External validity

External validity is the extent to which findings from a study apply to populations, settings, or times beyond those directly studied. It concerns the reach of a result, not whether the study measured or estimated its target correctly.

Population validity: The extent to which study findings apply to populations beyond the participants studied. It isolates the population dimension of external validity.
Internal validity: The extent to which a study supports a credible causal conclusion for the participants and conditions studied. A study can have strong internal validity yet limited reach beyond its sample.
Randomized controlled trial: An experiment that assigns participants randomly to interventions to estimate their effects. Trial participants and conditions may differ from those encountered in routine practice.
Generalizability: The applicability of findings beyond the specific observations, participants, or conditions studied. It is often used interchangeably with external validity, though definitions can vary.
Ecological validity: The extent to which study findings apply to real-world settings and conditions. It focuses external validity on whether study conditions resemble ordinary life.
Construct validity: The extent to which an operational measure or manipulation represents the intended theoretical concept. Results travel meaningfully only if the study's measures capture the intended concepts.
Pragmatic clinical trial: A clinical trial designed to estimate an intervention's effects under ordinary healthcare conditions. Its real-world design targets broader applicability than tightly controlled efficacy trials.
Reproducibility: The ability to obtain consistent results using the same data, methods, and computational procedures. Reproducing an analysis does not establish that its result applies to other settings.
Sampling bias: Systematic error arising when the process of selecting observations makes a sample unrepresentative of its target population. Biased recruitment can leave important populations outside a study's evidence.
Statistical conclusion validity: The soundness of inferences about statistical relationships, including whether estimates and tests are appropriate. Sound statistical evidence is needed before considering how far findings extend.

Linked from 28 pages