Estimation of convergent and discriminant validation by the multitrait-multimethod matrix: example of Personal Globe Inventory on Macedonian sample

Abstract

The goal of the paper is to demonstrate the procedure of evaluating convergent and discriminant validity of a set of measures by analyzing the multitrait-multimethod matrix on the example of the six Holland types of vocational interests (RIASEC) measured by using the Personal Globe Inventory (PGI). For this purpose 737 students of various universities from the Republic of Macedonia (FYROM) were asked to complete the macedonian version of the PGI. PGI follows the multimethod approach to measuring psychological constructs. It measures each of the six Holland’s vocational interests types in three separate ways – by using activity preference items, competence assessment items and vocational preference items. In this way it provides a total of 18 measures (6 x 3) for the six Holland’s vocational interest types. After MTMM analysis was conducted, the results showed that the obtained measures may be considered to possess convergent validity on the examined sample. Relative to discriminant validity, all measures fulfilled the less restrictive requirements for this type of validity, but not the more restrictive ones. Results showed the existence of a pronounced method effect, that resulted in increased intensity of correlations between measures obtained by using the same method of measurement, but without any significant effect on the structure of correlations between vocational interest type measures. The procedure for conducting the analysis of the multitrait-multimethod matrix was described in the text.