predicts the
conditions in the subsequent runs. The importance of randomization cannot
b.e over stressed. Randomization is necessary for conclusions drawn from
the experiment to be correct, unambiguous and defensible. Randomization
is preferred since alternatives may lead to biased results. The main point is
that randomization tends .to produce groups for study that are comparable in
unknown as well as known factors likely to influence the outcome, apart
from the actual treatment under study. The analysis of variance F tests
assume that treatments have been applied randomly.
By: Rafaqat
Replication
The repetition of an experiment on a large group of subjects is known as
replication. If a treatment is truly effective, the long-term averaging effect
of replication will reflect its experimental worth. If it is not effective, then
the few members of the experimental population who may have reacted to
the treatment will be negated by the large numbers. of subjects who were
unaffected by it. Replication reduces variability in experimental results,
increasing their significance and the confidence' level with which a
researcher can draw conclusions about an experimental factor.
Local Control
'J. Local control refers to grouping of the experimental units in such a way that
,,..,.! the units within a group (i.e., block) are more homogeneous than are units
in different groups. The experimental materials or conditions are more alike
within a group. Thus, the variation among experimental units within a group
is less than the variation would have been without grouping. This leads to
the "comparison of treatment effects under more uniform conditions or .on
the more uniform materials. For example, the total variation in Randomized
Complete Block Design (RCBD) is partjtioned into _:variation due to two
assignable causes, blocks and treatments, and .variation due to a non-
assignable cause or experimental error. This latter source of variation is
, reduced .as the variation due to block is remove&.
Experimental error= Total variation - Treatment variation - Block variation.
Chapter
, 9: Design and Analysis of.Experiments 144
A Quick Approach to Statistics with Questions.and An.swen
Treatment in Experimental Design

In expe~iments, a treatment .is something that researchers manage to
experimental units. For example, a com field is divided into four, each part
is 'treated' with a different fertilizer to see which produces the most corn; a
teacher practices different teaching methods on different groups _in his/her
class to see which yields the best results; a doctor treats a patient with a skin
condition with different creams to see which is most effective.
Treatments are administered to experimental units by 'level', where level
implies amoupt or magnitude. For example, if the experimental units were
given 5mg, IOmg, l5mg of a medication, those amounts would be three
levels of the treatment. 'Level' is also used for categorical variables, such as
Drugs A. B, and C, where the thre'e are different kinds of drug, not different
amounts of the same thing. In other words a treatment is a specific
combination of factor levels whose effect is to be compared with other
By: Rafaqat
treatments.
Factor of an Experiment
A factor of an experiment is a controlled independent variable; a variable
whose levels are set by t~e experimenter. A factor is a general type or
-category of treatments. Different treatments constitute different levels of a
factor. For example, three different groups of runners are subjected to
different training methods .. The runners are the experimental units, the
training methods, the treatments, where the three types of training methods
constitute three levels of the factor 'type of training'.
Analysis of Variance (ANOVA)

A mathematical process for separating the variability of a group of
observations into assignable causes and setting up various significance tests.
In other words analysis of variance is a statistical technique for analyzing
data that tests for a difference between two or more means by comparing
the variances "within" groups and variances "between" groups.
Assumptions:
Normality: The populations have normal distributions.

Homogeneity: The populations have the same variances. .
Randomization: The samples are random and independent of each other.
Chapter 9: Design and Analysis of Exferiments 145
Questions"".Answers
Q.J Define Model?

Ans. A Mathematical relationship, which relates changes in a given
response to changes in one or more factors.
Q.2 Define the term Design.

Ans. A set of experimental runs, which allows us to fit a ptirticular
model and e#imate our desired effects.
By: Rafaqat
Q.3 What is ~ant by the term Treatment Combination?

Ans. The combination of the settings. of several factors in a given
experimental trial. Also known as a run.
Q.4 Define variance components.

Ans; Partitioning .ofthe overall variation into. assignable components.
.
Q.5 What is one-way ANO VA?

Ans. The .one-way analysis of variance (ANO VA) allows us to compare
several groups of observations, all of which a.re independent but
possibly with a different mean for each group. A test of great
'J. importance fs whether or not all the means are equal.
1 The observations all arise from one ofseveral different groups (or
have been exposed to one of several. different treatments in an
experiment). We are. clas~ifying 'one-way' according to the group
or treatment.
Q.6 What is two-way ANO VA?

Ans. Two-way Analysis of Variance is a way of studying the effects of
two factors sepprately Jtheir main effects) and (sometimes)
together (their interaction effect). .
Q.7 What are the basic experimental designs?
Ans. Basic experimental designs are Completely Randomized Design
(CRD), Randomized .Complete 'Block Design (RCBD) and Latin
Square Design (LSD).
~~~----~~----~--~~~~~~~~~~~~~--~
Chapter 9: Design and Analysis of Expuiments 146
Q.8 EXplore the idea of completely randomized design.

A. The structure of the experiment in a completely randomized design
is assumed to be such that the treatments are allocated to the
experimental units completely at random.
Q.9 Explore the idea of randomized complete block design?

Ans. The randomized complete block design is a design in which the
subjecis are matched according to a variable, which the
experimenter wishes to control. The subjects are put into groups
(blocks) of the same size as the number of treatments. The
members of each block are then randomly assigned to different
treatment groups. '
Q.10 Explore the idta of Latin square design.

By: Rafaqat
Ans. Latin square (and related) designs are efficient designs to b/Qck
from 2 to 4 nuisance factors .
.Q.12 What is blocking?

Ans. This is the procedure.by which experimental units are grouped into
homogeneous clusters in an attempt to improve the comparison of
treatments by randomly allocating the treatmenis within each
cluster or 'block'.
Q.13 Explore the idea offactorial design.

Ans. A factorial design is used to evaluate two or more factors
simultaneously. The treatmenls are combinations of levels of the
factors. The advantages offactorial designs qver one-factor-at-a-
time experiments is that they are more efficient and they allow
interactions. to be detected.
Q.14 What is the effect?

Ans. Effect gives us how changing fhe settings of a factor changes the
response.
Q.15 What is main effect? .

Ans. The main effect is the average simple effect of a factor on a
dependent variable. It is the effect of the factor alone averaged
across the levels of other factors. In other words a main effect is a
Chapter 9: Design and Analysis of Experiments . 147
measurement of the average change in the output when a factor is

changedfrom its low level to its high level.
Q.16 What is interaction effect?

Ans. An interaction is the variation among the differences between
means for different levels of one factor over different levels of the
other facto~.
Q;17 What isfu:ed effect?

Ans. An effeet associated with an input variable that has a limited
number of levels or in which only a limited number of levels are of
interest to the experimenter.
Q.18 What is random effect?

By: Rafaqat
Ans. The random effect model does not provide knowledge of the
treatment effect at a particular level. It enables us to study the
variability due to the effect of treatment. Therefore the random
effect model is sometimes ca/led the component of variance. Or an
effect associated with input variables chosen at random from a
population having a large or infinite number ofpossible values.
Q.19 What is interaction effect?

Ans. It occurs when the effect of one factor on a response depends on
the level of another factor(s).
Q.20 What is the lack offit error?

Ans. Error that occurs when the analysis omits one or more important
terms or factors from the process model. Including replication in a
DOE allows separation of experimental error into its components:
lack Offit and random (j:Jure) error.
Q.21 What is meant by the term confounding?

Ans. A confounding design is one where some treatment effects (main or
interactions) are estimated by the same linear combination of the
experimental observations as some blocking effects. In this case,
the treatment effect and the blocking effect are' said to be
confounded Confounding is also used as a general term to
indicate .that the value of a main effect estimate comes from both
the main effect itself and also contamination or bias from higher
Chapter 9: Design and Analysis of Experiments 148
A Quick Approach to Statistics with Questions and.Answers
order interactions. Confounding naturally arise when full factorial

designs have to be run in blocks and the block size is smaller than
the number of different treatment combinations. It also occurs
whenever a fractional factorial design is chosen instead of a full
factorial design.
Q.22 What are Taguchi methods?

Ans. A technique for designing and performing experiments to
investigate processes where the output depends on many factors
(variables; inputs) without having to tediously and uneconomically
l
run the process using al/possible combinations of values of those
variables. By. systematically choosing certain combinations of il
variables it is possible to separate their individual effects.
By: Rafaqat
Q.23 What is the.balanced design?.

Ans. An experimental design where all cells (i.e. treatment
combinations) have the same number of observations.
Q.24 What is meant by the comparative design?

Ans. A design aimed at making condsions about one a przorz
impo.rtant factor, possibly in the presence of one or more other
"nuisance'' factors.
Q.25 What is meant by the term ortllogonality?

Ans. Two vectors of the same length are orthogonal if the sum of the
products of their corresponding elements is 0. An experimental
design is orthogonal if the effects of any factor balance out (sum to
zero) across the effects ofthe other factors.
Q.26 Define responses.

Ans. The output(s) of a process is known as response. Sometimes called
dependent variable(s).
Q.27 What are response surface designs?

Ans. A design of experiment thatfully explores the process window and
models the responses. These designs are most effective when there
are -less than 5 factors. Quadratic models are used for response
surface designs and at least three levels of every factor are needed
in the design.
Cllapter 9: Design and Analysis of Experiments 149
Q.18 Define scaling factor levels.

Ans. Transforming factor levels so that the high value becomes + 1 and
the low value becomes -1.
Q.19 What are screening designs?

Ans. A design of experiment that identifies which of many factors has a
significant effect on the response. Typicqlly screening designs have
. more than 5 factors.
Q.30 What are t/1e tests used/or mean comparison?

Ans. When treatment effect is declared significant, different tests are
used for comparing the treatment means. These tests include Least
Significance Difference (LSD) tes_t, Duncan Multiple Range Test
By: Rafaqat
(DMRTJ., Tuckey 's test, Scheffe 's test, orthogonal contrasts, trend
comparisons, etc.
Chapter 9: pesign and Analysis of Experiments 150
Exercises
Read the following statements carefQ!iy-and indicate whi~h statement is
"True" or "False":
1. The F-distribution is symmetrical,

2. The F-distribution isa distributioi:i of ratios of sample variances.
By: Rafaqat
3. An analysis of variance (ANOVA) is a useful tool for provii:ig or

disproving a null hypothesis of several means.
4. In an ANOV A an F-ratio of 1.00 -means that a significant
difference exists at both the 0.05 and 0.01 levels.
5. The F-ratio may be computed to detennine if there is a
significant difference among means.
6. In an ANOV A, the number of items in each column must be the'
same.
7. A one-way ANOVA is restricted to a maximum of seven
columns of data.
8. In a two.,way ANOV A the number of rows need not equal the
number of columns.
9. "f!le number of degrees of freedom _lost when computingthe
variance within columns is equal to the total number of
observations in all columns minus the number of columns.
10. In a two-way ANOV A the number of degrees offreedom for the

residual variance is equal to (r - 1) (c - 1). Where r denotes
rows and c denotes columns.
11. In a one-way ANOV A, it is assumed that the residual variation
is due to chance.
12. In a two-way ANOV A, it is assumed that the residual variation
is due to chance.
13. In an A.NOV A, it is assumed that the universes_ from which the
samples were selected are normally distributed.
By: Rafaqat
14. The number of degrees of freedom to be used in computing the

variance between the columns is equal to the number of columns
minus one (c - 1).
15. In a two-way ANOV A, there are three possible sources of
variation.
l!
'j
111111 Chapter 10
Analysis of Time Series
Time Series
By: Rafaqat
Any variable that is measured overtime in sequential order is called a time

series or data recorded at regular intervals of time is called time series for
example, daily city temperature, daily exchange rate of Pak. Rupee in US $,
monthly sale of a departmental store, annual yield of wheat of a country
etc.
Analysis of Time Series

The statistical methods used to make an inference about the pattern of the
time series data is known as the time series analysis. This analysis is used to
detect patterns of change in statistical information over regular intervals of
time.
Time Series Compopents

There are four components' of time series:
/. Secular Trend
2; Seasonal Variations
3. Cyclical Fluctuations
4. Irregular Variations
Secular Trend
The ba:;ic long-term movement in a time series that can be described by a
smooth line.
Chapter 10: Analysis of Time Serles 153
A Quick Approach.to Statistics with Questions and Answers
The trend is affected by changes in population, technology and productivity.

Its duration is more than one year and in many cases it is often more than IO
years. The steady increase in the cost of living recorded by the consumer
price index is an example of secular trend.
The methods of analyzing the secular trend are:
The method of freehand curve
The method of semi-averages
The method of moving averages
The method of least squares
Seasonal Variations
The seasonal variations are short-term movements occurring in a periodic
manner. These variations are mainly caused by the changes in the season.
These variations involves pattern of change within a. year that tend to be
By: Rafaqat
repeated year to year. For examplei.increase in number of flu cases every

winter, increase in use of electricity demand during summer, demand of.
liquid gas during winter, increase in sale of shoes near Eid, etc.
The methods of analyzing the seasonal ~ariations by seasonal indices are:
The Percentage-of-Annual Average Method

The Raito-to-Moving Average Method
The Ratio-to-Trend Method
The Link Relative Method
Cyclical Fluctuations
Cyclical fluctuation is a wavelike pattern describing a long-term trend that
"''t is generally apparent oyei' a number of years, resulting .in a cyclical effect.
A cycle is said to be completed when beginning with. a peak, the declining
curve reaches a low point, and then rising again reaching the next peak. By
definition, it.has duration of more than one year. Business cycle is the most
common example of such variation. .
Irregular Variations
Non-periodic or random fluctuations those are due to non~recurring or non-
periodic events such as strikes, wars, elections, deaths, and weather changes
etc.
Chapter 10: Analysis of Time Serles 154
-.,------
Questions-Answers
Q.J How can yo differentiate the cross-sectional and time series

data sets?
Ans. Time series data are data colleCled over a periods of time. Such
data are collected over differen~ intervals of time, such as daily,
weekly; monthly, quarterly, annually etc. , While cross-sectional
data are data on one or more variables collected at the same point
in time s.uch as individuals, consumers, firms, industry etc.
By: Rafaqat
Q.2 What is the purpose of the time series analysis?

Ans. The bas_ic purpose of the time series analysis is to use it for the
forecasting. Therefor.e, it is a decision making tool. Time series
analysis is us.ed for many applications such as: pconomic
Forecasting, SalesForecaSting, Budgetary Analysis, Stock Market
Analysis, Yield Projections, Process and Quality, Cqntrol,
lrtventory Studies, Workload Projections, Utility Studies. Census
Analysis, etc.
Q.3 Why the term "Series" is applied in the time series analysis?
Ans. We use the term time series to refer to any group of statistical
information accumulated.at regular intervals.
Q.4 On what we frequently, base our predictions and decisions?

Ans. Past patterns ofgrowth and changes are frequently used as a base
for predications and decisions. The businessman may, for example,
find that his sales hav'e risen about I 0% fr.om April. to May every
year for the past I 0-years, and use this to predict that his sales in
the next year will increase from April to May too.
-
J
A .QuickApproach to Statistics with Questions and Answers
Q.5 How can technological change affect the trend of a time series?
Ans. Technological change can cause up~ard or downward mov,ement
in the time series. For e3camp/e, the development of the tractor
caused a downward trend in the number of the mules, ox and
camels on the farms. However, the. development of the tractor
helped produce an upward trend in the sales of the petrol.
Q. 6 What time of the time series have the composition T x C x S x I?

Ans. Data classified by quarters, months and weeks have the
composition T x C x S x /.
Q. 7 What is the mathematical composition of time series'?

Ans. Annual time series: T x C x /. A monthly and Quarterly time
By: Rafaqat
series: T xC xS x I.
Q.8 In which wayypu graphic.ally represent the time series analysis?

Ans. Graphical representation of time series anqlysis is through the
graph oftime series; historigram.
Q.9 Why it important to chart a time series before choosing.the type

of trend to flt to the series?
Ans. Charting the trend enables the. one to make better choice of the
equation or method use to describe it.
Q.10 Give some rules for constructing time series line charts.
Ans. Some good rules are:
Make a chart wider than high
Place time on the horizontal axis and a scale for the vaiues on
the vertical axis, show only a few scale values
Sei tf!e scale so that the line (or Jines) will appear near the
center ofthe chart
Start the scale with zero urlless indexes are being charted
Make equal distance on the vertical scale represent equal
absolute amourtts
Plot po~nts to the middle of the periodfor eumulative data and
to the point ofthe time for non-cumulative data
Connect plotted points with the straight lines
Chapter I 0: Analysis of Time Series 156
Q.11 Why should the numerical scale of a time series line chart being
with the zero? Are there any exceptions?
Ans. The scale should begin with the zero becau3e the vase of the
reference is zero. An exception is the scale for an index, which has
the base of JOO. Another exception is when a logarithmic scale
(ratio scale) is used
Q.12 What does the semi-logarithmic scales do?

Ans. A semi-/Ogarithmic chart shows rates of the change. This chart is
useful for comparing two series that are measured in the same
units but different series that are measured in the same units but
different magnitude, such as rupees and millions of rupees. It is
also useful of comparing two series that are experienced in
different units, such as tons ofsteels and number of autos.
By: Rafaqat
Q.13 Define cu'mulative and non-cumulative data.

Ans. Cumulative data, often called period data, are those that cumulate
from the beginning of a period to the end of a period An example
is annual sales for each year for I 0 years. Non-cumulative data,
often called point data, are those that refer to a specific point of
time. An example would be the population of a province on July I
ofeach year for 10 years.
Q.14 What is the basic reason for measuring the secular trend of a
time series? Why is it important to know the reason?
Ans. The two basic reasons are:
To projectit and use it as a forecasting tool
To eliminate it from the original data so that other movements
can be studied independent from the trend's influence. When
we know, the reasons for measuring the trend of a .time series,
ii is better, able to choose an appropriate method
! .
Q.14 What type of trend is best one to use?

Ans. The type oftrend that is best is the one that ~est fulfills the purpose
ofmeasuring it.
i
Chapter J 0: AIJalysis of Time Series 157
, .
Q.15 In what direction does a secular trends mov'?
Ans. Secular trends can move up, down, or in both of the directions. i.e.
value of the wiriable tends to increase or decrease over long
period of the tin,e.
Q.16 What is the most widely used trend line? What equation is
generally used to describe a trend with one bend?
Ans. . The, least squares line is . the mostly commonly used The
trend equation is;
Y = a+bx+cx . 2
t
is used more than another equation to describe a trend with one
By: Rafaqat
bend .
Q.17 . Explain the meaning of tl~e coefficients of the equation
Y = a + bx + cx 2 .
I
Ans. The coeffident 'a' is the value of the trend as its origin, 'b' is the
general basic slope of the line and 'c' is th~ typical change in the
basic slope per unit lime.
Q.18 What is the purpose of.the measuring cycles of the time series?
Ans. Cycles of the time series usually measured so that they can be
studied and analyzed in order to find a way to forecast turning
~ point of the future cycles.
... Q.19. why are the turning points of the business cycles are different to
<..
predict?
Ans. Turning points of the business cycles are difficult to predict
because turning points of past cycle have not occurred at regular
interval oftime.
Q.20 What are the four phases of a business cycle?

Ans. The four phases are the recovery, prosperity, recession and
depression.

.)
..j
t
A Quick Approach toStatistics with Questions and Answers
Q.11 Why it is difficult to measure, tire typical business cycle.by direct

method? .. . .
.Ans. The lack of regularity in timing and magnitude of cyclic
fluctuations make it di/fic~/t to get a useful direc.i measure, such as
an "average" cycle, which represent a typica/business cycle.
Q.11 How is lag-lead relationship used?

Ans. Lag-lead relationships are sometimes used to predict the turning
points ofthe cycles of one series based on the turning points of one
or more series.
Q.13 How are the cycles isolated forlJI the other components of a time
series? -
Ans. Cycles are isolated so that they must be analyzed and studied
By: Rafaqat
without being biased by the effect ofother fluctuations .
. Q.14 Give an example of an industry that e:i:periences large irregular

fluctuations in the activity. Tell why these.fluctuations occur.
Ans. A good example would be construction because St:Vere hot wiather
(June-July) changes can be abruptly stop/slow down constr1Jction
activity. Another example is the stock market. The reason for these
fluctuations is humorous and many-are unknown. However, wars,
strikes, political speeches, etc. will cause such fluctuations in
prices ofstock.
Q.25 Why would it be difficult to predict the point 1'n time when a
business cycle will reach a peak or trough? .
Ans. The timing of cyclical peaks and troug,,. are difficult to predict
becquse ofthe irregularity in the timing ofthe past business cycles.
'I
r Q.26 Show symbolically how the influence of seasonal variations may
be removed from a monthly time series.
TxCxSx/
Ans.
s
=
TxCxl
A Quick Approach tu Statittics with Questions and Answers
(l-27 If the direct measure of the cyclical and irregular movements of

time series could be calculated, state symbolically .how their
combined influence could be removed from:
I) An annualtime series
2) A monthly time series
3) . A quarterly time series
Ans. Annual time series:= T xC x I IC.x I= T
Monthly time series and Quarterly time series
=TxCxSx//Cxl=TxS
Q.28 Why do seasonal variations show a high degree of regularity in

timing and amplitude? .
Ans. A high degree of regularity in the seasonal variations resulted
from the regularity of Holydays and seasonal bujling habits of
By: Rafaqat
people and regularity of harvest, production and comumption that

are caused by seasonal weather changes.
Q.29 What preliminary adjustment should somefliite be made in data

before measuring the typkal seasonal pattern? why?
Ans. Period data sometime need to be adjusted/or the number calendar
days or working days in a monlh. This would be done when one is
interested in measuring the rate of'seasonal activity rather than the
total seasonal activity.
Q.30 Why might a seasonal pattern changes over a long period of
time?
Ans. Changes in seasonal patterns over .a long period can be due to
. technological improvements. e.g. improved transportation changed
the seasonal consumption. of certain fruits and vegetables in the
cold climates.""
Q.31 What is the chief advantage of usllig the percent of 12-months

moving average method to compute a seasonal index?
Ans. The "percent of 12 month moving average"' method enables the
analyst to remove the effect of trend and cycles so that these
movements will not bias the seasonal index.
Q.32 For what types of business decisions ls a.seasonal index of sales

useful?
Chapter 10: Analysis of Time Series '160
. '
A Quick 'Approach to Statistics with Questions and Answers
Ans. Seasonal indices are useful for such decisions as diversification of

products, personnel practices (for example, seasonal employment
and vocation schedules), and advertising and promotions plans.
. Q.33 Why are most important monthly economic indicators adjusted

for seasonal variation?
Ans. The main purpose of the adjustments is to reveal the real basic
directions of indicators so that the strength of the economy can be
better evaluated.
Q.34 What ar.e the- two basic purpos.es for computing a seasonal
index?
Ans. Seasonal indices are computed in order to use them (1) to make
short-range forecasts and (ii) eliminate the effec( of seasonal
variations from the original data.
By: Rafaqat
Q.35 What does the seasonal index measure?

Ans. It measures the average or typical seasonal paitern found in the
time series.
Q.36 What Is the difference between a stable seasonal index and a

moving ~easonal index?
Ans. A stable seasonal index has only 12 seasonal indei numbers, one
for each month A moving seasonal index has. several index
numbers for each month of each year covered in the analysis.
Moving seasonal indices are used to measure seasonal patterns
tlfat are changing.
Q.37 If the storm prevents people from shopping for two days, wllat .
type offluctuation in the sales would this cause: trend, seasonal,
cyclical, or irregu,lar? Why?
Ans.. "Irregular", because storms are irregular in the timings of the
occurrence.
Q.38 Can we explain this/actor mathematically?

Ans. As these factors are unpredictable we do not. attempt to explain it
mathematically.
Q.39 Wiiy an irregular fluctuation in tile business activity impossible

to forecast?
Ans. Irregular fluctuations are impossible to forecast the ~ents cause
them cannot be predicted wiih certainty.
Q.40 Define Smoothing. .

Ans. Smoothing techniques are used to reduce irregularities (random
fluctuations) in time series data. They provide a clearer view of the
true underlying behavior of the series. In some time series,
seasonal variation is so strong it obscures any trends or cycles,
which are very important for the understanding of the process
being observed Smoothing can remove seasonality and makes
By: Rafaqat
long term fluctuations 'in the series stand out more clearly.
Q.41 Wllat are different types of smoothing?

Ans. Moving average smdothing, exponential smoothing, running
medians smoothing, etc. .
Q.42 What is tile most commonly used type of smoothing?

Ans. The most common type of smoothing technique is moving a\lerage
smoothing. Since the type of seasonality will vary from series to
series, so must the type ofsmoothing.
Q.43 Define moving average smoothing.

Ans. A moving average is a form of average, which has been adjusted to
allow for seasonal or' cyclical components of a time series. Moving
1 average smoothing is a smoothing technique. used to make the
long-term trends of a time series clearer. When a variable, like the
number of unemployed, or the cost of strawberries, is graphed
against time, there are .likely to be considerable seasonal or
cyclical components in the variation. These may make it difficult to
see the underlying trend These components can be 'eliminated by
taking a suitable moving average. By reducing random
fluctuations, moving average smoothing malies long term trends
clearer.
Q.44 What is forecasting?

Ans. In time series analysis, forecasting is a process of assessing the
magnitude of a time series variable which it will assume at sdme
future point of time. Forecasting is based on the assumption that
the past pattern and behaviour of a variable will continue in the
future.
By: Rafaqat
Chapter JO: Analysis of Time Series 163 .
A Quick Approach to Statisticswith Questions and Answers
Exercises
Exercise 10 (MCQs')
Q.1 Data concerning events over a period of time is called a:
(a) Time Series
(b) Moving Average
(c) Frequency Distribution
(d) Random Sample
By: Rafaqat
Q.2 Which of the following is not a component of time series?

(a) Seasonal Variation
(b) Cyclical Variation
{c) Variance
, (d) Trend
Q.3 In a time series, secular variation is:

(a) - A variation that occurs at regular intervals
(b) The difference between the largest and smallest item in
any one-year
(c) The range of time to which the series applies
(d) The long-term trend
Q.4 A cyclical varfation is:

(a) One that takes a number of years to complete
(b) One that goes round in circles
(c) One that occurs four times a year
(d) A regular change
A Quick Approach to Statistics with Ques#ions and Answers
. Q.S The number of cars sold by a car-dealer during just 6 months in

2003 was as following;
January February March April May June
18 16 28 51 47 55
What'is the first 3 monthly moving average?

(a) 20.67
(b) 31.67
(c) 42.00
(d) 51.00
Q.6 A histogram is:

(a) A frequency graph
(b) A time series graph
By: Rafaqat
(c) A graph-plotting mean against standard deviation

( d) A correlative frequency chart
Q. 7 The weekly takings in a shop over 5 weeks were as follows:
Week 1 2 3 4 5
Taking (Rs.) 98 112 161 109 101
The first 4 weekly moving averages was:

(a) Rs. 105.00
(b) Rs. 109.00
(c) Rs. 120.00
(d) Rs. 120.75
.Q.8 The following table shows (in thousands) the number of units of
electricity used by a firm over a period of two years.
Quarter. l 23 4 1 2 3 4
No. of units . 85 49 25 87 89 53 29 86
(lOOO's)
What is the value of the second 4-quarterly moving average?

(a) 61.50 thousand
(b) 62.50 thousand

(c) 86.50 thousand

(d) 64.75 thousand
Q.9 . Suppose you were considering a time series of data for the
quarters of 1992 and 1993. The third quarter of 1993 would be
coded as:
(a) 2
(b) 3
(c) 5
(d) 6
Q. to Assume that you have been given quarterly sales data for a five-
year period. To use the ratio-to-moving-average method of
computing a seasonal index, your first step will be
(a) Compute the four-quarter moving average
By: Rafaqat
(b) ., Discard highest and lowest value for each quarter

(c) . Calculate the four-quarter moving total
(d) None of these
A Quick Appro{li:h to Statistics with Questions and Answers
' llltll Chapter 11
Index Numbers
Index
By: Rafaqat
A numerical scale used to compare variables with one another or with some
reference number. Jn other words, a number or ratio (a value on a scale of
measurement) derived from a series of observed facts.
Index Number
An index number measures how much a variable changes. over time or
space. This is a statistical.measure to give average change in a variable or
group of variables with respect to time or space.
We calculate an index number by finding the ratio of the current value to a
base value then we multiply the resulting number by I 00 to express the
index as a percentage. This final value is the percentage relative. Note that
the index number for the base point in time is always hundred.
Generally,
. Current Value
1n d exN um b er= x 100 .
Base Va/11e [.
Base Period
The period of time for which data used as the base of an index number or
other ratio. In other words, it is a period. from which the changes _are
measured..
Chapter 11: lndf!X Numbers 167
Types of Index Numbers

There are three principle types of indices:
Price Index
Quantity Index
Value Index
Price Index
A price index is the one most frequently used. It compares levels of prices
from one period to another.
Price IndexNumber Current Price x I 00 = p n x I 00

Base Price Po
By: Rafaqat
For example, if the retail price of sugar is Rs. 18 in 1998 and Rs. 30 in 2005
from the foilowing Table then the index number for 2005 ob the base price
in 1998 will be
P.. 100 =-.-x
Po. =p1.,.,.2,.,;=-x JO 100 = 166.
Pm; 100 =-x 67.
. . Po P1998 18
It mean~ that if the price of sugar/Kg. was Rs. I 00 in 1998 then it becomes
Rs. i66.67 in 2005 or the said price gets 66.67% increase in 2005 when we
compare that in 1998.
The quantity P. is also called Price Relative.
Po
Quantity Index .
A quantity index measures how much the number or quantity of a variable
changes over time.
From the following Table, the quantity index for year 2005 using year 1998
as base:
25
qO,n =qim.2005=q.X100=qlm;X100= X100=186.67.
90 q199K 18
Value Index
The vatue index measures the changes in total monetary worth. In fact, the
value index combines price and quantity changes to present a more
informative index. '
Chapter I I: Index Numbers 168
From the following Table, showing the current and base year prices and the
quantities consumed (in 000' Kg.), the value index for year 2005 using year
1998 as base:
v. OO Vims . 750
v0 . =v1m2ms =-x 1 =--x 100 =-xlOO= 277.78.
. v.. v,.,.,. 270

Table: Prices (Rs./Kg.), Quantities (consumed in 000' Kg.) and
values for Sugar
(period '0' denotes 1998 while 'n' denotes 2005)
By: Rafaqat
Simple and Aggregate Index Numbers

Simple index numbers calculate price changes for a single item over time.
. Index numbers are more accurate if they are constructed using actual prices .
paid for a single commodity, product or service rather than the more general
aggregated index.
Aggregate index numbers calculate price changes for a group of related
items over time. Aggregate indices permit analysis of price changes for the
group of related products, such as price changes for apples, oranges,
mangoes, et:c.
Unweighted Aggregate Index Numbers

In unweighted index numbers, all the values considered are of equal
importance or equally weighted. Aggregate means that we add, or sum, all
the values.
It has two types:
Simple Aggregate Index
Simple Average ofRelatives
Simple Aggregate Index

It measures the percentage change in the aggregate prices of a number of
commodities, at different periods. It is computed by dividing the sum of the
given year prices of all commodities by the sum of all the base year prices
. of the same commodities and expressing the results as a percentage.
Chapter 11: Index Numbers 169
A Quick Approach to Stall"stics with Questions and Answers
p = LP. x 100.
o.. LPo
Simple Average of Relatives

It is the average of price relatives in an index obtained by taking the average
of the price relatives of the given commodities for each year and expressing
the results as a percentage.
p
"
1
= -M r(i!..!.)
Po
x I 00
,
where Mis the number of commodities und~ study.
.;
By: Rafaqat
Weighted Index Numbers

Sometimes, we have to attach greater importance to changes in some
variables than to others when we compute an index: This weighting allows
us to include more information than just the change in .Price over time.
Generally, such index numbers measure the change in the prices of a group
of commodities when the relative importance of the commodities has been
taken int~ account.
It has two types:
Weighted Aggregate Price Index Numbers

Weighted Average of Relative Price Index Number
Weighted Aggregate Price Index Numbers

It is constructed for an aggregate of items (prices) that have been weighted ..
These weights are usually, corresponding quantities produced, consumed _or
sold.
i" Following are the diffe~en~ weighted price index numbers:
:::
Laspeyres ' Price Index = LP.% xlOO.

LPoqo .
Passche 's Price Index = LP.q"xIOO

LPoq. .
Fisher's Ideal Price Index =

-.
Marsha/I-Edgeworth 's Price Index = L p ( q. + qJ x I 00
LPo(q. +q.)
Walsh's Price Index = L p ..rq;q: x I 00

LPoJq.q.
Weighted Average of Relative Price Index Number

It is computed by multiplying each price relative ( P.. ) by its weight,.
Po
summing these products and dividing by the sum of the weights. The
By: Rafaqat
weights are always the total values of the commodities.

Following are the different types of such indices:
PhlUSibility (Tests) of Index Numbers

Following are the tests available to check the plausibility of index numbers:
Time Reversal Test

Circular Test
P~rmutation or Price .Booncing
Commensurability
Factor Reversal Test
Chapter 11: lndexNumbers 171
A Quick Approach_ to Statistics with Questions and Answer~
Time Reversal Test

This test requires. that the index formula produced consistent results whether
it is calculated going from period 0 to period nor from period n to period 0.
More specifically, if the price observations for period . 0 and n are
interchanged then the resulting price index should be the reciprocal of the
original index;
Symbolically,
Circular Test
This test often called transitivity is a multi-period test (essentially a test of .
chaining). It requires that the product of a price index obtained by going
By: Rafaqat
from period 0 to period 1 and from period l to 2 be the same as going

directly from period 0 to 2.
Symbolicaily,
Poi x Pu = Po2
Generally,
Poi x P12 x .. x P.-1 . x P.o = l.
Permutation or Price Bouncfog

This test requires that, if the order of the prices in either period 0 or period n
(or both) is changed but not the individual prices, the index number should
not change. This test is appropriate in situations where there is considerable
, volatility in prices, for example, because of seasonal factors or sales
competitions.
CoQlmensurability
This test requires that if the units of measurement of the items are changed
(for example, form Kgs. to _Tonnes), then the price index will not change.
Factor Reversal Test

This test requires t~at the product of the price index number for any period
and an index of quantity obtained. from the formul1:1 by interchanging the
price and quantity terms should equal the ratio of expenditure in that period
to the base period expenditure.
Chapter 11: Index Numbers 172 c
A Quick Approach to Statistics with Questions a1'd Answers
In other words, if the factor 'p' is changed by the factor 'q', index formula
be interchanged (9r reversed) so that a quantity (or prke) index formula is
obtained, then the product of the two index numbers should equal the value
index number.
That is,
(Price Index) x (Quantity Index)= Value Index.
Consumer Price Index (CPI)

Consumer Price Indices measure changes over time in the general level of
prices of goods and services that a reference population acquires, uses or
pays for consumption. A Consumer Price Index is estimated as a series of
summary measures of the period-to-period proportional change in the prices
of a fixed set of consumer goods and services of constant quantity and
By: Rafaqat
characteristics acquired, used or paid for by the reference population. Each

of the elementary aggregate indices is estimated using a sample of prices for
a defined set of goods and services obtained in, or by residents of, a specific
region from a given set of outlets or other sources of consumption goods
and .ser\tices:
' .
'I
A Quick Approach to Statistics with Questions and Answer$
Questions-Answers
Q. l What is the main purpose for construction of index numbers?

Explain with examples.
Ans. The main use of index number comes when we are interested to
know that how much something has changed over a period of time.
We may want to know how much the prices of certain-daily used
items have increased so we can adjust our budget accordingly. A
factory' manager may .wish to compare this month's per unit
production cost with that ofpast six months. Or a medical research
team may wish to compare the number of malaria cases reported
By: Rafaqat
this year with the number reported in previous years. In each of

these situations, we require the degree of change with respect to
time. Typically, we use index numbers to measure such differences.
..,.. Quick Approach to Statistics with Questions and Answers
is employed: In chain base method every period is a base periiJd

for its next period
Q.5 What are the main characteristics of the base period?

Ans. The base period should be a normal period It means that a period
of economic stability and free from any major financial crisis
caused by inflation, depression, ~ars, labour unrest, strikes_,
natural disasters etc. This period should not be too far distant in
the past making the comparisons meaningless.
Q.6 What' is the m_ost appropriate time duration oft/U! base period?
Alis. This period is frequently one year but it may be as~sh'ort as one day
or as long as the average of a group ofyears.
By: Rafaqat
Q. 7 What the term "Weighting" reflects in measuring index

numbers?
Ans. Sometimes, we have to attach greater importance to changes in
some variables than to others when we compute an index. This
weighting allows us to include more ir(ormation than just the
change in price over time. The process whiCh assigns numerical ,,.,
coefficients (weights or weighting/actors) to each ofthe element in
a data set, in order to provide them with a desired degree of
impo;tance relative to one another.
Q.8 What is the most popular type ofprice Inda number?

Ans. 'The CPI (given above) is the most common' example ofprice index
number.
Q. 9 . What is the main purpose ofconstruction of CPI?

Ans. It measures overall price changes of a variety of consumer goods
and services and is used to define cost ofliving.
Q.10 What are the main methods.to compute CPI?

Ans. There are two methods used to construct CPl They are:
(i} Aggregate Expenditure Method
(ii) Household Budget Meth~d
Chapter 11: ittdex Numbers 175
A Quick Approach to Statistics with Questitms and Answers
Q.I 1 What are the main shortcomings of CPI?

Ans. The construction of CPI involves the sampling of goods and
services, the sampling errors and biases may affect fndices and
render them to suspect.
In case .of certain goods, it is difficult to collect prices actually.
needed For example, the prices for clothing usually relate to cloth
and not to tailored clothes.
Q.1). What is Producer Price Index (PPI)? .

Ans. PP/ is a composite figure of producer's prlces of representative
commodities included in the market basket. It is used to measure
monthly or yearly change in producer's prices of key commodities
in the manufacturing sector. It is used to deflate production
indicators and serves as the dejlqtor in the estimation of
By: Rafaqat
manufacturing production in the system ofnational accounts.
Q.13 What are the main shortcomings in construction of index

numbers?
Ans. The main shortcomings are:
(i) It is very difficult to collect prices of the commodities
included in the construction of indices as well as to take
into account all changes in quantity or product.
(ii) Sampling error is expected to be found in the calculation
of Indices. ..
(iii) The choice of a normal period is a difficult task as, at a
time, more ihan one period clln be considered as normal
for all segments ofthe economy.
,,
1
Exercises

\
Q.1 To measure changes in total monetary worth, one should calculate:

(a) Price index
(b) Quantity index
( c) Value index
By: Rafaqat
(d) CPI
Q.2 If an index number calculations oYer 8 years with a base value of

100 gave an index for 2003 of 120, what would be the percentage
relative for 2003? ".4
<. .
a). 100
b) 120
c) 880
.. d) 960
Q.3 Which of the following describes an advantage of using the

Laspeyers' method?
. (a) Many commonly used quantity measures are not tabulated
for every period.
(b) Changes in consumption patterns are taken into account.
(c) One index can be easily compared with another.
(d) (a) and (c) but not (b)
Q.4 When computing a we.ighted average of relative index, we would

be best able to compare indices from various periods if:.
(a) Base values were used as p,,q,,
(b) Current values were used as p,,q,,
( c) Fixed values were used as p,,q,,
(d)' Either base or fixed values were used as p,,q,,
<"'.
\\,,
i
'l
,/'
A Quick Approach to Statistics with Questions.and Answe1
Q.5 Commodities subject to considerable price variations could best b

measured by:
(a) Price.index
(\>) Quantity index
(c) Value index
(d) CPI
Q.6 A base period can ~e described as a normal period if:

(a) It is neither the peak nor the trough ofa fluctuation
(b) It must be the most recent period for which we have data
(c) There was no inflationor deflation of prices during th1
period.
(d) {a) and (c) above
Q. 7 The weights used in a quantity index are:

By: Rafaqat
(a) Percentages of total quantity

(b) Prices
(c) Average of quantities
(d) None of these
Q.8 To measure how much the cost of some variable changes over
time, we would use:
(a} Inflation index
(b) Quantity index
(c) Value index
(d} None of these
Q.9 When the base year values are used as weights, the weighted
average of relative price index is the same as:
(a) The Paasche's index
(b) The Laspeyers' index
(c) The unweighted average of relative price index
. (d) None of these
~I
Chapter 11: IndexNufnhers 178
A QuickApproach io Statistics with .Questions and Answers
Q.10 A primary difference between average of relatives and aggregate

methods is that:
(a) Aggregate methods sum all prices b~fore finding the
ratio
(b) Average of relatives methods sum all prices before
finding the ratio
(c) Aggregates methods are useful only for price indices
(d) (a) and (c) but not (b)

By: Rafaqat
Read the following .statements carefully and indicate which statement is

"True" or "False":
L The price index numbers are used to measure changes in a

particular group of prices and help us in comparing the
movement in p~ices of one commodity with another.
2. Index numbers give very valid comparison of changes in
variable over long periods.
3. All index numbers are not suitable for all purposes.
4. lndex numbers cannot be used to measure enrollment changes in
'
an institution.
5. There are two ways for the selection of base period.
6. The arithmetic mean; median and geometric mean are the ~sual
measures for the calculatian of average in index numbers.
7. Weighted aggregate index number uses the relative importance I;
of the index numbers.
8. Fisher ideal index has a theoretical advantage over other index

numbers as i! is the only index that follows the time reversal,
factor reversal and circular test.
9. A CPI is used to measure changes in the composite price of a
specified basket of goods and services during the given period as
compared with the base period.
io. The indices calculated from different index numbers may not
agree.
11. Fisher ideal index is the hybrid of two index numbers.
12. Index numbers cannot measure differences in a given variable in
By: Rafaqat
several locations.
13. A simplest form of a composite index is a weighted aggregate
index.
14. An index number is alWl!YS found by taking the ratio of current
value to a base value and multiplying by I 00.
15. The simple average of relatives method divides the weighted
sum by the sum of weiWits.
I
~I
Chapter 11: IndexNumbers 180
Chapter 12
1111111
Nonparametric Statistics
By: Rafaqat
Nonparametric Tests
Parametric tests require assumptions about the nature or shape of the
populations involved; nonparametric tests do not require such assumptions;
Consequently, nonparametric tests of hypotheses are often called
Distribution Free Tests.
Some of these tests are. Sign test, Wilcoxon Signed-Rank test, Runs test;
Mann-Whitney Utest, Kruskal Wallis test, etc.
Nonparametric tests may be, and often are, more powerful in detecting
population differences when certain assumptions are not satisfied.
Nonparametric methods have a number of clear ;:tdvantages over parametric

methods. The advantages and disadvantages are as under;
Advantages: .
1. They do not require us to make the assumption that a population is
.distributed in the shape of normal curve or another specific shape.
2. Generally, they are easier to handle and to understand.
3. Unlike the parametric mythods, nonparametric methods can often be
applied to non-numeric data such as the gender of survey
respondents.
4. Some times, even formal ordering or ranking is not required.
Disadvantages:
I I. They ignore a certain amount of information.
2. They are often not as efficient or sharp as the parametric tests.
Chapter 12: Nonpar:ametric Statistics 181
A Quick Approqch to Statistics with Questions and Answers
3. Nonparametric tests typically require data to be rank-ordered

and/or grouped according to nominal classifications.
Sign Test
The sign test is designed to test a hypothesis about the location of a
population distribution. It is most often used to test the hypothesis about a
population median, and often involves the use of matched pairs, for
example, before and after data, in which case ittests for a median difference
of zero.
We can use a signed test;
(i) To test claims about the median of the paired differences for two
dependent samples
(ii) To test claims about certain types of nominal data
By: Rafaqat
(iii) To test the claim. made about the median of a single population.
The Sign test does not require the assumption that the population is
normally distributed. In many applications, this test is used in place of the
one sample t-test when the normality assumption is questionable. This test
can also be applied when the observations in a sample of data are ranks, that
is, ordinal data rather than direct measurements.
Wilcoxon Signed Ranks Test

The signed test considers only whether each data value is above (+) or
below (-) the value M given for the median in the null hypothesis. It does
not take into account how much each data value is above or below M thus a
large amount of pertinent information is ignored. The Wilcoxon Signed
Ranks test incorporates these informations. The Wilcoxon Signed Ranks
test is designed to test a hypothesis abot1t the location (median) of a
population distribution. It often involv.es the use of matched pairs, for
example, before and after data, in which case it tests for a median difference
of zero.
In many applications, this test is used in place of the one sample t-test when
the normality assumption is questionable. It is a more powerful alternative
to the sign test, bukloes assume that the population probability distribution
is symmetric.
This test can also be applied when the observations in a sample of data are
ranks, that is, ordinal data rather than direct measurements.
Chapter!2: Nonparametric Statistics 182
~ .
Mann-Whitney UJ'est
The Mann-Whitney U test is one of the most powerful nonparametric tests
for comparing two populations. It is ~sed as a test of comparison of medianS
or means of two populations using independent samples.
The Mann-Whitney test does not require the assumption that the differences
between the two samples are normally distributed. In many applications, the
Mann-Whitney test is used in place of the two sample t-test when the
normality assumption is questionable.
This test can also be applied when the observations in a sample of data are
ranks, that is, ordinal data rather than direct measurements. This test is also
known as Mann-Whitney Utest and Wilcoxon Rarik-Sum test.
Kruskal-Wallis Test .
By: Rafaqat
The Kruskal-Wallis test 'is a nonparametric test used to compare three or

more populations. It is used to test the null hypothesis that all populations
have identical distribution functions against the alternative hypothesis that
at least two of the samples differ only with respect to location (medici), if
at all. .
It is the analogue to the F-test used in E\nalysis of variance. Whlle analysis
of variance test . depends on the assumption that all populations under
comparison are normally distributed, the Kruskal-Wallis test places no such
restriction on the comparison. It is a logicaiextension of the Mann-Whitney
test..
It is also known as H-test. This H statistic has a distribution that can be
approximated by the Chi;square distribution as long as each sample has at
least five observations. It is therefore a right-tailed test.
Runs Test
In studies .where measurements are made according to some well defined
ordering, either in time or space, a frequent question is whether or not the
average value. of the measurement is different at different points in the
syquence. The runs test provides a means of testing this. In other words; it is
a grocedure for testing the randomness of data.
Run:
A run is a sequence of data that exhibits the sam~ characteristic; the
sequence is preceded and followed by different data or no data at all. In
other words, it is maximal sequence of similar elements.
Chapter 12: Nonparametric Stat~tics 183
The number of runs in the sequence,.SS F S FFF SSS F, is 6.
Spearman's Ran~ Correlation Test

Spearman's Rank Correlation test is a nonparametric test for testing the
significance of Spearman 's . Rank correlation that is the nonparametric
counterpart of that parametric measure. Rank correlation is the measure of
the correlation that exists between the two sets of ranks, a measure of the
degree of association between the variables that we would not have been
able to calculate otherwise.
Kolmogorov-Smirnov Test
For a single sample of data, the Kolmogorov-Smimov test is used to test
whether or not the sample of data is consistent with a specified distribution
function. When there are two samples of data, it is used to test whether or
By: Rafaqat
not these two samples may reasonably be assumed to come from the same
distribution. The Kolmogorav-Smimov Test is therefore another measure of
the goodness of fit of the theoretical frequency distribution as was the Chi-
square test. The Kolmogorov-Smirnov test does not require the assumption
that the population is normally distributed.
Chapter 12: Nonparametric Statistics 184
..
A Quick Approach to Statistics with Questions and Aftswers
Questions-Answers
Q.J What are the usual normal-theory-based procedures for

hypotheses testing?
Ans. For hypothesis testing z-test, t-test, Chi-square test, F-test are
usual normal-theory-based procedures.
Q.2 When do we use nonparametric tests?

Ans. The following are some situations in which the use of
nonparametric tests is appropriate;
By: Rafaqat
(i) The hypothesis to be tested does not involve a population

parameter.
(ii) The data have been measured on a scale weaker than that
required for the parametric procedure that would otherwise
be employed For example, the data may consist of count or
rank data,
(iii) The assumptions necessary for the valid use of parametric
procedures are not met.
.(iv) Results are needed in a hurry and calculations must be done
by hand
Q.3 What is the nonparametric alternative of the two sample I-test for
independent samples?
Ans. Usually, when we have two samples that we want to compare
concerning their mean value for some variable of interest, we
would use the I-test for independent samples; nonparametric
alternatives for this test is the Mann~Whitney U test.
Q.4 What is the. nonparametric alternatives of the ANOVA, when

samples are independent? . .
Ans. If we have multiple groups, .we would use analysis of variance
(ANOVA); the nonparametric equivalents to this method are the
Kruskal-Wa/lis test.
_Chapter 12: Nonparametric Statistics 185
.A Quick Approach to Statistics with Questions and Answers
What are the nonpatametric alternatives of the ANOVA, when

samples are dependent?
Ans. If there are 1r1ore than two varlables that were measured in the
same data, then We would customarily use repeated weasures
ANOVA. Nonparametric alternatives to this method are
Friedman's two-way analysis of variance and Cochran Q test (if
the variable was measured in terms of categories, e.g., "passed"
vs. ''failed''). Cochran Q is particularly useful for measuring
changes infrequencies (proportions) across time.
Q.6 What are the nonparametric qlternatives of the two sample t-test
' for dependent samples?
Ans. Ifwe want to compare two variables measured in the same sample
By: Rafaqat
we would customarily use _the matched pair t-tesl (or I-test for
dependent samples). Nonparametric alternatives to this test are the
Sign test and Wilcoxon's matched pairs test.
Q.7 What npnparametric tests are used to . test the relationship

between variables?
Ans. To express a relationship between two variables one usually
computes the correlation coefficient. Nonparametric equivalents to
the standard correlation coefficient are Spearman'R, Kendall Tau,
and coefficient Gamma. If the tWo variables of interest are
categorical in nature (e.g., ''passed" vs. ''failed" by "male" vs.
''female'') appropriate nonparametric statistics for testing the
relationship between the two variables are the Chi-square test, the
Phi coefficient, and the Fisher exact test. In addition, a
simultaneous test for relationships between multiRle cases is
available i. e Kendafl coefficient of concordance.
Q.8 For what purpose Kendall coefficient of concordance is used?

Ans. Kendall coefficieni of concordance is often used for expressing
inter-rater agreement among independent judges who are raiing
(ranking) the same stimuli. It is extension of rank correlation
coefficient.
Q.9 What can be determined with the Mann-Whitney test? Wl}at kind
of data are used/or this test'! .
Ans. The I-test and z-test are useful for testing whether two samples
have been drawn from populations that are assumed to , be
normally distributed and which have equal means and variances.
However, the Mann-Whitney test enables us to make a significance
test without assumptions, although the data/or the Mann-Whitney
test are assumed to be continuous and which must be ranked
Q.10 What is the purpose of the Kruskal-Wallis H test'! What type of

data are used for this test?
A11s. This test is used to te~t the hypothesis that the several samples
come from populations with the same means: It requires no
assumptions about the distributions of the populations from which
the samples were drawn except that the population data are
By: Rafaqat
continuous. To make this test the data for all samples must be
pooled a.nd ranked in order ofmagnitude.
Q.11 What is the null hypothesis when we use the Spearman Rank
correlation coefficient test?
Ans. The null hypothesis is that there is no significant co"elation
. between the two rankings.
Q.12 Discuss the merits and demerits of Kolmogorov-Smirnov(K-S)

goodness offlt test with comparison of Chi-square goodness offit
test
Ans. There are some merits and demerit$ of K-S goodness offit test with
comparison of Chi-square goodness offit tesJ:
(i) The K-S test does not require that the observations be
grouped as in the case with the Chi-Square test. The
consequence of this difference is that the K-S test makes use
of all the information present in a set of table.
(ii) The K-S test can be used with any size of sample. It will be
g recalled that certain minimum sample sizes are required for
g the use ofthe Chi-square test.
(iii) K-S test is not applicable when parameters have to be
submitted from the sample. The Chi-squqre test may be ilsed
in these situations by reducing the degrees offreedom by 1
for each parameter estimated.
:6
Q.13 Define non-randomness.

Ans. A very large or very small number of runs in a sequence show non- -
randomness.
Q.14 What are the major nonparametric tests used to test the
randomness in the data?
Ans. Runs test for randomness and Kendal Tau test are the major tests
used to test the randomness in the data.
Q.15 Give any example/or runs test?

Ans. Suppose that, as part of a screening proiram for heart disease,
men aged 45-:-65 years have their blood cholesterol level measured
on entry to the study. After _many months it is noticed that
cholesterol levels in this population appear somewhat higher in the
By: Rafaqat
Winter than in the Summer. This could be tested formally using a

Runs test on the recorded data, first arranging the measurements
in the date order in which they were collected
. '
Exercises
Q.1 The sigri test is: .

(a) Less powerful than that of the Wilcoxon signed rank test
(b) More powerful than the paired sample ~-te.st
(c) More powerful than the Wilcoxon signed rank test.
(d) Equivalent to Mann-Whitney t~st
By: Rafaqat
Q.2 The rtonparametric equivalent of an unpaired samples t-test is the:

(a) Sign test
(b). Wilcoxon signed rank test
(c) Mann- WhitneyUtest.
(d) Kruskal-Wallis Test
Q.3 The M~-Whitney Utest is preferred to a't-test when:

(a) Data are paired. ,
(b) Sample sizes are small
(c) The assumption of normality is not met.
(d) Samples are dependent
Q.4 When using the sign test, if two scores are tied, then we:
(a) Count them
(b) Discard them
(c) Depends upon the scores
(d) None of these
Q.S The sign test assumes that the:

(a) Samples are independent
(b) Samples are dependent
(c) Samples have the same mean
(d) None of these
/
A Quick Approach to $tatistics with Questions and Answer
Q.6 When testing for randomness, we can use:

(a) Mann-Whitney Utest
(b) Sign test
( c) Runs test
(d) None of these
Q.7 The Runs test results in rejecting the null hypothesis c 1
randomness when:
(a) There is an unusually large number of runs
(b) There is an unusually small number of runs
(c) Either of the above
(d) None of the above
Q.8 The Wilcoxon fa~k-sum test can be:

By: Rafaqat
(a) Upper tailed.

(b) Lower tailed
Q.9 The Wilcoxon rank-sum test compares:

(a) Two populations
(b) Three populations
(c) A sample mean to the population mean.
( d) Any number of populations
Q.10 The Wilcoxon signed rank is used:.

(a) Only with independent samples
(b) Only in matched pair~ samples
(c) As an alternative to the Kruskal-Wallis test
(d) To test for randomness
Q.11 Which of the following test use rank sums?

(a) Ftest
(b) Chi-square and Sign tests
( c) Runs test
(d) Kruskal-Wallis and Wilcoxon tests
Chapter 12: Nonparametric Statistics 19'
Q.12 '
Which of the following tests must be two-sided?
(a) Kruskal-Wallis
(b) Wilcoxon Signed rank
(c) Runs test
(d) 'sign test
Q.13 In testing for the difference between two populations, it is possible

to use:
(a) The Wilcoxon rank-sum test
(b) The sign test
Q.14 In a Wilcoxon rank-sum test:

By: Rafaqat
(a) Ties never affect the decision

(b) Ties always affect the decision
(c) Ties within one sample may affect the decision
(d) Ties between the two samples may affect the decision
Q.15 The Spearman rank-correlation test requires that the:

(a) Data must be measured on the same scale
(b) Data at least ordinal scaled
(c) Data must be from two ,independent samples
(d) Data must be distributed at least approximately as a t-
distribution
Q.16 To perform a runs test for randomness the data must be:
(a) Qualitative
(b) Quan_tjtative
(c) Divided into at least two classifications
(d) Divided into exactly two classifications
Q.17 To compare the annual income of engineers with those of clerks,

two random samples are obtained n1 = 20 and n1 = 20. A decision
was made to use the Wilcoxon rank-sum test to determine if clerks
earn more than,engineers. Which of the following can be a valid
reason for this decision.
(a) The sample are too small to estimate and d
(b) The samples do not adequately represent the populations
Ch I. ter 12: Nonptilametric Statistics
1
l9 l
( c) The samples are from non-normal distributions

(d) The samples are not C>f equal size
. Q.18 Three brands of coffee are rated for taste on a scale of I to 10. Six
persons are asked to rate each brand so that there is a total of 18
observations. The appropriate test to determine if three -brands taste
equally goodis:
(a) One way analysis of variance
(b) Wilcoxon rank-sum test
(c) Spearman rank difference
(d) Kruskal-Wallis test
Q.19 If a Chi-square goodness of fit test has 6 categories and an N=30,

By: Rafaqat
then the correct number of degrees of freedom is:

(a) 5
(b) 6
(c) 28
(d) 29.
Q.20 Comparing the times-to-failure of radar transponders made by

firms A, B, and C, based on an airline's sample experience with the
three types of instruments, one may well call for:
(a) A Kolmogorov-Sm~mov test
{b) A Kruskal-Wallis te~
(c) A Wilcoxon rank~silm test
( d) A. Spearman rank ooqelatioii test
Q.21 Which of the follo~ing tests is~ most)ikeiy. asse5sing this null
hypothesis: Ho: The number of vio~atii>ns per aPartJneni in the
population of all city apartments is binomially distributed with a ,
probability of success in any one trial of p =0.3:
(a) The Kolmogorov-Smirnov test
(b) The Kruskal-Wallis test
( c) The Mann-Whitney test
(d) The Wilcoxon signed-rank test
..
A Quick Approach to Statistics with Questions ad Answn
Q.22 In the Kruskal-Wallis test of k samples, the appropriate number of

degrees of freedom is
(a) k
. (b) k-1
(c) nt-1
(d) n-k
Q.23 When compare to parametric methods, nonparametric methods are

(a) Less accurate
(b) Less efficient
(c) Computationally easier
(d) (b) and (c) but not (a)
By: Rafaqat

Read th~ following statements carefully and indicate which statement is
"True" or "False":
1. Ties between paired observations should be deleted before a sign

test is made.
2.. The Mann-Whitney Utest is a two tailed test.
3. The Mann-Whitney U test can be used to test if two samples
came from the same population or identical populations.
4. If six or more observations are in each sample, the Kruskal-
Wallis Htest is distributed as Chi-~quare.
' 5. The Kruskal-WaUis H test can be used to determine if several
samples came from populations with equal means.
6. A Chi-square analysis should not be originally rriade when there
are fewer than S observations in one or more of the cells.
7. The runs test is upper tailed.
8. The Wilcoxon signed rank test may be used whenever the sign
test is applicable.
9; The binomial distribution can be used to provide probabilities
for outcomes when the sign test is used.
10. One of the main advantage of the nonparametrk tests is that the
underlying assumptions .are often less restrictive than those of
parametric tests.
I I. Parametric tests are easier to compute and therefore more
desirable than nonparametric tests.
By: Rafaqat
12. Nonparametric tests are more powerful than parametric tests

because they are not bound by population assumptions.
13. Parametric tests are appropriate when the dependent variable is
the number of correct trials out 6f a fixed number of trials.
14. The nonparametric tests apply only to populations that have no
parameters.
15. A nonparametric test is often less efficient than its parametric
counterpart, because a nonparametric test utilizes less of the
information contai~ed in the sample.
Answers to Exercises
Chapter 1: Introductory Statistic~

By: Rafaqat

1. d 2. d 3. b 4. a 5. a.

1. False 2. True 3. True 4. False 5. False
6. True 7. False 8. True 9. True 10. True
11. True 12. False 13. True 14. False 15. False
16. False 17. False 18.. True 19. True 20. False
21. False 22. True 23. False 24. False 25. True
26. False 27. True '28. False 29. True 30. False
31. True 32. False 33. False 34. True 35. True
36. False 37. False 38. False 39. True 40. True

A Quick Appro~ch to Statistics with Questions and Answers
Chapter 2: Basic Probability Theo'ry
l. d 2. a 3. d 4. b 5. a
6. d 7. c 8. c 9. d 10. c
11. d 12. c 13. a 14. b 15. a
16. b 17. a 18. c 19. b 20. d
By: Rafaqat

1. True 2. True 3. False 4. True 5 False
6. True 7. . True 8. False 9. True 10. False
11. True 12. True 13. True 14. True 15. False
16. False 17. False 18. False 19. False. 20. True
21. True 22. True 23. True 24. False 25. False
26. False 27. False 28. False 29. True 30. False
Chapter 3: Random Variables
1. False 2. True 3. False 4. Ttue 5. True
6. False 7. False 8. True 9. True 10. False
i-
' A Quick Approach to Statistics with Questions and Answers
Chapter 4: Discrete Probability Distributions

I. b 2. c 3. d 4. a s. c
6. b 7. a 8. d 9. b 10. d
11. d 12. d 13. c 14. a IS. d
E~ercise 4.2 (Tru~alse)

By: Rafaqat
I. False 2. Tr:ue 3. True 4. False 5. False

6. False 7. False 8. .True 9. True 10. True
Chapter S:: Continuous Probability Distributions

Exercise 5.1 ~CQs')
1. b 2. d .3. c 4. c s. b
6. b 7. .d 8. d 9. d 10. b
11. c 12. c 13. c 14. b IS. d
16. b 17. d is. c 19. b 20. b
Exercise S.2 (True/False)

I. False 2. .False 3. True
.
4. False 5. True
6. True. 7. True 8. False 9. True JO. True
11. False 12. False 13. True 14. False 15. Tni.e
j
Chapter 6: Regression & C~nrelation
Exercise 6 (MCQs')
1. a 2. c 3. c 4. b 5. b
6. a 7. a 8. c 9. d 10. c
11. b 12.. c 13. b 14. c 15. d
16. b 17. b 18. d 19. b 20. b
By: Rafaqat
Chapter 7: Sampling
1. True 2. False 3. True 4. False 5. False
6. False 7. True 8. True 9. True 10. True
11. True 12. True 13. True 14. False 15. False
Chapter 8: Statistical Inference
Exercise 8 (True/F~lse)
l. False 2. False 3. True 4. False 5. True
6. False 7. True 8. True 9. True 10. False
16. False 17. False 18. False 19. True 20. False
21. True 22. True 23. False 24. False 25. False
26. False 27. True 28; False 29. True 30. True
.A Quick Approach to Statistics with Questions and Answers
Chapter 9: Design and Analysis of Experiments
. Exercise 9 (True/False)
1. Fats~ 2. True 3. True 4. False 5. True
6. False 7. False 8. True 9. False 10. True
11. True 12. Tnie 13. True 14. True 15. True
Chapter 10: Analysis of Time Series

By: Rafaqat
Exercise 10 (MCQs')
1. a 2. c 3. d 4. b 5. a
6. a 7. c 8. b 9. c 10. c
Chapter 11: Iridex Numbers

1. c 2. b 3. d 4. d 5. b
6. d 7. b 8. d 9. b 10. a .
1. True 2. False 3. True 4. False s. True

6. True 7. True 8. True 9. True 10. True
11. True 12. False 13. False 14. True 15. False
A Quick Appro.ach 'to Statistics with Questions and Answers
Chapter 12: Nonparametric Statistics
1. a 2. c 3. c 4. b 5. b
6. c 7. c 8. c 9. a 10. b
11. d 12. d 13. c 14. b 15. b
16. d 17. 6 18. d 19. a 20. b
21. a 22. b 23. d
By: Rafaqat

1. True '2. False. 3. True 4. True 5. True
11. True 12. False 13. False 14. False 15. True
1111111
Bibliograp_hy
I. Brockwell, P. J. and Davis, R. A. (2002). "Time Series Theory and

Methods", Springer Verlag, New York.
2. Chaudhry, S. M. and Kamal, S. (1996),_ "Introduction to Statistical
Theory" Part l, II, 6th ed., Ilmi Kitab Khana, Lahore, Pakistan.
By: Rafaqat
3. Clark, G. M. and Cooke, D., (1998), "A Basic Course in

Statistics" 4th ed., Arnold, London.
4. Conover, W. J: (1999), 1'Practical Nonparametric Statistics", 3rd
Edition, John Wiley and Sons, New York.
5. Des Raj and Chandhok, P. (1998), "Sample Survey Theory",
Narosa Publishing House, New Delhi.
6. Draper, N.R. and Smith, H. (2004)," Applied Regression
Analysis", John Wiley and Sons. New York.
7. Gujrati, D. (2003), "Eco_nometrics", .John Wiley and Sons, New
York.
8. Hogg, R.V. and Craig, A.T. (2001), "Introduction to Mathematical
Statistics". Prentice Hall, Engle wood Cliffs, New Jersey.
9. Lindgren, B:w. (1998), "Statistical Theory", Chapman and Hall,
New York.
IO. Lyman Ott, Michael, T. Longnecker, R. (2001), "An Introduction
to Statistical Methods and Data Analysis", Brooks Cole, 5th ed.
Bibliography 20 I
A Quick Approach to Statistic$ with Questions and Answers
l l. McClave, J. T., Benson, P. G. and Sincich, T. (1998), "Statistics

, for Business & Economics" 7rlt ed., Prentice Hall, New Jersey.
12. Montgomery, D.C. (2000), "D~ign and Analysis of Experiments",
John Wiley, New York.
13. Mood, AM., Graybill, F.A. and Boss, D.C. (2003), "Introduction
tQ the Theory ofStatistics", McGraw H.ill. New York
14. Mukhopadhayay, P., (2005), "Theory and Methods of Survey
Samplint', Prentice-Hall, India.
15. Richard I. Levin and David S. Rubin. (1997), "Statistics for
Management", Prentice Hall.
By: Rafaqat
16. Rosenbalatt., R.(2001 ), "Basic Statistical Methods and Models for

the Sciences", Univ. of Texas.
17. Spiegel, M.R., Schiller, J. L. and Sirinivasan, R. L. (2000)
"Probability and Statistics", 2nd ed. Schaums Outlines Series.
McGraw Hill. N.Y.
18. Stirzaker, D. (1999), "Probability and Random Variables",.
Cambridge University Press, Cambridge.
19. Walpole, R.E., Mayer, R.H and Mayer, S.L. (1998), "Probability
and Statistics for Engineers and Scientisf' 6rJt ed., Prentice Hall,
N.Y.
20. Weiss, N. A. (2002), "Introductory Statistics" 4th ed., Addison-
. Wesley Pub. Company, Inc.
Bibliography 202
1111111111
Subject Index
Class Boundaries, 5
A Class Interval (Width), 5
Class Limits, 4
Addition Rule of Probability, 28 Class Mark (Midpoints), 5
By: Rafaqat
Alternative Hypothesis, 129 Classification, 4

Analysis of Time Series, 153 Cluster Sampling, I IO
Analysis of Variance (ANOVA), Coefficient of Determination, 93
145 Coefficient of Variation (CV), 9
Arithmetic Mean, 7 Collectively Exhaustive Events,
Axiomatic Definition of 25
Probability, 26 Combinations, 29
Commensurability, I 7 I
B Complementary Events, 25
Composite Hypothesis, 130
Base Period, I 67
Conditional Probability, 27
Bernoulli Trial, 49
. Consistency, 128 .
Bias, I I I
Consumer Price Index (CPI), 173
Binomial Distribution, 49
Continuous Data, 2
Continuous Probability
c Distributions, 63
Census, 11 Continuous Random Variable, 41
Convenience Sampling, 107
Central Limit Theorem (CLT),
Correlation, 94
69
Characteristic Function, 45
Coverag~ Error, 111
Charts, 6 Critical Region, 131
'Critical Value, 131
Chebychev's Inequality, 45
Cumulant-Generating Function,
Chi-Square Distribution, 72
44
Circular Test, 171.
Subject Index 203
A Quick Approach to.Statistics with Questions and Answers
Cumulative Distribution Factor of an Experiment, 145

Function, 43 Factor Reversal Test, 171
Cumulative Frequency, 4 Frequency,.4.
Cumulative Frequency Polygon Frequency Curve, 6
(Ogive), 6 Frequency Distribution, 4
Cyclical Fluctuations, 153 Frequency Polygon, 6
D G
Data, l Gamma . Distribution, 70
Deciles, 8 Geometric Distribution, 52
Descriptive Statistics, l Geometric Mean, 7
Design of Experiment (DOE), Graph, 5
143 Grouped Data, 4
Diagram, 6
By: Rafaqat
Discrete Data, 2 H
Discrete Probability
Distributions, 49 Harmonic Mean, 7
Discrete Random Variable, 41 Histogram, 5
Dispersion, 9 Homogeneity, 145
Hypergeometric Distribution, 51
Hypothesis, 128
E
Efficiency, 128 I
Equally Likely Events, 24
Estimate, 127 Independent Events, 25
Estimation, 127 Independent Random Variables,
Estimation Error, 112 44
Estimator, 127 Index, 167
Event, 24 Index Number, 167
Event Space, 24 Inferential Statistics, 1
Expected Value, 43 Interval Scale, 3
Experiment, 143 Irregular Variations, 153
Experimental Unit, 143
Explained Variation, 93
Exponential Distribution, 65 l
Judgement Sampling, I 08 ]
]
F
1
F -Distribution, 77 1
Subject Index 204 }
K Normality, 145
Null Hypothesis, 129
Kolmogorov;Smimov Test; 184
Kruskal-Wallis Test., 183
0
L Obse~ation, v
One"'.Sided (One-Tailed) Test,
.Law of Total Probability, 28 131
Local Control, 144 Ordinal Scale, 3
Outcome, 23
M
p
Mann-Whitney U Test, 183
Mean Deviation, 9 Parameter, l 06
By: Rafaqat
Measure of Central Tendency, 7 Partial Correlation, 94

Measurement Error, 112 Percentile, 8
Median, 7 Permutation or Price Bouncing,
Method of Least Squares, 92 171
Mode, 8 Permutations, 29
Moment-Generating Function, 44 Poisson Distribution, 54
Moments, 10 Population, I 05
Multinomial Distribution, 56 Power of the Test, 132
Multiple Regression, 92 Price Index, 168
Multiple Regression Correl~tion Primary Data, 2
Coefficient, 94 Probability, 2
Multiplication Rule of Probability Density Function, 42
Probability, 28 Probability Distribution, 42
Mutually Exclusive or Disjoint Probability Function, 42
Events, 24 Processing Error, 112
Purposive Sampling, I 07
N P-Value, 132
Negative Binomial (Pascal)

Distribution, 53
Q
Nominal Scale, 3 Qualitative Data, 2
Nonparametric Statistics, 181 Quantitative Data, 2
Nonparametric Tests, 181 Quantity Index. 168
Non-response Error, 112 Quartile, 8
Non-sampling Error, 111 . Quartile Deviation, 9
Normat Distribution, 67 Quota Sampling, I 08
Subject Index' 205
R Simple Regression, 92
Skewness, IO
Random Experiment, 23
Random Variables, 41
Randomization, 144
Range, 9
,(~tio Scale, 3
Regression Analysis, 91
Regression Equation/Model, 91
Snowball Sampling, I 08
Standard Deviation, 9
Standard Error of Estimate, 93
Standard Normal Distribution, 69
Statistic, 106
Statistical Hypothesis, 129
Statistical Inference, 106
I
Regression Line, 91 Statistical Methods, 1
Relative Frequency, 4 Statistics, 1
Relative -Frequency Definition Stratified Random Sampling, 109
of Probability, 26 Subjective Probability, 25
Replication, 144 Sufficiency, 128
By: Rafaqat
Rule of Addition, 29 Systematic Random Sampling,

Rule of Multiplication, 29 109
Runs Test, 183
T
s TabulatiQn, 4
Sample, l 0_5 Test Statistic, 130
Sample Design, 106 Testing of Hypothesis, 127
Sample Space, 24 Time Reversal Test, 171
Sampling, 105 Time Series, 153
Sampling Bias, 111 Total Variation, 93
Sampling Error, 111 Treatment in Experimental
Sampling Frame, 105 Design, 145'
Sampling Unit, I 05 Trial, 23
Scatter Diagram, 91 Two-Sided Test, 131
Seasonal Variations, 153 Type of Statistics, 1
Secondary Data, 2 Type-I Error, 130
Secular Trend, 153 Type-II Error, 130
Sign Test, 182 Types of Index Numbers,_ 168
Significance Level, 130
Simple Aggregate Index, 169
Simple Average of Relatives,
u
169 Unbiasedness, 128
Simple Hypothesis, 129 Unexplained.Variation, 93
Simple- Random Sampling, 109 Uniform Distribution, 63
Subjett Index 206
v Weighted Index Numbers, 170

Weighted Mean, 7
Value Index, 168 Wilcoxon Signed Ranks Test,
Variance, 9 182
w z
Weighted Aggregate Price Index Z-Score (Standard Score), 9
Numbers, 170
Weighted Average of Relative
Price Index Number, 170
By: Rafaqat
Subject Index 207

A Quick Approach To Statistics by G.R.pashA

Uploaded by

Copyright:

Available Formats

A Quick Approach To Statistics by G.R.pashA

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Quick Approach To Statistics by G.R.pashA

Uploaded by

Copyright:

Available Formats

By: Rafaqat

Chapter 1: Introductory Statistics

Chapter 6: Regression & Correlation 91

Chapter 7: Sampling 105

Chapter 8: Statistical Inference 127

Chapter 9: Design and Analysis of Experiment 143

Chapter 10: Analysis of Time Series 153

Chapter 11: Index Numben 167

Chapter 12: Nonpatametric Statistics 181

Answers to Exercises 195

Subject Index 203

Statistics is the collection of methods- for planning experiments, obtaining

Types of Data (Nature) .

Chapter I: Introductory Statistics 2

Chapter I: Introductory Statistics J

Chapter I: Introductory Statistics 4

Class Mark (Midpoints)

Class Interval (Width)

In the table below, class interval (width)= 196-191=201-196=5.

Chapter 1: Introductory Statistics s .

Cumulative Frequency Polygon (Ogive)

\"-a.,.. l: Introductory Statisdcs

Measure of Central Tendency

A measure of central tendency is a value at the center or middle of a data

Empirical Relation among Mean, Median, and Mode

Chapter I: Introductory Statistics 8

The absolute deviation of an element of a data set is the absolute difference

Coefficient of Variation (CV)

Z-Score (Standard Score)

Chapter I: Introductory Statistics

Mean 2'.: Median ;:: Mode; Positively Skewed

Chapter 1: Introductory Statistics 10

Q.2 What is meant by interpretati{Jn of data?

Q.3 Can accurate data be misused? Explain. .

comparison of the annual number of automobile. accidents in the

Q.5 Why is the presentation of the data necessary?

Chapter I: Introductory Statistics 11

Q.8 Why do we construct a histogram?

Q.9 Discuss the advantages of the polygon over the histogram.

Q.10 In what ways 11fretplmcy polygon is different from a histogram?

Q.12 Give chief characteristics of arithmetic mean, media11, mode,

Chapter 1: Introductory Statistics 12

The arithmetic mean can be computed even if all that is

group, is not changed in value by changes in the ot}J.er

If any of the values of the variable is zero, the harmonic

Q.14 Why cannot the arithmetic mean be computed from an open-end

Q.15 . Why ah average computed from a frequency diStribution is not

Q.18 In order to use the formula for the mode:

Mode= L+ (f. - fo) xh

Chapter 1: Introductory Statistics 15

Q.11 In 1003, the mean (arithmetic) monthly income of the families in

Rs. 10,900 cannot indicate a market size ai all because nothing is

Q.13 Why do we need measure of dispersion even when we have

Q.14 For small sample size of 5 or less, which measure of dispersion is

Q.25 Why is the measurement of the . dispersion computed for a

Chapter/: Introductory Statistics 16

Ans. Due to same Grouping Error as discussed in answering for above

Q.27 Which measure of the dispersion, is most widely used?