Convergence Failures in Logistic Regression

A frequent problem in estimating logistic regression models is a failure of the likelihood maximization algorithm to converge. In most cases, this failure is a consequence of data patterns known as complete or quasi-complete separation. For these patterns, the maximum likelihood estimates simply do not exist. In this paper, I examine how and why complete or quasi-complete separation occur, and the effects they produce in output from SAS ® procedures. I then describe and evaluate several possible solutions.