Esther Ulitzsch, University of Oslo
Confirmatory mixture models for investigating contextual correlates of response behavior in ecological momentary assessments
Location: RB 101 Auditorium (Rajska Building)
In ecological momentary assessment (EMA) studies, respondents report on their current behaviors and experiences on several occasions throughout the day, for multiple days. This in-depth data collection offers a unique window into human behavior and experiences. The validity of conclusions drawn from EMA data, however, rests on the assumption that respondents interact with the administered measures in the same way on all measurement occasions. In this talk, I focus on inattentive responding as a prominent example of deviating interactions. When inattentive, respondents provide their responses without investing effort into carefully evaluating the items. As a result, EMA data may be contaminated with responses that do not reflect what researchers want to measure. I illustrate how translating subject-matter theory on respondent behavior into confirmatory mixture models facilitates studying the occurrence of deviating interactions as well as their contextual correlates (e.g., time of day), discussing both item responses and process data as data sources that can be used for model formulation. I end with a brief note on other types of heterogeneous respondent interactions that could be captured and investigated with appropriately formulated confirmatory mixture models.
About the Speaker
Esther Ulitzsch is an Associate Professor at the Center for Educational Measurement (CEMO) at the University of Oslo, Norway. She holds a PhD from Freie Universität Berlin, Germany. Before joining CEMO, she served as a Research Associate at the Department for Educational Measurement and Data Science at the IPN – Leibniz Institute for Science and Mathematics Education in Kiel, Germany. Her research focuses on psychometric model development (e.g., for identifying aberrant response behavior), exploratory tools for unstructured process data, and (Bayesian) small-sample latent variable modeling. Her dissertation on the use of response times for modeling missing values was awarded the NCME Brenda Lloyd Dissertation Award and the Gustav A. Lienert Doctoral Thesis Award from the Expert Group of Methods and Evaluation of the German Psychological Society.