Missing values for potentially observable variables are ubiquitous in psychometric and related social science scenarios for a variety of reasons, including those that may or may not be planned, anticipated, or controlled by the assessor. This chapter covers Bayesian perspectives on missing data as they play out in psychometrics through the lens of Rubin’s (1976) missing-data framework. The key insight is this: when there is a “hole in the data,” a Bayesian perspective tells us to build a predictive distribution for what we might have seen, given all the data we did see, our model for the data had it all been observed, and a model that expresses our belief about the mechanisms that caused part of the data to be missing ...
Get Bayesian Psychometric Modeling now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
