Publication details
Reanalýza testu DISMAS: reliabilita a obsahová validita
| Title in English | Reanalyzing the Dismas Test: Reliability and Content Validity |
|---|---|
| Authors | |
| Year of publication | 2016 |
| Type | Article in Periodical |
| Magazine / Source | E-psychologie |
| MU Faculty or unit | |
| Citation | |
| Web | http://e-psycholog.eu/clanek/273 |
| Field | Psychology |
| Keywords | DISMAS; CFA; confirmatory factor analysis; DIF; content validity; reanalysis |
| Description | The goal of this paper is the reanalysis of the Czech “Assessment of the structure of mathematical abilities” test (Traspe and Skalková, 2013), designed to assess problems related to the development of mathematical abilities in children aged approx. 5–11 years. The test contains 14 developmental scales and total scores – a total of 22 test scores with percentile norms. This study uses normative (N = 878) and clinical (N = 877) samples and focuses on three objectives: (1.) the estimation of composite scores reliability using stratified Cronbach's alpha; assessment of content validity and test fairness using (2.) a series of confirmatory factor analyses and (3.) differential item functioning analysis (DIF). Reliability estimates, which took into account the multidimensional structure of composite scores, led to a two-fold (in the case of total score, a three-fold) decrease in standard errors and narrower confidence intervals. Structural models supported the assumption of a weak factorial invariance between grades 2 to 5, except the Computing Automation subtest (the relationship of which with overall math ability strengthens with age). However, the factorial structures for first graders and younger children were different and there was no clear factor structure in the clinical sample. Results also suggested that the Mathematical Ideas subtest can serve as a screening method of the overall level of mathematical abilities. Single scales were not shown to be invariant according to the DIF analyses between grades and samples, which might mean that lower scores do not simply imply lower levels of mathematical ability, but also qualitative differences. This paper offers further recommendations for test use in common assessment situations. |