Survey Design & Analysis

Expert Statisticians

Several Westat staff are faculty at the Joint Program in Survey Methodology consortium (external site), a consortium of the University of Maryland, the University of Michigan, and Westat. JPSM is the nation's oldest and largest program offering graduate training in the principles and practices of survey research.

Westat statisticians are also

  • Fellows of the American Statistical Association
  • Members of the International Statistical Institute
  • Serving as editors and referees for statistical journals
  • Publishing papers, teaching courses, and making presentations at conferences

Statistical Sample Design

Our statisticians work closely with project teams in applying customized, optimally efficient sampling, weighting, and estimation procedures to diverse research projects. Westat has played a key role in refining methods and enhancing the effectiveness of

  • Area probability sampling
  • Weighting
  • Data imputation
  • Automated variance estimation
  • Production of small-area estimates

Westat has a number of national master samples, optimally designed for major national surveys. Our statisticians also use many other databases to select the most efficient samples for surveys of special populations.

Survey Instrument Design

Westat's instrument design, evaluation, and analysis professionals apply qualitative research methods to designing and revising surveys. We have experience with all survey modes, including self-administered (e.g., paper, ACASI, CAPI), interviewer administered, Web, and interactive voice response (IVR). Methods include expert review, cognitive interviews, usability interviews, forms review, and focus groups.

Our staff have backgrounds in sociology, anthropology, and various psychology disciplines. Our expertise includes

  • Instrument evaluation and design/redesign
  • Pretesting and usability evaluation
  • Questionnaire translation

Data Analysis

Westat is thoroughly versed in the analysis of data for many different subject areas from many types of research designs. We use

  • Mathematical models to predict the answers to "what if" questions
  • Econometric methods to simulate or forecast changes in program outcomes
  • Biostatistical methods to assess the effectiveness and safety of experimental therapies

We also use multivariate inferential techniques, including

  • Linear, logistic, and random effects regression modeling
  • Hierarchical linear modeling
  • Generalized estimating equations
  • Factor analyses
  • Multidimensional scaling
  • Cluster analyses

Statistical Software

Westat uses and supports several statistical software applications.