Threats to Validity: construct-irrelevant variances contributing to performance under-representation on Graduate Record Exam (GRE)
Ali Panahi

The fundamental rationale for advancing the present investigation is traced back to the recognition that construct-irrelevant variances and construct under-representation threaten the value implications and the consequential validity of the tests, hence spoiling the appropriateness and usefulness of the test scores for the purpose of decision-making (Mesick, 1989). Accordingly, this article deals in the main with the potentially interfering variances giving rise to performance under-representation on Graduate Record Exam. To these tight objectives, after undertaking the pilot testing project, dividing the subjects into two groups and sketching out the instruments, the run t-test as well as the self-reported protocols ultimately revealed that some statistically significant differences existed in both anxiety and performance of the two groups, i.e., math less-interested-in test-takers and math more-interested-in ones as a function of anxiety and math-section-related face validity, i.e., test bias. Also, the verbal section of GRE aiming at measuring the intelligence of the test takers is argued. Pedagogically assumed, the concentration is placed on the premise that the extraneous and interfering parameters culminate in misinterpretation and misuse of the tests in which confidence is put so as to make some global or local decision. Hence, they must be to the feasible extent controlled and manipulated so as to enhance the validation of and eschew misconstruing and misapplying the intended operationalized trait and score meaning for judgmental undertakings and social enterprises.

Full Text: PDF