Assessment expertise project: validity of assessment
When candidates take a test, whether it is a written examination, a practical performance, or submitting a portfolio of work, it is taken so that an examiner can assess the candidate’s performance relative to a standard. There is normally a reason for doing this: to select people for further education or training; to assess suitability for a specific job; to certify what individuals can do in a particular domain; and so on.
Validity is essentially the concern that the inferences drawn from a candidate’s performance and any decisions based on these inferences are reasonable and fair. In some cases, demonstrating validity is relatively straightforward, but in most cases, the verdict will not be as apparent.
There are two main threats to the validity of any test use (Messick, 1989). The first is that the test does not measure everything that it should (construct underrepresentation). The second threat to validity is that the test measures things that it should not (construct irrelevant variance). In practice, these threats are likely to be relatively subtle, perhaps especially in the case of construct irrelevant variance, where statistical analyses of the item scores may draw attention to a problem in the first instance.