Abstract

This study was conducted to compare the stability of validity between two scale construction methods(empirical keying versus factorial scaling) of biographical data. In addition to the relative validity of weighting systems, biographical and job performance data from 300 saleswomen were collected. 194 cases were used in item analyses and 106 cases were used in the cross-validation. After a series of item analyses were conducted, 23 items among 114 biographical items which had been developed by Owens(1976) were included to the validation study. 5 factors of 23 biographical items were observed by common factor analysis. The results showed that the validity of empirical keyed biodata was higher than that of the factorial scaling method, but that significant validity shrinkage was observed in empirical keyed biodata. Regression-weighting and equal-weighting methods were superior to other weighting systems. The implications, some possible limits and the future directions of the present research were discussed.