Evaluating test validity: reprise and progress

The AERA, APA, NCME Standards define validity as ‘the degree to which evidence and theory support the interpretations of test scores for proposed uses of tests’. A century of disagreement about validity does not mean that there has not been substantial progress. This consensus definition brings together interpretations and use so that it is one idea, not a sequence of steps. Just as test design is framed by a particular context of use, so too must validation research focus on the adequacy of tests for specific purposes. The consensus definition also carries forward major reforms in validity theory begun in the 1970s that rejected separate types of validity evidence for different types of tests, e.g. content validity for achievement tests and predictive correlations for employment tests. When the current definition refers to both ‘evidence and theory’ the Standards are requiring not just that a test be well designed based on theory but that evidence be collected to verify that the test device is working as intended. Having taught policy-makers, citizens, and the courts to use the word validity, especially in high-stakes applications, we cannot after the fact substitute a more limited, technical definition of validity. An official definition provides clarity even for those who disagree, because it serves as a touchstone and obliges them to acknowledge when they are departing from it.