Why constructs exist
A thermometer reads temperature, but no instrument reads optimism. Researchers instead define what a construct should look like in practice, then gather indicators such as questionnaire answers, reaction times or ratings by other people. The construct is the explanation offered for why those indicators move together.
Testing whether a construct is real
A measure has construct validity when it behaves as the underlying idea says it should. It ought to correlate with things the theory links it to, and fail to correlate with things the theory says are unrelated. Reliability, meaning consistent results, is necessary but does not settle the matter, since a measure can be consistent and still capture the wrong thing.
Common problems
Two scales with different names can measure nearly the same thing, and one label can hide several distinct ideas. Researchers also debate whether constructs refer to real states of people or are convenient summaries. Reading any entry in this index is easier with that caution in mind.
Further reading
- Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281-302.