Does Taking a More Holistic View of Personality Improve its Predictive Utility? A Comparison Multiple Regression, Fuzzy Cluster Analysis, and Indirect Mixture Modeling for Predicting Leadership Effectiveness.

摘要

When using personality to predict leadership outcomes, researchers typically use either bivariate correlations or additive, linear multiple regression models. Recently, however, some researchers have suggested that the relationship between personality and leadership may be more complex than typically represented in the literature. The purpose of the current study was to evaluate the predictive utility of two under-utilized statistical modeling techniques that take a holistic approach to modeling the personality-leadership relationship -- fuzzy cluster analysis and indirect mixture modeling. These statistical techniques were applied to an archival data set containing personality and leadership effectiveness information for 619 department managers at a grocery chain in the United States. Using this data, four statistical models were tested and compared in terms of their overall fit, predictive validity, and generalizability: (1) a traditional main-effects only multiple regression model, (2) a regression model that includes theoretically relevant interactions and nonlinear effects, (3) fuzzy cluster analysis, and (4) indirect mixture modeling. Results indicated that although indirect mixture modeling outperformed both multiple regression and fuzzy cluster analysis across all four leadership criteria in the development sample, this technique experienced the most shrinkage in the validation sample. In contrast, the main-effects multiple regression models explained a small, but significant amount of variance in the leadership effectiveness outcomes, the magnitudes of which were stable across samples.