Construct-Related Validity Vis-A-Vis Internal Structure of the Test

This study aimed to establish the content, construct and predictive validity evidence of a locally-made college admission and placement test for a private sectarian university. It employed empirical methods for test validation and reliability analysis using a sample of 262 college freshman students. Content-related validity evidence was initially established through the development of Tables of Specifications for the various subtests and experts‘ review on the alignment of test item content and objectives. Construct-related evidence of validity was established by exploring the internal structure of the test via factor analysis. The procedure yielded two factors and examination of the items comprising the subtests that loaded heavily on each factor suggested that the test measured two distinct but interrelated skills; namely: (1) Language and Literacy Skills and (2) Analytical Reasoning Skills, across three content areas in English, Mathematics and Science. Results from correlating test scores and first year cumulative Grade Point Average (GPA) of the students yielded a substantial overall predictive validity coefficient of 0.643. Reliability analysis using internal consistency measures yielded a high reliability coefficient Cronbach Alpha value of 0.968. These findings indicated that the proposed college admission test was a good predictor of student’s academic performance in the first year of college work, and thus, valid for its intended purpose.

Keywords – content-related validity, construct- related validity, factor analysis, predictive validity, college admission test