External validity & generalizability

A brilliant therapy works perfectly — on 40 undergraduate psychology majors in a soundproof lab on a Tuesday. Would it work with real clients in a busy community clinic? That leap is external validity: the extent to which study findings generalize beyond the specific people, settings, and conditions studied.

There are two flavors examiners test. Population validity asks whether results generalize from the sample to the broader population — would findings from those undergraduates hold for older adults, other cultures, or clinical populations? A narrow, unrepresentative sample limits population generalization. Ecological validity asks whether results generalize from the artificial setting and conditions of the study to real-world environments — does a tightly controlled lab result hold up in a messy, distraction-filled clinic?

Here's the tension examiners love: internal and external validity often trade off. The tighter you control a study to prove the treatment caused the change (high internal validity), the more artificial it becomes, which can shrink external validity. Threats to generalization include the Hawthorne effect (people behave differently because they know they're being observed), reactive/interaction effects of testing, and selection–treatment interactions where the treatment only works for the particular sample studied. The fix is representative sampling and replication across diverse people and settings — one study rarely settles anything.

Clinical Pearl: Keep the pair straight — internal validity = did the treatment cause the effect; external validity = does the effect generalize. External validity has two subtypes the NCE probes: population (to whom?) and ecological (to what settings/conditions?).