Definition · Interviewing and assessment
Assessment Validity
Definition
Assessment validity is the degree to which evidence supports using a test or selection method for a specific hiring decision. Predictive validity, the form employers care about most, is how well scores collected at hiring predict later job outcomes such as performance ratings, quality scores, sales or retention.
Also called: Predictive validity, Test validity, Validity of selection methods
Formula
Validity coefficient (r) = correlation between assessment scores and a later job outcome, for the same group of hires
- Assessment score
- Each hire's score on the test or interview at the time of hiring
- Job outcome (criterion)
- A measure of performance collected later, such as 90-day QA score, supervisor rating or sales
- r
- Pearson correlation, from -1 to 1; higher positive values mean the assessment predicts the outcome better
Worked example
A contact center gives 220 new agents a 20-minute call simulation at hiring. Six months later it correlates simulation scores with each agent's average QA score and finds r = 0.32. That is a useful predictor for one assessment. A typing test given to the same group correlates at r = 0.05 with QA scores, so it predicts almost nothing about call quality and is a candidate for removal.
Types of validity evidence
The EEOC's Uniform Guidelines on Employee Selection Procedures (1978) recognize three kinds of validation:
- Content validity: the test samples the actual work. A data-entry test for a data-entry job is the classic case.
- Criterion-related validity: test scores correlate with job outcomes. Predictive validity tests applicants and measures outcomes later; concurrent validity tests current employees and compares with their current performance.
- Construct validity: the test measures an underlying trait, such as conscientiousness, that is shown to matter for the job.
Most employers rely on content evidence for work samples and criterion evidence for everything else.
Validity vs. reliability
Reliability is consistency: would the candidate get a similar score tomorrow, or from a different rater? Validity is whether the score means what you use it for. A test can be reliable and useless, like a precise measure of something unrelated to the job. It cannot be valid without being reasonably reliable. Unstructured interviews often fail on reliability first, because two interviewers rate the same candidate differently.
What good numbers look like
Single selection methods rarely exceed a validity coefficient of about 0.5. Values around 0.2 to 0.4 are typical for well-built structured interviews, work samples and ability tests, and combining methods usually raises the overall prediction. A coefficient below about 0.1 means the method adds little. Small samples make coefficients unstable, so a study on 30 hires tells you less than one on 300. Ask vendors for criterion studies on roles similar to yours, with sample sizes, rather than a single headline number.
Why it matters legally
Under Title VII, if a selection procedure causes disparate impact, the employer must show it is job-related and consistent with business necessity. Validity evidence is how you show that. Without it, a test that screens out one group at a higher rate is hard to defend, regardless of how sensible it looks. Validation also informs where to set a cut score.
How to run a simple local check
You do not need a psychometrician to start. For any pre-employment assessment you use at volume, store each hire's score, pick one or two outcomes that matter (90-day retention, QA average, time to proficiency) and correlate them after six months. Even rough results tell you which steps earn their place in the funnel and which only add drop-off.
Frequently asked questions
What is predictive validity in hiring?
It is how well scores from a selection method, collected when candidates apply, predict their later job outcomes such as performance, quality or retention.
What is the difference between test validity and reliability?
Reliability is whether a test gives consistent scores. Validity is whether those scores are meaningful for the decision you make with them. A valid test must be reliable, but a reliable test is not necessarily valid.
What is a good validity coefficient for a selection test?
Coefficients around 0.2 to 0.4 are typical for useful methods, and few single methods exceed about 0.5. Combining several valid methods usually improves prediction.
Which selection methods have the highest validity?
Research generally places structured interviews, job knowledge tests, work samples and cognitive ability tests among the stronger predictors, with unstructured interviews and years of experience weaker.
Do employers have to validate pre-employment tests?
US law requires validation evidence when a test causes adverse impact against a protected group. Validating before that happens is good practice and makes the test easier to defend.