Revalidating a measurement instrument of spatial thinking ability for junior and high school students

Abstract

Spatial thinking is a set of cognitive abilities that enable people to organize, reason about, and mentally manipulate both real and imagined spaces. One of the available measurement instruments is the Spatial Thinking Ability Test (STAT). Given the critical need for spatial thinking ability measurement for junior and high school students, and the popularity of STAT to measure spatial thinking ability, revalidation of STAT is necessary as STAT was developed primarily for university students and validation of the original STAT was based on the classical test theory from which the findings are notoriously sample dependent. We used Rasch modeling to revalidate STAT as it allows parameters to be mutually independent and measures to be interval. The sample included 1340 junior and high school students. Item fit statistics, Item Characteristics Curves, unidimensionality test, and the Wright map provided evidence for the construct validity of STAT measures. The reliability of the instrument was moderate. Wald test for item measure invariance of individual items showed that among sixteen items seven items were variant in measures. The Anderson LR test indicates that the Rasch difficulty measures of STAT were not adequate for invariance. There was no DIF between two subsamples based on gender, suggesting fairness of the instrument in terms of gender. The above results suggest that STAT possesses certain degrees of validity, reliability, and fairness, although there is still room for further improvement.