195-30: Single Event Timing Model with Heterogeneity: Predicting the Time of the First Home Purchase by Immigrants

It is well known that unobserved heterogeneity can produce misleading estimates in survival analysis (see for example, Allison, 1995, p.233). The goal of this paper is to show how to estimate and apply a single event timing model with heterogeneity to real data with SAS. We consider three different models: An Exponential-gamma model, when individual-level time-to-event is exponentially distributed, and the event-rate is gamma-distributed A Weibull-gamma model, when individual-level time-to-event has a Weibull distribution, and the eventrate has a gamma-distribution (Exponential-gamma is a particular case of this model) A latent class Weibull model that allows for heterogeneity in both shape and scale parameters. Thus, the assumption is that several segments of individuals exist with their own fixed but unknown values of shape and scale parameters.