Modeling Item Response Theory with Stochastic Variational Inference
This work addresses scalability issues in psychometric modeling for researchers and practitioners, enabling more flexible and efficient analysis of large-scale human response data, though it is incremental in applying existing variational inference techniques to IRT.
The authors tackled the computational bottleneck in fitting Item Response Theory (IRT) models to large datasets by introducing a variational Bayesian inference algorithm, which achieved higher log likelihoods and accuracy in imputing missing data across five datasets compared to alternatives.
Item Response Theory (IRT) is a ubiquitous model for understanding human behaviors and attitudes based on their responses to questions. Large modern datasets offer opportunities to capture more nuances in human behavior, potentially improving psychometric modeling leading to improved scientific understanding and public policy. However, while larger datasets allow for more flexible approaches, many contemporary algorithms for fitting IRT models may also have massive computational demands that forbid real-world application. To address this bottleneck, we introduce a variational Bayesian inference algorithm for IRT, and show that it is fast and scalable without sacrificing accuracy. Applying this method to five large-scale item response datasets from cognitive science and education yields higher log likelihoods and higher accuracy in imputing missing data than alternative inference algorithms. Using this new inference approach we then generalize IRT with expressive Bayesian models of responses, leveraging recent advances in deep learning to capture nonlinear item characteristic curves (ICC) with neural networks. Using an eigth-grade mathematics test from TIMSS, we show our nonlinear IRT models can capture interesting asymmetric ICCs. The algorithm implementation is open-source, and easily usable.