PSY8765: Testing and Assessment in Workplace Psychology

This question gets at one of the central practical tensions PSY8765 addresses throughout its coverage of workplace assessment, because the answer reveals why I/O psychologists cannot simply optimize a selection system for raw predictive validity alone, and must instead navigate a genuine tradeoff between several simultaneously important considerations. The validity generalization research, most notably Schmidt and Hunter's extensive meta-analytic body of work spanning decades and a very wide range of occupations, consistently finds that measures of general cognitive ability show meaningfully strong correlations with job performance criteria across a broad swath of job types, particularly jobs involving greater complexity and information-processing demands, and that this validity holds up reasonably well (generalizes) across different organizational settings rather than being highly situation-specific — making cognitive ability tests, on pure predictive-validity grounds alone, one of the more attractive selection tools available to I/O practitioners. However, a large and equally well-documented body of research has found that cognitive ability tests frequently produce meaningful score differences across certain demographic groups, in a pattern that can produce adverse impact under the EEOC's four-fifths guideline and the broader disparate impact legal framework covered in PSY8755 — meaning an employer relying heavily or exclusively on cognitive ability testing risks a selection system that, even if highly predictive of performance, disproportionately screens out members of a legally protected group, triggering legal scrutiny and a need for the employer to defend the procedure's job-relatedness and necessity. This creates a genuine practical dilemma for I/O practitioners: a selection tool can be highly valid in a narrow statistical sense (it does predict job performance reasonably well) while simultaneously creating significant adverse impact and legal exposure, and PSY8765 emphasizes that practitioners cannot resolve this tension by simply picking the single most statistically predictive tool and ignoring its adverse impact profile, nor can they responsibly ignore validity altogether in favor of whatever tool happens to minimize adverse impact regardless of how poorly it actually predicts performance. The practical resolution the field has developed, and which PSY8765 covers directly, involves combining multiple predictors thoughtfully — research demonstrates that certain combinations, such as pairing cognitive ability measures with structured interviews, situational judgment tests, or personality assessments (which tend to show smaller group differences than cognitive ability tests alone), can produce a composite selection system that retains much of the predictive validity benefit of cognitive ability testing while measurably reducing the overall level of adverse impact compared to relying on cognitive ability testing in isolation. This is precisely why contemporary I/O practice rarely recommends a single-predictor selection system built purely around cognitive ability testing, instead favoring carefully validated, multi-predictor batteries designed with both predictive validity and adverse impact simultaneously in view from the outset of the selection system's design.