Regression v. CHAID
Message posted by Jim on February 15, 2000 at 12:00 AM (ET)

I am trying to evaluate two alternative methods of identifying predictive variables in a customer satisfaction survey. The dependent variable is an overall satisfaction question using a 7-point scale. The rest of the questionnaire consists mainly of yes/no and agree/disagree (dichotomous) questions. The two alternative approaches in contention are stepwise regression and CHAID. As a non-statistician I am reasonably familiar with the concept behind each method but I am not confident enough to determine which is better for this task. Can anyone help?