Skip to Main Content
The National Academies of Sciences, Engineering and Medicine
Committee on National Statistics
Division of Behavioral and Social Sciences and Education
Statistical Methods and Estimates for Policy Use

Over several decades CNSTAT has undertaken a range of studies to improve statistical methods and estimates for use in policy making and implementation. A landmark 1991 study evaluated microsimulation models that are used extensively by executive branch and congressional agencies for assessing the pros and cons of alternative tax and transfer programs; its recommendations led to better understanding, documentation, and hardware and software platforms for these important public policy tools. Subsequent studies examined models for retirement income and Medicare policies. A major study in the late 1990s evaluated small area estimation models that combine administrative records with survey data to produce more reliable estimates than are possible with survey data alone: the estimates produced by one such model, the Census Bureau’s Small Area Income and Poverty Estimates (SAIPE) program, are used to allocate federal funds to school districts to support disadvantaged students. In the late 2000s, CNSTAT completed a major study for the Food and Drug Administration on methods for reducing and compensating for missing data in clinical trials of alternative treatments for medical conditions. Other policy-relevant methodological studies have evaluated the accuracy of the polygraph for lie detection, alternative methods for estimating the size of markets for illegal drugs, and measures of disparity for women-owned small businesses in federal contracting.

Using Multiple Data Sources to Improve Federal Statistics

CNSTAT is currently engaged in studies for individual statistical agencies and for the federal statistical system more generally to improve federal statistics for policy use through combining multiple sources of data. The goal is to explore how to foster a paradigm shift in federal statistical programs—from the current paradigm of providing users with the output from a single census, survey, or administrative records source. The new paradigm would include combinations of diverse data sources from government and the private sector. These data would be combined with state-of-the art methods to give users richer and more reliable statistics leading to new insights about policy and socioeconomic behavior. In order to accomplish this, a wide variety of issues will need to be addressed, including access to alternative sources of data for federal statistical programs; designs of statistical processes that foster privacy protection, transparency, objectivity, timeliness, replicability, efficiency, and continuity of statistical series; concepts, metrics, and methods for assessing the quality and utility of alternative data sources for federal statistics; statistical models for combining data from multiple sources; and metrics and methods for evaluating the quality of combined-information estimates. Through these studies we hope to identify “proof of concept” examples as well as priorities for needed research and development.

Completed Projects and Workshops

Using Multiple Data Sources and State-of-the-Art Estimation Methods in Federal Statistics: Frameworks, Methods, and Assessment

Methods for Integrating Multiple Data Sources to Improve Crop Estimates

Research Methodologies and Statistical Approaches to Understanding Driver Fatigue Factors in Motor Carrier Safety and Driver Health

Options for Estimating Illegal Entries at U.S.-Mexico Border

Strengthening Data Science Methods for Department of Defense Personnel and Readiness Missions


Relevant Reports


CNSTAT Home About CNSTAT CNSTAT Members CNSTAT Staff CNSTAT Mission Other Sites of Interest News and Events Upcoming Events CNSTAT News Public Seminars and Symposia Our Work Approaches to Estimating the Prevalence of Human Trafficking in the US Challenges and New Approaches for Protecting Privacy in Federal Statistical Programs Contingent Work and Alternate Work Arrangements Developing Indicators of Educational Equity Transparency and Reproducibility of Federal Statistics for the National Center for Science and Engineering Statistics Using Models to Estimate Hog Production Workshop on Improving the American Community Survey Workshop on Social Science Modeling for Big Data in the World of Machine Learning Principles and Practices for a Federal Statistical Agency People, Projects, and Publications 2018-2019 Presentations from Meetings and Workshops Publications CNSTAT Reports Principles and Practices (P&P) Work With Us Our Study Process Study Process Brochure Getting to Know the Committee Process Overview for Sponsors Guide for Study Sponsors