Validity

The word validity is primarily a measurement term, having to do with the relevance of a measuring instrument for a particular purpose, but it has been broadened to apply to an entire study. A research investigation is said to have internal validity if there are valid causal implications and is said to have external validity if the results are generalizable.

As far as measurement is concerned, the most important property of a measuring instrument is the extent to which it has been validated with respect to some gold standard whose validity has been assumed to be taken for granted. For example, if scores on a test of mathematical aptitude (the instrument to be validated) correlate highly with scores on a subsequent test of mathematical achievement (the gold standard), all is well, and the aptitude test would be regarded as valid.

In the early 20th century, the methodological literature referred to three kinds of validity: (1) content validity (expert judgment, i.e. postulation of the gold standard itself), (2) criterion-related validity (the type of validity mentioned in the previous paragraph), and (3) construct validity (the extent to which the scores obtained using a particular measuring instrument agreed with theoretical expectations). There were also subtypes of criterion-related validity {concurrent and predictive) and construct validity {convergent and discriminant), but more recently the general label "construct validity" not only has become more popular but also has been alleged to include content validity and criterion-related validity as well.