Latent Class regression is a relatively new analytic technique that has been shown to be superior to more traditional techniques such as cluster analysis. In Latent Class regression, the procedure seeks to accomplish two tasks simultaneously:
The purpose of each regression model is to determine why members of that segment do or do not prefer a given brand. For example, let’s say brand preference for some respondents is driven by perception of value and low price. For other respondents, perceptions of quality and prestige drive brand preference. Further, let’s assume a latent class segmentation was conducted which identified these two segments of customers. Each segment would have its own regression model. Each regression model would explain why people in that segment prefer or do not prefer a given brand. That is, people in segment one would be very price sensitive as reflected in their regression model containing a large coefficient on the perceived price and value variables. Conversely, people in segment two would have a regression model containing large coefficients on brand attributes that reflect quality and prestige. All of these brand attributes, if influential to the decision to choose one brand over another, would appear in the regression model for a given segment. The factors that did appear in the model would vary across segments, reflecting the different decision dynamic that exists across segments.