Statistical Methods

Analysis and Modeling

Gaining knowledge and insights are the goals of analysis and modeling. Westat statisticians are well-versed in a variety of analytic and modeling techniques, ranging from simple linear regression to complex multivariate linear and log-linear models, hierarchical Bayesian models, to propensity and predictive models, casual inference, and multilevel modeling and small area estimation. These techniques are often used in conjunction with statistical data integration or simulation methods. Modeling techniques can be applied to a multitude of data types, such as probability or non-probability based samples, administrative records, observational or design-based data. Westat works in collaboration with clients to support transparency and reproducibility of analytic results, providing documentation and annotated programs of the applied methodology.

Data Harmonization and Data Linkage

Bringing together data from disparate sources can help answer complex questions or solve complicated problems. Data harmonization is frequently needed to unify disparate data fields, formats, or dimensions. Using statistical methods for data harmonization, such as standardization and imputation, allows greater data utility. Data linkage methods are used to connect records or units across the different sources. Westat brings a depth of knowledge and experience in these areas to support client analyses, research, and evidenced-based decisions.

Data Privacy

Westat is a highly respected and trusted industry leader in statistical confidentiality and privacy protection. Our statisticians apply a variety of rigorous methods to quantify the risk of disclosure (using statistical models and record linkage), develop data treatments for risk mitigation (including consideration for a formal privacy framework), and to provide creative data access solutions (such as a flexible table generator). To ensure confidentiality and privacy protection of respondent data, Westat works collaboratively with our clients to balance the risk of disclosure with the usefulness of the final data products.

Survey Statistics

Since our inception in 1963, Westat has been world-renowned for expertise in survey statistics. We bring innovative statistical approaches throughout the entire process, from frame creation, to complex sample designs and sample selection, data collection and extraction, data preparation and cleaning, imputation, weighting, nonresponse bias analyses, estimation and variance estimation, to data access and dissemination. Westat statisticians are adept at applying these methods to traditional populations of interest, such as people, households, or businesses, as well as novel populations, such as a corpus of publications, electronic health records (EHRs), or road segments. The purpose is to design high-quality, cost-effective, representative data products for research, analyses, and decisionmaking.