Threats to internal validity

You run a stress-reduction group, everyone improves, and you proudly credit your intervention — but what if a citywide holiday break relaxed everyone anyway? Internal validity asks one blunt question: can we truly say the treatment caused the change, or did something else sneak in? Examiners love the named list of threats because each one is a rival explanation for your results.

Here's the classic roster. History — an outside event (a natural disaster, a news story) happens during the study and affects outcomes. Maturation — participants simply change over time (they grow older, wiser, more tired, or naturally recover). Testing — taking the pretest itself changes performance on the posttest (practice effects). Instrumentation — the measuring tool or the observers drift or change between pretest and posttest. Statistical regression (regression to the mean) — participants selected for extreme scores drift back toward the average on retesting, with or without treatment. Selection — groups differ from the start because participants weren't randomly assigned. Attrition (also called mortality) — participants drop out unevenly, leaving non-comparable groups.

A useful memory trick: random assignment is the great equalizer. It neutralizes selection and spreads maturation, history, and regression evenly across groups, which is why the true experiment is the gold standard for internal validity. When you read a study, hunt for the design feature (control group, random assignment, matched testing conditions) that rules each threat out.

Clinical Pearl: Memorize the roster — history, maturation, testing, instrumentation, statistical regression, selection, attrition/mortality — and remember that random assignment to groups is the single best defense against most of them; that pairing is what the NCE tests.