We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Naijia Liu (Princeton University)
Essays on Model Selection and Honest Inference
Selection committeee: Xun Pang (Tsinghua University), Dean Knox (Princeton, recused) and Yiqing Xu (University of California, San Diego)
Citation:
On behalf of this year's John T. Williams Dissertation Prize Committee (Dean Knox (recused), Yiqing Xu, and Xun Pang), I am pleased to announce that the 2019 John T. Williams Dissertation Prize was awarded to Naijia Liu (Princeton University)'s dissertation proposal "Essays on Model Selection and Honest Inference." Naijia’s work takes on two important issues in data analysis for the social sciences. In the first two essays, she integrates machine learning and a latent utility model to correct the over-fitting problem in the current practice of missing data imputation and to encompass missing not at random (MNAR) problems. In the third essay, Naijia works on the challenging problem of model selection in text analysis. She proposes an original method to estimate the number of topics by approximating the marginal likelihood of Latent Dirichlet Allocation topic model. The dissertation proposal provides excellent applications of the techniques to problems of interest, clearly demonstrating that the proposed methods are promising with broad applications in political science.