Seminar at SMU Delhi

Logistic regression is a commonly used method for handling binary responses. In this talk, we study a situation when a portion of the responses have been misclassified. Pseudo-likelihood is used to estimate the regression parameter and we study the asymptotic properties of the estimator. We also show that the bootstrap is consistent in this set up. A small numerical study is presented to illustrate the theoretical results. If time permits, we will also present a brief overview of empirical process tools that were used to solve this problem.