Business Commons^™

Applications Of Business Analytics In Marketing: Joint Modeling Of Correlated Multivariate Outcomes, Sumin Han

Doctoral Dissertations

In this dissertation I develop a general regression methodology for mixed multivariate outcomes. This methodology extends the generalized linear mixed model paradigm (glmm) to allow for correlated multivariate normal random effects across regression equations for differing outcomes. This methodology, referred to as joint modeling, is particularly useful in business and marketing applications where multiple outcomes of varying data type must be analyzed simultaneously with regression.

I apply joint models to binary and continuous measures of customer loyalty in a large multinational survey of car owners. Survey respondents’ word-of-mouth and desire to switch brands were used as proxies for attitudinal loyalty …