Behaviours versus Demographics as Identifiers of CHAID Splits

Demographic variables have traditionally been adopted as descriptors for segmentation studies due largely to their inherent convenience. This paper uses data from a large empirical study to investigate the predictive power of a range of segmentation variables. Specifically, the ability to identify segment membership using descriptive demographics is compared to the identifying ability of other behaviours. Segments are determined by propensity to react to particular marketing stimuli. Propensity to react is measured using the Juster verbal probability scale. For the product category examined, results highlight a stronger ability of behaviours in identifying segment membership. That is, propensity to react differently to marketing stimuli (i.e. behaviour that determines segment membership) is often strongly correlated with other post buying behaviours (of different products). Results as such highlight the additional benefit that behavioural descriptors can provide to segmentation studies in addition to having significant implications for the facilitation of marketing and communication strategies for segments. That behaviours should correlate well with other behaviours is not surprising. That behaviours are not often used to identify segment membership is surprising. This paper concludes with suggestions for segmentation practice.