• Version
• Download 120
• File Size 743.64 KB
• File Count 1
• Create Date August 2, 2018
• Last Updated August 2, 2018

The Evolution of Validity and Modern Psychometrics: Do We Need to Revisit Item Validity?

Early validity studies were concerned with the discriminative capacity of items. Were examinees with high total test scores answering items correctly and were examinees with low total test scores answering items incorrectly (see, for example, Swineford, 1936)? The statistics used for determining whether an item was valid were the biserial and point-biserial correlations. The school of thought headed up by Lindquist questioned item validity studied this way since a subjective, human ingredient was needed to offset the possibility that ambiguous or structurally deficient items could be missed more by higher ability examinees. It is interesting to note that this view of validity lasted until the works of Cronbach (1971) and Messick (1989). Implied in this earlier view of validity was that it was the responses to items that are validated. When items are validated in this way, unless one is interested in ratings for establishing content validity, of course responses to items are central to item validity claims. Nevertheless, such a view of validity lasted a number of decades through an array of different conceptions regarding criterion-related validity and construct validity, until Cronbach (1971) raised the point from a more psychological perspective that it was the interpretation of test scores that is validated rather than the test per se. This approach to understanding the concept of validity has evolved into Kane’,s (2013) interpretive argument. Item invalidity and ambiguity were no longer the emphasis in validity studies since the focus of validity was now on the interpretation.

Attached Files