Doria Xiao is a Ph.D. candidate in Social Research Methodologies at the University of California, Berkeley, with an anticipated graduation in Spring 2025. Her research focuses on psychometrics and quantitative methods in education, with particular emphasis on Bayesian statistical modeling, latent variable modeling, longitudinal and hierarchical modeling, as well as integrating machine learning into educational assessments. A significant part of her work addresses identifiability challenges in Bayesian Growth Mixture Models (GMMs) and improving model evaluation.
Doria's research aims to understand student growth trajectories by identifying distinct subpopulations characterized by varying developmental patterns. Working under the guidance of Dr. Sophia Rabe-Hesketh, she has contributed to developing diagnostic tools to detect convergence issues in MCMC estimations and has offered recommendations for mitigating issues. She is also exploring model selection and evaluation approaches in Bayesian and frequentist contexts for latent variable and hierarchical models.
In her work at the Berkeley Evaluation and Assessment Research (BEAR) Center, Doria collaborates with interdisciplinary teams on educational measurement. She has also worked on leveraging machine learning to enhance the efficiency of scoring open-ended responses and improve the accuracy of educational assessments.
Doria has served as a graduate student instructor for courses such as Hierarchical and Longitudinal Modeling and Data Analysis in Educational Research II. She is committed to fostering an inclusive and supportive learning environment for students, integrating active learning strategies to help bridge the gap between theory and practical application.
Specializations and Interests
Bayesian Statistical Modeling; Latent Variable and Longitudinal Modeling; Psychometrics and Educational Assessment; Machine Learning Integration in Education; Equity and Fairness in Educational Measurement
