Content Validity at the Crossroads

Validity

“…Ironically, the manner in which Yalow and Popham (1983) phrase their position reveals its essential flaw. They contend that content -60 -validity "resides" in a test, whereas validity is a property of.…”

“…However, Yalow and Popham (1983) argue that "such potential unfairness resides in the use of the score-based inference, not in the inference itself" (p. 13).…”

“…Nonetheless, Yalow and Popham (1983) have identified. a critical problem in certification testing, one that might be at least in part resolved by focussing on the appropriate relevance question to be asked.…”

“…Such tests should always measure genuine desired outcomes of the educational process rather than merely the outcomes to be expected from existing instructional materials and methods. (p. 671) Accordingly, if certification tests were constructed to represent ultimate domain objectives, as Yalow and Popham (1983) also desire, these should prove useful in evaluating the coverage and effectiveness of both the curriculum and the instruction, as well as in certifying -118 -student competence in terms of tasks that -by social or expert consensus -they should be able to perform. Then, accountability for unsatisfactory student competence might inpinge on the teacher and the curriculum developer and not wholly on the student.…”

Fifty reasons why student achievement gain does not mean teacher effectiveness

“…Popham (1983) has identified four data-sources for describing whether students have received instruction that would enable them to perform satisfactorily on a test: (1) observations of classroom transactions; (2) analyses of instructional materials; (3) instructor selfreports; and (4) student self-reports. Although he views these sources as methods for determining the adequacy of test preparation (Yalow & Popham, 1983), they can be considered as techniques for gathering evidence of instructional validity. Unfortunately, Popham's (1983) evaluation of those techniques suggests that the process of estimating the percentage of a standardized test that has been covered by teaching has numerous methodological problems related to executing the datagathering procedures (see Leinhardt, 1983;Schmidt et al, 1983).…”