Abstract

The purpose of this research was to examine the validity of empirical scaling methods and factor scaling methods of biographical items. 154 biographical items were administered to 4813 insurance salespersons. The subjects were divided into three groups: large(4118 salespersons), middle(835), and small(393) group. For each group, empirical scaling methods (unit weighting, regression weighting, WAB, Stokes and correlation coefficient) and factor scaling methods (unit weighting, regression weighting) were employed to examine the variation of the validity coefficients. The results revealed that the larger the sample, the more stable the cross-validation. It was also found that in large groups, empirical scaling methods (unit weighting, WAB, Stokes, correlation coefficient) and factor scaling methods (unit weighting) showed stable validity in crass-validation. The regression weighting, however, showed meaningful decreases in validity when cross-validated for both the empirical scaling method and the factor scaling method. Nevertheless, the regression weighting was more valid than the other weighting methods for the development of empirical scaling and for the cross validation sample. Furthermore, there is no statistically significant variation with respect to the size of validity coefficient in the developing samples for the three groups. Because regression weighting is more easily available than any other weighting method, it should be considered to use the regression weighting method for future research. Lastly, the limitations of this study and future research directions were discussed.