Performance Prediction Models 2 Flashcards

In TAPAS performance prediction research, what does 'incremental validity' specifically refer to?

Answer: The degree to which TAPAS scores improve predictive accuracy beyond what is already explained by cognitive ability tests like the ASVAB

Incremental validity measures how much unique predictive variance TAPAS personality dimensions contribute over and above existing predictors such as cognitive ability. Because personality and cognitive ability are largely independent, TAPAS can meaningfully improve overall prediction of job performance criteria beyond ASVAB scores alone.

When researchers 'cross-validate' a TAPAS prediction model on a new sample, what phenomenon are they primarily trying to quantify?

Answer: Validity shrinkage — the reduction in predictive accuracy caused by capitalizing on chance relationships in the derivation sample

Cross-validation exposes validity shrinkage: a prediction equation optimized on one sample will over-fit to chance covariances in that sample, and its R² will drop when applied to a fresh sample. Quantifying shrinkage is essential before a TAPAS model is operationalized for selection.

How does 'criterion contamination' differ from 'criterion deficiency' in evaluating TAPAS performance prediction models?

Answer: Criterion contamination occurs when the criterion measure is influenced by knowledge of TAPAS scores, inflating apparent validity; criterion deficiency occurs when the criterion fails to capture all relevant performance dimensions

Criterion contamination introduces spurious validity: if a rater knows an employee's personality profile and lets that knowledge color the performance rating, the TAPAS–criterion correlation is artificially inflated. Criterion deficiency, by contrast, deflates validity by omitting key performance dimensions from the criterion measure. Both threaten the accuracy of TAPAS prediction model evaluations, but in opposite directions.

What is 'differential prediction' in the context of TAPAS, and why is it important for fair employment decisions?

Answer: Differential prediction examines whether the TAPAS prediction equation produces equally accurate and unbiased performance forecasts across demographic subgroups, such as gender or ethnicity

Differential prediction analysis tests whether a single regression equation fits all subgroups equally — examining intercept and slope differences across groups. If the model systematically over- or under-predicts performance for one subgroup, its use in selection would be unfair regardless of overall validity, making this analysis a legal and ethical requirement.

In TAPAS prediction research, what is the purpose of examining a 'suppressor variable' within a prediction composite?

Answer: A suppressor variable increases the composite's validity by removing irrelevant variance from another predictor, even though the suppressor itself may correlate weakly with the criterion

A suppressor variable may have a near-zero correlation with the criterion but a meaningful correlation with error variance in another predictor. Including it in the composite 'suppresses' that irrelevant variance, allowing the primary predictor's true relationship with the criterion to emerge more clearly and boosting overall composite validity.

What does 'validity generalization' mean when applied to TAPAS performance prediction models across military occupational specialties (MOS)?

Answer: Validity generalization is the empirical finding that situational specificity is largely an artifact of sampling error, and that TAPAS validity coefficients are more consistent across MOS contexts than classic study-by-study comparisons suggest

Meta-analytic validity generalization research, pioneered by Schmidt and Hunter, shows that much of the apparent variability in predictor–criterion correlations across studies is due to statistical artifacts (sampling error, range restriction, criterion unreliability). Applied to TAPAS, this supports transporting validated prediction models across MOS settings rather than requiring a new local validation study for every specialty.