A fire department develops a physical fitness test by having subject matter experts identify the physical demands of firefighting and then designing test components that simulate those demands. This best illustrates which validation strategy?
APredictive criterion-related validity
BConcurrent criterion-related validity
CContent validity
DConstruct validity
Content validity is established by demonstrating that the test content representatively samples the important aspects of the job domain. By having SMEs identify physical demands and then designing test components that directly simulate those demands, the fire department is building a content-valid test. No criterion data (actual job performance scores) are collected — the argument rests on the correspondence between test content and job content.
Question 2 True / False
A validity coefficient of r = .30 between a selection test and job performance should be considered useless because it explains only 9% of the variance in performance.
TTrue
FFalse
Answer: False
The 'variance explained' interpretation (r² = .09) dramatically understates practical utility. In selection contexts, even modest validity coefficients yield substantial gains when applied across many hiring decisions. Taylor-Russell tables and utility analysis show that a test with r = .30 can produce large improvements in workforce quality, especially when the selection ratio is low (many applicants per opening). Schmidt and Hunter estimated that the dollar-value utility of valid selection is enormous when aggregated across an organization.
Question 3 Short Answer
What is restriction of range, and how does it affect observed validity coefficients?
Think about your answer, then reveal below.
Model answer: Restriction of range occurs when the sample used to assess validity has less variability on the predictor or criterion than the full applicant population — typically because only those who were hired (selected) are included. This attenuates the observed correlation between predictor and criterion, making the test appear less valid than it actually is.
If an organization only hires applicants who scored above 70 on a test, the hired sample will have a compressed range of test scores (70-100 instead of 0-100). With less variance in the predictor, the correlation with the criterion is mathematically reduced. Correction formulas exist to estimate what the validity coefficient would have been in the unrestricted population. This is why predictive validity designs — hiring all applicants regardless of test scores — are methodologically superior, though rarely practical.
Question 4 Multiple Choice
Validity generalization research by Schmidt and Hunter challenged which long-held belief about selection test validity?
AThat cognitive ability tests have any validity at all for predicting job performance
BThat validity is situation-specific — that a test valid in one setting might not be valid in another
CThat content validity is a legitimate validation strategy
DThat structured interviews are superior to unstructured interviews
Before validity generalization (VG) research, the prevailing view was situational specificity — that a test's validity could vary dramatically across settings, requiring local validation for every new context. Schmidt and Hunter used meta-analysis to show that much of the observed variability in validity coefficients across studies was due to statistical artifacts (sampling error, restriction of range, criterion unreliability). After correcting for these artifacts, the true variability was much smaller, suggesting that cognitive ability test validity generalizes broadly across jobs and settings.