Regression with latent variables

This talk will provide a brief overview of some of some techniques, highlighting the advantages and disadvantages of each, with particular reference to the data types usually encountered in the social sciences. The overview will touch on naïve methods based on the use of latent variable scores, and on methods for correcting and / or avoiding the biases associated with such analyses. The talk will conclude with a brief description of some recent applications to probit and logistic regression with latent predictor variables, and with suggestions for future research.