Assessing the Construct Validity of Rational Biodata Scales

Awareness of the theoretical and practical limitations of empirically keyed biodata instruments has led to an increasing interest in rational keying techniques that group items into scales having substantive meaning. This research investigated the construct validity of rational biodata scales for measuring temperament constructs. The biodata items were carefully developed and re- fined to enhance construct relevance, response variability, relevance to the target population, readability, nonintrusiveness, and to minimize social desirability bias. In Experiment 1, results from a sample of 229 U.S. Army civilians indicated that rational biodata scales (a) can approximate temperament mea- sures of the same constructs, (b) have lower correlations with social desirability than temperament measures, and (c) show criterion-related validity. In Experi- ment 2, job incumbents rated the desirability of both the temperament and biodata response options. Results showed that the biodata scales are more subtle than temperament scales measuring the same constructs. The implications of these findings are discussed.