Income and occupation as explanatory variables: Their power combined vs. separate

Summary This study put a new explanatory variable, relative occupational class income (Y/O), through multivariate analysis in order to test the variable's predictive and marketing segmentation powers. Y/O was tested against income and occupation used separately as explanatory variables. The mid-1960s new-car market was used for the analysis. With respect to the power to predict individual buyer behavior, the new variable did not prove itself to be superior to income and occupation used separately. With respect to market segmentation, however, Y/O often identified more meaningful market segments, where the probability of buyers choosing the new car class involved was obviously much higher (or lower) than was true for the entire market. Whether it would be worthwhile to use the combination variable, Y/O, instead of occupation and income separately as segmentation variables is, nevertheless, a matter of marketing judgment and not of statistical power. It obviously takes time, money, and effort to build a combination variable such as Y/O, and it is often a complex procedure to use and interpret the results from such a variable. In a statistical sense, the difference in power in using Y/O as compared to using occupation and income separately in a multiple regression procedure is trivial. Y/O now needs to be tested as a market segmentation variable on other types of products. Yet, on the basis of the results from this study (and the previous studies involving income versus social class), one wonders if it is not better to use such demographic variables separately as explanatory variables.