Construct Validity of TIMSS 2011 Mathematics Cognitive Domains for Turkish Students

Abstract

Article History: Received 14.06.2014 Received in revised form 23.07.2014 Accepted 12.10.2014 Available online 01.03.2015 This study investigated empirical evidence for construct validity of the TIMSS 2011 Mathematics test by using sample of fourth and eighth grade Turkish students (n=4091). The purpose of this study was to analyze construct validity of the TIMSS 2011 mathematics test by using Turkish students’ data. A confirmatory factor analysis model was formed with using cognitive domain definitions for each mathematics items presented by TIMSS assessment framework. In this three-factor model, observed variables (items) were associated to three latent variables (mathematics cognitive domain), namely knowing, applying and reasoning. Fit indices of confirmatory factor analysis, conducted by LISREL, showed mixed results. For eighth grade booklets, the proposed model was supported by the data, however, for fourth grade booklets the proposed model was not supported completely. Besides, seven of the items were found to be correlated with other cognitive domains than with the defined one. Evaluation of these items showed that four of them were problematic. These results showed that for the eighth grade, the mathematics cognitive domain defined by TIMSS also works for the Turkish sample and that a scale score for each cognitive domain could be used confidently. However, for the fourth grade, a scale score for each cognitive domain should be used with caution for Turkish students. © 2015 IOJES. All rights reserved