• Version
• Download 147
• File Size 101.16 KB
• File Count 1
• Create Date August 2, 2018
• Last Updated August 2, 2018

A Preoperative Index for Construct Validity

Judgmental procedures fall short of restoring the desirable attributes of the test after it hasbeen administered. Preventive strategies must replace negating efforts. Perfect key reliability forinstance can always be ensured during the construction of the test before using it. Apparently,attributes such as inter-subject reliability, predictive validity, concurrent-validity etc. cannot bepredicted before obtaining the empirical data. So far as the construct validity is concerned there aresome a priori aspects independent of responses given by the participants. The inclusion error(irrelevant impurities diffused into the items) can be detected and removed before the subjects areexposed to the test. Also exclusion error (misrepresented intent) can be identified and essentialcontent can be supplied in advance. In order to describe the relevancy between the intent of the testmaker and the effect as distinguished by the expert(s) a numerical index based on the Shannon’,sconcept of entropy is introduced in this study. Items in the instrument are tallied into categories(taxonomical levels, sub-constructs etc.) as intended by the test-maker. Same set of items arechecked in categories as distinguished by the expert(s). Observed frequencies are cross-tabulatedon a contingency table to compute entropy values.Keywords: Construct Validity, Content Validity, inter-subject reliability, information theory

Attached Files