Measure Of Validity

Criterion validity is the final measure of validity. A criterion is an instrument that measures the same variable. Correlations can be conducted to see whether different instruments measure the same variable. There are three ways to determine criterion validity:

1. Convergent validity—it declares that an instrument is greatly correlated with instruments measuring similar variables.

2. Divergent validity—shows that an instrument has low correlation with instruments that measure different variables. In this case, for example, there should be a weak relationship between an instrument that measures self-efficacy and one that measures motivation.

3. Predictive validity—means that the instrument should be highly correlated to future criterions. For

Each year more than one million applicants compete to obtain a seat at one of the state universities. The selection procedure at Iranian universities is carried out through university entrance examinations (UEE) administered for different levels of education namely bachelor, master’s, and PhD. Besides being an academic competition, the UEE has turned into a seal of prestige in the Iranian society. UEE powerfully influences many spheres of social, academic, and educational lives of the applicants that the event could be a relevant subject to current developments and challenges in language testing and assessment. Consequently, the construct validity of this test needs due attention. Although some sporadic validity investigations (Barati & Ahmadi, 2010; Razavipour, 2010; Tahmasbi & Yamini, 2012) have been conducted on the UEE test at M.A. levels, to the best knowledge of the authors, no single study about the validity of the present test has been carried out yet.

According to Messick (1995), although validity is a unitary concept, six distinguishable aspects of construct validity can be highlighted as means of addressing the notion of validity as a unified concept. These six aspects of construct validity are content, substantive, structural, generalizability, external, and

It adds to the content aspect of construct validity the need for empirical evidence of response consistencies or performance regularities reflective of domain processes (Loevinger, 1957). The cognitive modeling of the response process of the test takers is an evidence to support the substantive aspect of construct validity.

The structural aspect of validity addresses the degree to which the scoring model matches the structure of the test (Messick, 1995). As Wu & Adam (2007, p.21) argued, “An aggregated item score is meaningful just when all the test items tap into the same latent variable. Otherwise, one outcome score for different dimensions is uninterpretable, since the same total score for students A and B could mean that student A scored high on latent variable X, and low on latent variable Y, and vice versa for student B”. Hence, the dimensionality of the test is a determining factor in the choice of its scoring