Sonja Winter

Sonja Winter is an Assistant Professor in the Statistics, Measurement, and Evaluation in Education program at the College of Education and Human Development, University of Missouri. She holds a Ph.D. in Quantitative Methods, Measurement, and Statistics from the University of California-Merced and completed a postdoctoral fellowship at the Missouri Prevention Science Institute, University of Missouri.

Her research interests focus on advanced quantitative methods, particularly Bayesian statistics, structural equation modeling (SEM), psychometrics, and longitudinal data analysis. Her work emphasizes the thoughtful application of priors in Bayesian inference, prior sensitivity analysis, and improving model selection practices in SEM. She leads the Winter Lab, which investigates Bayesian methods, measurement, and latent variable modeling.

Her recent research, including a 2023 publication on under-fitting and over-fitting in Bayesian SEM, critically examines the performance of fit indices and highlights their limitations in detecting overfitting, urging researchers to adopt more robust model evaluation practices. She also promotes the use of prior predictive checks and posterior predictive checks for model validation.

Sonja actively contributes to methodological education through teaching graduate measurement courses and public engagement, such as her appearance on the Learning Bayesian Statistics podcast, where she discussed the intuitive alignment of Bayesian methods with research questions, the challenges of small samples and complex data structures in educational research, and the elegant handling of missing data in longitudinal studies through Bayesian imputation.

She collaborates with colleagues like Dr. Brian Keller on methodological tools such as BLIMP (Bayesian Latent Interaction Modeling Program) for advanced modeling and missing data handling. Her work aims to bridge complex statistical methodology with practical application in educational and developmental psychology research.