An Empirical Case for Residual Measures of Status Inconsistency Effects

Properly manipulated residuals resulting from "status predicting status" regressions constitute a precise and unique measurement of status inconsistency without the definitional and methodological problems and limitations of simple and simultaneous cross-classification techniques associated with the typical "dummy variable regression" approach. Residuals entered as independent variables in multiple regression models including usual status variables allow determination of "inconsistency effects" separate from main effects of status dimensions without sacrificing variability and predictability and precluding analyses of higher-order interaction terms. Several theoretically-based, nonlinear, and modified dummy-variable transformations of residuals are suggested which avoid linear determinancy while measuring the relative influence of status and degrees of directional and nondirectional status inconsistency. Several residual regression models are discussed and empirically compared: 2-variable and multivariable status-inconsistent types and full regression models of statuses and residual inconsistents. Results of tests with selected models predicting political attitude empirically demonstrated the practical and theoretical utility of the residual regression method of measuring status inconsistency effects relative to those of status. Implications for future mobility and status inconsistency research are outlined. (Author)