Glossary
Internal validity
What is internal validity?
Internal validity is the extent to which a study can establish a causal relationship between the treatment and the outcome, ruling out alternative explanations.
Threats to internal validity
Threats to internal validity include
- confounding variables,
- selection bias,
- maturation (participants change over time regardless of treatment),
- history (external events influence outcomes),
- testing effects (the measurement itself changes behavior),
- and regression to the mean.
Random assignment prevents systematic assignment based on participant characteristics, but it does not guarantee that groups differ only in treatment in a particular sample. Differential dropout, measurement problems, or other aspects of study conduct can still weaken causal inference. Strong causal design need not require an artificial setting.
Applied example
A study showing that students who use a new app score higher on exams may lack internal validity if students self-selected into the app group, because motivated students who chose to use the app might have differed even without the app.
Why it matters
Internal validity matters when evaluating causal research: for a causal claim, researchers must assess whether the observed difference can reasonably be attributed to the treatment.
Sources and further reading
Campbell and Stanley’s original discussion of validity distinguishes the causal question from the question of generalization. It also notes that randomization does not produce perfect initial equivalence in every sample.