We conduct research on statistical design, modeling, and analysis methods for the Census Bureau's data collection, analysis, and dissemination programs.

Data obtained by the Census Bureau report on people's behavior and condition: Who they are. How they live. Where they live. What they do. What they produce. Once data are collected, we need to determine: How do we use data to produce official statistical products? How do we assess and quantify the uncertainty and quality of official statistical products? Statistical methods help us obtain and do things with data to provide answers to these and related questions.

Statistical methods are procedures that help with:

1. the statistical design of an investigation (e.g., census, sample survey, administrative records, model building);
2. the summarization of the findings from the investigation; and 
3. the inferences that can be drawn from sample data about an entire general population. 

In addition to quantification of uncertainty, statistical methods also help us use “un-designed data” (e.g., commercial data and opportunity data). When existing statistical methods are inadequate for addressing new and challenging data questions, improved or new statistical methods must be found.

Statistical research is the rigorous development of improved or new statistical methods grounded in probability and statistical theory. We focus on methods in eight areas of statistical research and maintain expertise in each.

Research & Expertise

• Missing Data & Observational Data Modeling Methods
• Record Linkage & Machine Learning Methods
• Small Area Estimation Methods
• Spatial Analysis & Modeling Methods
• Sampling Estimation & Survey Inference Methods
• Time Series & Seasonal Adjustment Methods
• Experimentation, Prediction, & Modeling Methods
• Simulation, Data Science, & Visualization Methods

Selected Projects

• Seasonal Adjustment
• The Ranking Project
• Comparisons of A State with Each Other State (Data Visuals)
• Estimated Rankings of All States (Data Visuals)