An Approach to Estimating Logit Models of a Single Decision Maker's Choice Behavior

Randall G. Chapman, University of Alberta
ABSTRACT - This paper describes an approach to estimating multinomial logit models of a single decision maker's choice behavior. This approach is based on a recently developed "explosion" (decomposition) procedure for logit models. This approach uses the same data routinely generated for conjoint analysis modeling efforts -- preference rank order information on a set of multi-attribute product or service configurations which vary on their attribute settings. A comparison of conjoint and logit modeling approaches in two empirical case studies indicates that the two modeling approaches perform almost identically in terms of internal validity. Since the logit model has several important advantages over the traditional conjoint model, these results suggest that logit modeling of individual choice behavior should receive favorable consideration by marketing analysts interested in modeling preference or choice behavior at the level of an individual decision maker.
[ to cite ]:
Randall G. Chapman (1984) ,"An Approach to Estimating Logit Models of a Single Decision Maker's Choice Behavior", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 656-661.