Investigating the Measures of Relative Importance in Marketing Research

Abstract

Determining the relative importance of various predictors in a marketing research model is important for both theoretical and practical reasons. To date, the most commonly used methods to assess relative importance have involved examining either the regression coefficients or zero-order correlations of each predictor. Unfortunately, these indices are problematic when the predictors are correlated, as is the case with many of the drivers of service-provider switching, loyalty studies, satisfaction models and other marketing research. In this paper, we introduce Dominance Analysis to an audience of researchers in marketing research and empirically demonstrate its usefulness for assessing predictor relative importance. Using a Monte Carlo simulation, we first compare the accuracy of five traditional methods used in marketing research assessing relative importance and comparing them to Dominance Analysis. There are theoretical, as well as empirical, advantages to using Dominance Analysis over other methods, and these are discussed in the context of an empirical example using data drawn from a larger study of auto-repair service customers (n = 355).