Dr. Sterba’s research interests span three important areas and integrate theoretical insights with new empirical results. These areas each pertain to multivariate models used extensively in the behavioral sciences. In her study of these models, Dr. Sterba seeks to clarify hypotheses that can appropriately be tested and conditions under which models yield accurate results. She has extended classes of finite mixture and multilevel models to accommodate data structures that are commonly found in the behavioral sciences, but for which existing models were inadequate. Dr. Sterba’s clear writing style makes her work accessible not only to methodologists but also to substantive researchers, and she often provides user-friendly computer code or programs to carry out analyses. As such, her research has far-reaching implications for substantive applications in the behavioral sciences.
Her first area of research concerns finite mixture models, which are commonly used to capture population heterogeneity. Her papers have greatly advanced social scientists’ understanding of the interpretation and performance of mixture models. In particular, Dr. Sterba’s research has clarified how mixture models are related to each other and to multilevel models, how mixture models can be adapted to accommodate different kinds of missing data, how they can recover interaction effects, and how they can be used in assessing explained variance.
A second focus of Dr. Sterba’s research is the widespread use of item parceling, or combining small groups of items to produce more continuously and normally distributed variables. Dr. Sterba had the insight that the selection of which items to include in which parcels introduced a source of additional variability in model results that was not well understood. In a series of papers, Dr. Sterba demonstrated that different allocations of items to parcels could influence parameter estimates, model fit, and the ranking of models. She then developed methods and software tools to ameliorate this problem. Her work upended conventional thinking regarding the impact of different assignments of items to parcels, showing that even under ideal conditions this aspect of parceling can be highly consequential.
Her third major focus is on analyzing partially nested data, which frequently arises in clinical and educational settings when some individuals are nested within groups (e.g., group therapy treatment) and others are not (e.g., wait-list control). Ignoring partial nesting leads to biased standard errors for treatment effects. She introduced structural equation models and extended multilevel models to accommodate partially nested data for a variety of designs. Her work on this topic greatly increased modeling options for partially nested data.
Dr. Sterba is an associate professor and director of the Quantitative Methods Program at Vanderbilt University and she received her PhD from the University of North Carolina at Chapel Hill. She is the recipient of the 2018 Anne Anastasi Distinguished Early Career Contribution Award (APA), 2015 Rising Star Award (APS), and 2015 Cattell Early Career Research Award (SMEP). Additionally, she has taken on important service roles that enhance communication of quantitative modeling advances to the scientific community and public, including serving as Program Chair of the American Psychological Association, Division 5.
