The utility of racial subgroup analysis in developing a weighted application blank

The objective of this research was to determine whether application blank information was differentially related to job success for 258 white and 138 Negro clerical employees of a large petroleum company located in a southwestern city. One-third of each ethnic group was randomly assigned to a holdout group for cross-validation purposes. The remainder, which comprised the validation sample for each croup, were separated into high and low criterion groups on the basis of a composite criterion, corrected for tenure, which included job level, performance ratings, and salary. Responses to 19 application blank responses were weighted for each ethnic group using Strong’s (1926) vertical percent method. After the scoring key was developed for each group, it was used to score the application blanks of its own holdout group as well as those of the other holdout group. A correlational analysis indicated that the white key was a significantly better predictor of job success for its own group than for the Negro group. The Negro key demonstrated validity for both groups. The white scoring key did demonstrate differential validity and a scatter-gram between weighted scores and criterion scores for the two ethnic groups based on this key was plotted. It shows that, in this case, unfair discrimination would occur against the better performers of the Negro group. It was also found that the utility of application blank information as a predictor of job success was significantly increased by developing separate weighted scoring keys.