Blackwell Reference Online

Extract

The term “content validity” was initially introduced in 1954 ( American Psychological Association, 1954 ) to mean the adequacy of test content as a sample of a larger content domain. Claims of adequacy were to be based on the “representativeness of the test content in relation to the universe of items adopted for reference” ( American Psychological Association, 1954: 13 ). Such claims evaluated test development without recognizing the difference between test development and drawing inferences from test scores (i.e., the basic concept of validity). Three conferences on content validity held at Bowling Green State University concluded that “there is no such thing” ( Cranny, 1987 ). The 1985 Standards referred not to content validity but to “content‐related evidence of validity,” showing that test content is “appropriate relative to its intended purpose” (see American Educational Research Association et al., 1985: 90 ). A test intended to measure a factor‐analytically determined construct need not sample all kinds of test items defining the factor. There may be “appropriate” evidence of validity even with a single‐item type among the many that have in the past served to define the factor. Valid inferences from scores require content appropriate for such inferences, regardless of the nature of the measurement or of the attribute (construct) being measured, but it is not the case ... log in or subscribe to read full text