Validity

A bathroom scale that reads exactly 300 pounds every single time is beautifully reliable — but if it's supposed to measure your IQ, it's useless. That gap is validity: does the test actually measure what it claims to measure? It's the single most important quality of any assessment, and examiners test the three big types relentlessly.

Content validity asks whether the items adequately sample the domain — does a depression inventory cover the full range of depressive symptoms, not just sleep? Think of it as expert-judged coverage (a spelling test drawn only from words starting with 'A' has poor content validity). Criterion-related validity asks whether scores relate to an external benchmark, and it splits by timing: concurrent validity correlates the test with a criterion measured at the same time (does a new anxiety scale match a validated one given today?), while predictive validity correlates the test with a future outcome (does the GRE predict grad-school GPA?). Construct validity — the broadest and most theoretical — asks whether the test measures the abstract trait it targets (intelligence, resilience). Its two workhorses: convergent validity (correlates highly with measures of the same construct) and discriminant validity (correlates low with measures of different constructs).

Now the relationship examiners adore: reliability and validity are linked, and reliability sets a ceiling on validity. A test can be reliable without being valid (consistently measuring the wrong thing), but it cannot be valid without being reliable — you can't hit the target dependably if your scores bounce around. Reliability is necessary but not sufficient for validity.

Clinical Pearl: Reliability is consistency; validity is accuracy — and the one-way street examiners test cold: a test can be reliable without being valid, but never valid without being reliable.