Validation and validity beyond Messick

Abstract

Following the texts of the Messick (1981: 10; 1980: 1023) formulations more closely, another representation of his well-known “Facets of validity” matrix is possible. This matrix can be read as four claims about language testing (“The technical adequacy of inferences made from test scores depends on multiple sources of empirical evidence; The appropriateness of inferences made from test scores relates to the detrimental or beneficial consequences...” and so forth). This representation still follows Messick’s argument, but rather than validity, articulates the coherence of a number of assessment concepts. Such concepts as the technical adequacy of our assessment instruments, their appropriateness, the technical meaningfulness (interpretation) of their measurements, their utility, their social impact and public defensibility indicate that we can usefully reconceptualise not only validation and validity, but all of our efforts at designing assessments responsibly. The current debates in South Africa about standardisation and equivalence can be deepened if we examine ways of going beyond conventional notions of validation and validity, and take responsible design criteria to constitute the overriding condition(s) for the development of assessment instruments.