Annals of the International Communication Association

ISSN: 2380-8985 (Print) 2380-8977 (Online) Journal homepage: [Link]

The Influence of Speech Communication on the

Image of a Political Candidate: “Limited Effects”
Revisited

Keith R. Sanders & Thomas J. Pace

To cite this article: Keith R. Sanders & Thomas J. Pace (1977) The Influence of Speech
Communication on the Image of a Political Candidate: “Limited Effects” Revisited, Annals of the
International Communication Association, 1:1, 465-474, DOI: 10.1080/23808985.1977.11923700

To link to this article: [Link]

Published online: 02 Nov 2017.

Submit your article to this journal

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

[Link]
THE INFLUENCE OF SPEECH COMMUNICATION ON THE
IMAGE OF A POLITICAL CANDIDATE: "LIMITED
EFFECTS" REVISITED 1

KEITH R. SANDERS and THOMAS J. PACE

Southern Illinois University

Drawing on the selective exposure hypothesis and related lit~ature, this paper presents
the results of a "natural experiment" designed to test hypotheses regarding the demo-
graphic composition and partisan affiliations of the audiences attracted to political
campaign speeches. The immediate effects on candidate image of the speeches were also
measured. Seventeen audiences of a Republican gubernatorial candidate completed a
nine-scale semantic differential before he spoke and immediately afterward. Data
analysis revealed that, although a majority of!hose in the candidate's audiences were
Republican, partisan composition shifted significantly from one audience situation to
another. In addition, the candidate's speeches produced significant positive shifts in
image across all audience situations. Finding the "limited effects" perspective uninspir-
ing as an explanation for these outcomes, the authors invoke, tentatively, a "uses and
gratifications" rationale.

It is widely assumed that, in the age of television, posure one of the two basic precepts of his highly
political oratory is an anachronism about which heuristic theory of cognitive dissonance. Berelson
little of value remains to be learned. Heavily influ- and Steiner (1964) presented it as a basic principle
enced by a "limited effects" communication of human behavior. The Behavioral Sciences Sub-
model, most scholars who hold an opinion on the panel (1962, p. 57) of the President's Science Advi-
matter appear to believe that they know, at least in sory Committee asserted flatly, " ... individuals
general terms, who comes to hear the speeches of a engage in selective exposure ... ", and Richard
political candidate and what effect, if any, these Hofstadter ( 1966) included selective exposure in his
speeches have (Scammon & Wattenberg, 1968; description of the "paranoid style" in American
Lamb & Smith, 1968; Swanson, 1972). The imme- politics.
diate audiences of political candidates are thought Indeed, such conclusions as the foregoing
to be relatively homogeneous, highly partisan, and seemed, at the time, irrefutable, when one consid-
rather resistant to change. It is reasoned that those ered the dearth of studies which supported the no-
who attend political meetings are so firmly commit- tion of selectivity. For example, after their classic
ted to their opinions and to their candidate, that study of the 1940 presidential election campaigns,
there is little room for further conversion and little Lazarsfeld, Berelson, and Gaudet (1948) con-
likelihood of a reduction in their commitment. At cluded that "voters turn to propaganda which af-
best, it is thought that political rallies serve to rein- firms that validity and wisdom of their original
force pre-existing attitudes (MacNeil, 1968; Kei- decision" (p. 90). In studying exposure patterns
Jey, 1960). resulting from an information campaign favorable
Those who hold this point of view could, until to the United Nations, Star and Hughes (1950)
recently, find a great deal of support for their posi- found exposure was considerably higher among
tion in the literature of selective exposure. Through persons who were already favorable to the United
the mid-1960s no psychological principle was more Nations. Schramm and Carter ( 1959) found that a
widely accepted than the assertion that individuals larger percentage of Republicans than Democrats
are motivated to seek information which supports had watched a telethon given by the Republican
their opinions and to avoid information which con- candidate. It was evidence of this sort, and more,
tradicts them. Fe stinger ( 1957) made selective ex- which led to universal acceptance of the doctrine of
466 Communication Yearbook I

selectivity and the implication that political mes- because it contains much useful information about
sages, especially during campaigns, produce little day-to-day trends in the business and industrial
attitude change (Blumler & McLeod, 1974; Sanders community. The position one takes when explain-
in press). ing selectivity, de facto or motivated, dictates, to a
However, Freedman and Sears (1975) published large extent, the inference he will draw about the
a review of the experimental studies which had probability that partisan political messages will
tested the selective exposure hypothesis and con- change the attitudes of those who attend partisan
cluded that "experimental evidence does not dem- political rallies.
onstrate that there is a general psychological ten- If one is a "limited effects" theorist, holding the
dency to avoid nonsupportive and to seek out sup- view that political messages in campaigns exert
portive information" (1965, p. 94). Not convinced few influence partly because of selective exposure,
by the Freedman and Sears analysis, Mills ( 1968) then one must hold that political oratory will have
reviewed the same set of studies, plus some which few if any additional effects on its immediate audi-
were done after the Freedman-Sears review, and ence. Unlike the more or less passive selectivity
decided the evidence warrants the conclusion that involved in deciding to watch one television pro-
people tend to seek out supporting information and gram over another, the taking of oneself to a politi-
avoid discrepant information. Katz ( 196 ~- has also cal rally requires effort. In fact, it is difficult to
called for a less than total rejection of the imagine a more overt act of selective exposure.
hypothesis, based upon field research in mass Sears and Whitney ( 1973, p. 259) contend that:
communications. Sanders and Newman ( 1971) "Face-to-face political meetings may elicit more de
found in a limited test of Hofstadter' s description of facto selectivity than communications in newspa-
the paranoid political style that political pers, magazines, TV and radio because they attract
"paranoids" do not actively avoid discrepant in- fewer people and hence probably only the most
formation. In fact, they found some evidence that involved partisans." If Sears and Whitney are cor-
such persons actively seek discrepant information, rect, political meetings constitute a propitious set-
distort it, and use it as data in support of their world ting in which to test the hypothesis that partisan
view. messages have little impact on partisan audiences.
Unfortunately, few empirical studies have been
conducted in this setting.
Motivated Versus De Facto Selectivity
Although speech-communication. scholars have
The major issue in the controversy has to do with given a great deal of attention to political public
the "voluntary" or "motivated" nature of selectiv- speaking, it has, as Swanson (1972) rightly con-
ity and not with the selective exposure per se. All tended, come almost exclusively from those who
disputants agree that de facto selectivity has been hold a noncmpirical orientation. In this literature
well documented. Persons are exposed more to there is much speculation regarding the composi-
communication with which they agree, than to tion of audiences and the impact of speeches on
communication with which they disagree. Howev- immediate audiences. In fact, a review (Trent,
er, this does not demonstrate the presence of a 1975) of the most pertinent literature reveals only
general psychological predisposition in favor of two field studies on the direct effects of political
supportive information. Selectivity could be due to oratory.
a variety of other factors, such as the availability of Brooks ( l 967) studied the influence on attitudes
more supportive information in the individual's toward the candidates of speeches given at political
natural environment (Sears, 1968), the greater per- rallies in Pittsburgh during the Goldwater-Johnson
ceived utility of some information over other infor- campaign. He found that Goldwater was successful
mation, or a combination of reasons. A conserva- in changing audience attitudes toward himself on
tive Republican may read the Wall Street Journal two of the five scales used in the study. Johnson,
because it includes supportive information, is in the however, failed to produce a significant change on
offices and homes of all of his colleagues, and/or any of the five scales.
POLITICAL COMMUNICATION 467

Kaid and Hirsch ( 1973) found that a 1972 cam- groups. and political rallies. At least two, but not
paign speech by Edmund Muskie produced a positive more than three, audiences were chosen from each
shift in mean image score as measured by twelve of the last six weeks of the campaign during which
semantic differential scales, but no change in the the largest number of speeches were given. Audi-
overall image structure as indicated by factor analy- ences were selected from a variety of geographical
sis of pretest and posttest scores. A follow-up study areas within the state.
conducted two to three weeks after the Muskie rally The selection of subjects within the 17 audiences
indicated that, while the changes in mean image was less complicated. In audiences of thirty or
scores held, image structure had changed. The re- fewer, largely civic and fraternal organizations, the
searchers concluded that'' A single appearance by a entire audience served as subjects. In larger groups,
political candidate can result in a favorable shift in such as political rallies, subjects were selected as
image, and that shift can persist over time." they entered the meeting by randomly handing
Both the Brooks ( 1967) and Kaid and Hirsch them a questionnaire or by moving through an al-
(1973) studies call in to question inferences drawn ready assembled audience and passing out ques-
from the much embattled "limited effects" model tionnaires. Subjects were not chosen from any par-
(Becker, McCombs, & McLeod, 1975) that politi- ticular place in the meeting room or from those
cal partisans are the exclusive audience of political ~ rriving early or late. Questionnaires were in the
oratory and that such oratory has little impact. As a hands of subjects before the candidate was intro-
further test of the implications of this model, this duced.
study was designed to test the following hypoth- A total of763 questionnaires were distributed. Of
eses: this number, 553 (or about 80%) were retrieved.
Complete and usable data were obtained from 491,
H 1: At least a majority (51%) of those who come to
or 64%, of the total: 127 from civic and fraternal
hear a political candidate speak will be mem-
groups, 189 from university related groups, and 175
bers of his political party.
from political rallies.
H2 : Sex, age, and political party affiliation will not
vary significantly from one audience situation Design
to another during a political campaign.
H3 : The speeches of a political candidate will have
The pretest-posttest design pattern was used
no significant influence on his image as per-
across all audiences. All data were gathered by the
ceived by his immediate audiences.
researchers. Before the candidate began speaking,
subjects were given a three page questionnaire call-
PROCEDURES ing for information regarding their sex, age, politi-
cal party affiliation, and presidential preference. It
Subjects also asked whether they had seen the candidate on
television or on billboards; whether they had heard
The subjects from whom data were gathered for him on radio, or read about him in newspapers; and
this study were selected from 17 different audiences whether they had heard him speak in person earlier
who came to hear a political speech by Paul Eggers in the campaign. The individual introducing the
during the final six weeks of his 1968 campaign as candidate was asked to tell the audience that a study
Republican candidate for governor of Texas. The was being conducted and to call for their coopera-
17 audiences were chosen in an attempt to include in tion.
the study a representative sample of the immediate The questionnaire also contained two nine-scale
audiences to whom the candidate spoke. semantic differentials and instructions for comple-
In about the proper proportions they represented tion. Subjects were asked to complete one set of
the three different types of audiences to whom the scales before the candidate was introduced and an
candidate spoke during the period under study: civic identical set after the speech and the question and
and fraternal organizations, university-related answer period.
468 Communication Yearbook I

Measuring Instrument or designated their affiliation as "other." A com-

parison of 297 with the hypothesized value of 250
The nine scales used in the semantic differential (51%) produced a z of 3.16 which was significant
were derived especially for this study. Sixty-seven beyond .05. Thus, when one looks at the total
graduate and undergraduate students on the campus sample, there were clearly more representatives of
of Southern Illinois University at Carbondale lis- the candidate's party than of all other parties com-
tened to an audio tape of Paul Eggers giving a bined.
speech and being interviewed. As they listened, However, a somewhat different picture emerges
they were asked to write down adjectives describing when one tests the stability of party affiliation
Eggers as they came to mind. From these adjectives across audience situations. Civic, fraternal, and
35 recurring scales were selected and converted to university audiences were split almost in half be-
bipolar semantic differential form. These 35 scales tween Republicans and non-Republicans while
plus an additional 10 which had been used to assess 81% of those who attended rallies were Republican
the image of participants in the 1960 Kennedy- and 19% non-Republican. Testing the number of
Nixon debates comprised a 45-item semantic dif- Republicans who attended speeches in civic/
ferential which was administered to another group fraternal contexts against those who attended in
of 153 undergraduate and graduate students after university contexts produced a nonsignificant z of
they had viewed a video tape of Eggers giving a .20 (p>.05). However, the same statistic revealed
speech and/or being interviewed. significant differences when the number of Repub-
Factor analysis of the data obtained from this licans at civic/fraternal speaking events and the
experiment produced I 0 scales with high loadings number at university related events were compared
on an evaluative factor, a dynamism factor, and a with the number of Republicans at political rallies.
weakly defined "ideological" factor. These 10 These scores were, respectively: z=6.60, (p<.05),
scales were tested on the campus of the University and z=6.40, (p<.05). A political speaker can,
of Texas in mid-September just prior to the period therefore, expect a highly partisan audience at a
under study. Seventy-one subjects who had heard of rally but not in the other two contexts studied here.
the candidate and were Texans of voting age com- Political party affiliation does vary from one audi-
pleted the semantic differential on the concept ence situation to another.
"Paul Eggers." These data were factor analyzed
and produced a factor structure basically similar to Hypothesis 2
the one obtained in Illinois, suggesting that the
candidate's image was being defined in basically A review of the data in Table I on the sexual
three dimensions of meaning. Thus, eight of the composition of political audiences indicates that in
original I 0 scales were used in the study reported all three audience situations males tended to pre-
here. Two were added to give strength to what dominate. Overall, 67% ofthose present were male
appeared to be a stable, but heretofore weakly de- and 33% were female. There was some variation
fined, "ideological" factor. across audience types. Although comparison of the
number of males in the civic/fraternal setting with
RESULTS the number of males in university setting failed to
produce a significant difference (z=[Link], p>.05)
Hypothesis 1 as did a comparison of males in civic/fraternal audi-
ences with males who attended political rallies
The data summarized in Table I clearly indicate (z= I. 75, p> .05), the number of males in univer-
that hypothesis I is confirmed. More than a major- sity audiences did differ from the number at politi-
ity of those who came to hear the candidate speak cal rallies (z=3.19, p<.05). Thus, the sexual com-
were members of his party. Sixty percent of those position of political audiences can vary across audi-
who came were Republicans while 21% were Dem- ence situations.
ocrats, and 18%were either American Independents The same general pattern holds true for the age
POLITICAL COMMUNICATION 469

TABLE 1
Frequency Distribution and Percentages on Sex, Party Affiliation, and
Age of Subjects in Civic/Fraternal, University, and Political Rally
Audience Situations

Audience Situations
Variable Civic-Frat. Univ. Pol. Rally Total

Sex
Male 87(69%) 140(74%) 103(59%) 330(67%)
Female 40(31%) 49(26%) 72(41%) 161(33%)

Totals 127 189 175 491

Party
Republican 64(50%) 92(49%) 141(81%) 297(60%)
Democrat 35(28%) 62(33%) 5( 3%) 102(21%)
Amer. Ind. 10( 8%) 10( 5%) 7( 4%) 27( 5%)
Other 18(14%) 25(13%) 22(13%) 65(13%)

Totals 127 189 175 491

Age
0 - 20 13(10%) 75(40%) 21(12%) 109(22%)
21 - 30 39(31%) I08(51%) 45(26%) 192( 39%~
31 - 40 44(35%) 5( 3%) 59(34%) 108(22%)
41 - 50 17(13%) 1( 1%) 35(20%) 53( 11%)
51 - 60 10( 8%) 9( 5%) 19( 4%)
61 plus 4( 3%) 6( 3%) 10( 2%)

Totals 127 189 175 491

Mean Age 33.9 22.2 32.3 28.8

variable. As Table I reveals, there was a considera- (t=I56.68, p<.05). Thus, hypothesis 2 must be
ble difference in the range and variation around the rejected. Sex, age, and political party affiliation all
mean from one audience type to another. The t-test tend to vary from one audience type to another.
for differences between means of independent sam-
ples indicates that the mean age of the civic/fraternal Hypothesis 3
audiences differed significantly from the mean age
of the university audiences (t=84.35, p<.05), that Table 2 summarizes the data collected in the
the mean age of civic/fraternal audiences differed testing of hypothesis 3. It indicates overwhelmingly
significantly from the mean age of the political rally that hypothesis 3 must be rejected. The speeches of
audiences (t=215.92, p<.05), and that the mean the candidate under study had a significant influ-
age of university audiences differed significantly ence on his image in all three audience situations.
from the mean age of political rally audiences The scores which served as raw data for the means
470 Communication Yearbook I

TABLE 2
Means, Standard Deviations, and F. Ratio for Pretest and Posttest
Scores on the Semantic Differential

Aud. Sit. N Pretest S.D. Post test S.D. F-Ratiol

Civic/Frat. 127 34.12 6.03 37.41 5.21 75.47

University 189 34.05 5.15 38.86 5.15 132.48

Rally 175 36.82 6.09 39.58 5.06 173.50

Over-All 491 35.05 5.88 37.97 5.09 413.74

1 All F-Ratios were significant beyond the .001 level

TABLE 3
Factor Loadings Yielded by Varimax Analysis of Pretest and Posttest
Data Gathered from Civic/Fraternal Audiences

Pretest
Factor 1 (.58)1 Factor 2 (.42)

Unresponsive-Responsive .74 Talker-Doer .71

Bad-Good .74 Old Ideas-New Ideas .71
Unfair-Fair .69 Weak-Strong .55
Dishonest-Honest .68
Liberal-Conservative .66
Spender-Saver .31

Post test
Factor 1 (.62) Factor 2 (.38)

Unresponsive-Responsive .69 Bad-Good .60

Old Ideas-New Ideas .65 Weak-Strong .55
Unfair-Fair .61 Liberal-Conservative .51
Dishonest-Honest .58 Spender-Saver .37
Talker-Doer .56

1 Percent of common variance

POLITICAL COMMUNICATION 471

TABLE 4
Factor Loadings Yielded by Varimax Analysis of Pretest and Posttest
Data Gathered from University Audiences

Pretest
Factor 1 (.85) 1 Factor 2 (.15)

Bad-Good .79 Spender-Saver .45

Dishonest-Honest .74 Liberal-Conservative .31
Unfair-Fair .70
Unresponsive-Responsive .70
Weak-Strong .66
Old Ideas-New Ideas .59
Talker-Doer .46

Post test
Factor 1 (.86) Factor 2 (.14)

Unfair-Fair .80 Spender-Saver .49

Bad-Good .75 Liberal-Conservative .30
Weak-Strong .72
Unresponsive-Responsive .72
Dishonest-Honest .65
Old Ideas-New Ideas .54
Talker-Doer .52

1 Percent of common variance

reported in Table 2 ranged from a low of 9 to a high ence perceptions of the candidate between the pre-
of 45 for each subject. In other words, the lowest testing and posttesting of the civic/fraternal audi-
score any subject could give was one on each of the ences. For example, posttest factor 2 retained only
nine scales, for a total of 9, and the highest score one of the three scales it held in the pretest, dropped
was 45 or a 5 on each of the nine scales. The means two scales, and added three. However, the results
rep6rted in Table 2 revealed a positive shift in the from an application of program SIMFAC do not
candidate's image. support this conclusion. SIMFAC compares factor
As a supplement to the analysis of variance, we structures and provides a coefficient which is indic-
performed a varimax orthogonally-rotated factor ative of the extent to which the factors are related.
analysis on the pretest and posttest data gathered in Pretest factor I correlates highly with posttest factor
each of the three audience situations. With the 2 (r=.90, R2 =.81), and pretest factor 2 correlates
eigenvalue set at 1.00, the first run yielded a one highly with posttest factor I (4=.96, R2 =.92). It
factor solution on all six sets of data (three pretests can, therefore, be concluded that those who came to
and three posttests). However, it did appear that hear the candidate on civic/fraternal occasions did
there might be a second related factor. Tables 3, 4, not restructure their image of the candidate after
and 5 contain the scales and Ioadings extracted hearing him speak.
during a second analysis wherein a two factor solu- Table 4 illustrates a similar result for those who
tion was requested. heard the candidate in university settings. There is
At first glance, Table 3 might lead one to con- obviously great similarity between pretest factor I
clude that there was a substantial change in audi- and posttest tactor I , and between pretest factor 2
472 Communication Yearbook I

TABLE 5
Factor Loadings Yielded by Varimax Analysis of Pretest and Posttest
Data Gathered from Political Rally Audiences

Pretest
Factor 1 (.68)1 Factor 2 (.32)

Dishonest-Honest .86 Talker-Doer .65

Unfair-Fair .82 Spender-Saver .63
Unresponsive-Responsive .76 Old Ideas-New Ideas .61
Bad-Good .76
Weak-Strong .70
Liberal-Conservative .50

Post test
Factor 1 (.65) Factor 2 (.35)

Bad-Good .82 Talker-Doer .69

Unfair-Fair .80 Old Ideas-New Ideas .65
Weak-Strong .72 Spender-Saver .54
Dishonest-Honest .70
Unresponsive-Responsive .47
Liberal-Conservative .18

1 Percent of common variance

and posttest factor 2. The Ioadings for each factor DISCUSSION

are similar, as is the amount of variance accounted
for by each factor. SIMFAC compared pretest fac- This study, which has elsewhere been called a
tor I with posttest factor 1 and yielded a similarity "natural experiment" (Sanders & Pace, 1969), suf-
coefficient of .99 (R2 =.98). A similar comparison fered from the problem of the nonrespondent, as do
of pretest factor 2 and posttest factor 2 produced a r many nonlaboratory studies. And in this study non-
value of .93 (R2 =.88). Thus, although the analysis respondents were not systematically examined.
of variance results reported in Table 2 indicate that However, a cursory analysis of the incomplete data
there was a significant shift in candidate image, this produced by some subjects and brief interviews
shift was not accompanied by a restructuring of the with a few nonrespondcnts suggest that some did
candidate's image. not understand the instructions and that others were
The data reported in Table 5 on political rallies less friendly to the candidate than were respon-
follows a general pattern almost identical to the one dents. Thus, it is possible that the image pretest is
evident in Table 4. Pretest factor 1 and posttest somewhat inflated and, perhaps, the posttests as
factor I correlate highly {r=.97, R2 =.95), and pre- well. This does not, in our judgment, compromise
test factor 2 correlates highly with posttest factor 2 the worth of the study, because each subject served
{r=.99, R2 =.98). Again, analysis of variance re- as his own control. Whether viewed on an
sults show a shift in image, but this shift does not audience-by-audience, or a scale-by-scale basis, as
include any significant restructuring. we did in a preliminary analysis of variance, or in a
POLITICAL COMMUNICATION 473

more collective audience-situation-by-audience- enced. This suggests, however tentatively, that

situation context, as we have done in this report, the there may be a strong relationship between audience
shift in image is substantial. needs and expectations and the immediate, and,
More importantly, however, this study raises perhaps, long-term effects of political messages,
new questions regarding the adequacies of the" lim- and that this relationship may be much more com-
ited effects'' perspective as an explanation for polit- plicated than the selective exposure hypothesis im-
ical information-seeking behavior and communica- plies.
tion outcomes. The partisan diversity of the audi- At any rate, we do not believe that changes in
ences which came to hear Paul Eggers, as well as candidate image of the magnitude found in this
their demographic heterogeneity, suggest that many study can be explained satisfactorily as mere rein-
persons came for reasons other than to have their forcement of pre-existing attitudes toward the can-
partisan dispositions reinforced. Party affiliation as didate. Even if such changes could be explained in
an indicator of audience composition or audience that fashion, we doubt that such an explanation
response may be losing its strength. There is much would hold the long-term heuristic value of the
evidence from other sources that identification with yet embryonic uses and gratifications rationale
the major parties is decreasing and that the power of which we have offered, ex post facto. There is an
party affiliation as a predictor of voting behavior is indication in this study, and in others, that the
diminishing (Dreyer, 1971-1972). As voters be- insistence by many speech communication theorists
come less interested in party cues and more atten- and rhetorical critics that oral communication be-
tive to the characteristics of individual candidates, havior is inherently an independent variable should
which recent data (De Vries & Tarrance, I 972; At- be abandoned. To paraphrase, members of a politi-
wood & Sanders, 1975) suggest, the reasons for cal audience may do more with a political speech in
attending a political meeting may become more terms of satisfying their needs than the speech may
complicated. do to them. (Katz, Blumler, & Gurevitch, 1974).
In fact, so inadequate is the "limited effects" Political speeches should be viewed as dependent
perspective, one is tempted to abandon it as a variables as well as independent variables. Wide
heuristic stimulus and as an explanatory tool, in ranging, diverse, and complex audience needs may,
spite of its profound influence on the planning and in the long-run, be more predictive of communica-
conduct of this study. In its place, one is inclined tion outcomes than either source or message vari-
to substitute the recently reemergent uses and ables, the traditional concerns of speech communi-
gratifications model (Biumler & McQuail, 1969; cation process and criticism. The time has come to
Blumler & Katz, 1974) which offers several possi- investigate this possibility.
ble explanations for the motivations behind political
information-seeking of the kind studied here. This NOTE
approach would suggest that when members of the I. The authors wish to express their appreciation to Professor
electorate make the effort necessary to attend a Douglas Bock of the University of Florida and to Professor
Joe Steams of Western Kentucky University for their help in
political meeting, they may be exhibiting evidence gathering data; to Paul Eggers, the I 968 Republican Candi-
of a variety of needs, including the need to learn date for governor of Texas, for allowing unrestrained access
more about the issues, the need to be seen in proper to his campaign appearances; and to the Office of Research
and Projects, SIU-C, for its monetary support.
political company, the general need to'' feel better''
about the political process, or the explicit need to
REFERENCES
have their attitudes toward the candidate changed in
a positive direction. There are data from a recent ATWOOD, E., & SANDERS, K. Perception of information
study of voter behavior in Ohio (O'Keefe, I 976) sources and likelihood of split ticket voting. Journalism
indicating that those who were most influenced by Quarterly, I 975, 52, 3, 42 I -428.
BECKER. L.B., McCOMBS, M.E., & McLEOD, J.M. The
mass communication over the course of the cam- development of political cognitions. In S.H. Chaffee (Ed.),
paign were those who most expected to be influ- Political communication. Beverly Hills: California, 1975.
474 Communication Yearbook I

BERELSON, B.R., & STEINER, [Link] behavior. New MacNEIL, R. The people machine. New York: Harper & Row,
York: Harcourt Brace, 1964. 1968.
BLUMLER, J.G., & McQUAIL. D. Television in polirics: lrs MILLS. J. Interest in supporting and discrepant information. In
uses and influence. Chicago, Illinois: University of Chicago Robert P. Abelson, et al., (Eds.), Theories of cognitive
Press, 1969. consisrency. Chicago, Illinois: Rand McNally and Com-
BLUMLER, J.G., & KATZ, E. (Eds.), The uses of mass com- pany, 1968,771-776.
munications. Beverly Hills, California: Sage, 1974. O'KEEFE, G. The uses and gratifications approach and political
BLUMLER, J.G., & McLEOD, J.M. Communication and voter communication research. Political Communication Review,
turnout in Britain. In T. Leggart (Ed.), Sociological theory 1976, I, 3, 8-11.
and survey research. Beverly Hills, California: Sage, 1974. Report of behavioral sciences subpanel. President's Science
BROOKS, W.D. A field study of the Johnson and Goldwater Advisory Committee. Washington, D.C.: Government
campaign speeches in Pittsburgh. The Solllhem Speech Printing Office, 1962.
Journal, 1967, 32, 273-281. SANDERS. K.R. A critique of contemporary approaches to the
DeVRIES, W., & TARRANCE, V.L. The ticket-spliuer. Grand study of political communication. In R. Davis (Ed.}, Pro-
Rapids, Michigan: Eardmans, 1972. cadings of the spuch communication association summer
DREYER, E.C. Media use and electoral choices: Some political conference, 1975. Falls Church, Virginia: Speech Com-
consequences of information exposure. Public Opinion munication Association (in press).
Quarterly, 1971-72,35, 3, 545-553. SANDERS, K.R., & NEW MAN, R.P. John A. Stormerand the
FESTINGER, L. A theory of cognith·e dissonance. Stanford, Hofstadter hypothesis. Central Srates Speech Journal, 22,
California: Stanford University Press, 1957. 1971, 218-227.
FREEDMAN, J.L., & SEARS, D.O. Selective exposure. In SANDERS, K.R .. & PACE, T.J. The natural experiment as a
Leonard Berkowitz (Ed.), Advances in experimental social research methodology in speech communication. In James
psychology, Vol. 2. New York: Academic Press. 1965. E. Roever (Ed.}, Proceedings of the Speech Association of
HOFSTADTER, R. The paranoid style in American politics, America summer conference on research and action. New
and other essays. New York: Knopf, 1966. York: Speech Association of America, 1969, 80-87.
KAID, L.L., & HIRSCH, R.O., Selective exposure and candi- SCAMMON, R.M., & WATTENBERG, B.J. The real major-
date image: A field study over time. Central States Speech itv. New York: Coward-McCann, 1970.
Journal, 1973, 24, 48-51. SCHRAMM, W., & CARTER, F. Effectiveness of a political
telethon. Public Opinion Quarterly, 1959, 23, 188-190.
KATZ, E. On reopening the question of selectivity in exposure SEARS. D.O. The paradox of de facto selective exposure with-
to mass communications. In Robert P. Abclson, et al.. out preferences for supportive information. In Robert P.
(Eds.). Theories of cognitive consistency. Chicago, Illinois: Abelson, et al., (Eds.), Theories of cognitive consisrency.
Rand McNally, 1968, 788-796. Chicago, Illinois: Rand McNally, 1968.
KATZ, E., BLUMLER, J.G .. & GUREVITCH. M. Utilization SEARS, D.O., & WHITNEY, R.E. Political persuasion. In I. de
of mass communication by the individual. In J.G. Blumler Sola Pool, et al., (Eds.), Handbook of communication.
and E. Katz (Eds.), The uses of mass communicarion. Bev- Chicago, Illinois: Rand McNally, 1973.
erly Hills, California, Sage, 1974. STAR, S.A., & HUGHES, H.M. Report of an educational
campaign: The Cincinnati plan for the United Nations. Amer-
KELLEY, S., JR. Political campaigning. Washington, D.C.: ican Journal of Sociology, 1950, 55, 389-400.
The Brookings Institute, 1960. SWANSON, D.L. The new politics meets the old rhetoric: New
LAMB, K.A., & SMITH, P.A. Campaign decision-making: directions in campaign communications research. Quarterly
The presidential election of 1964. Belmont, California: Journal of Speech, 1972, 53, 31-40.
Wadsworth, 1968. TRENT, J.S. A synthesis of methodologies used in studying
LASARSFELD, P.F .. BERELSON, B., & GAUDET. H. The political communication. Central States Speech Journal,
people's choice. New York: Duell, Sloan and Pcarce, 1948. 1975, 26, 287-297.