Multivariate multilevel models for attitudes toward statistics: multi-disciplinary settings in Afghanistan

Abstract

The present paper focuses on examining students' attitudes and perception of statistics in Afghanistan universities and the factor structure of the statistical anxiety rating scale (STARS). In total, 209 undergraduate students from different disciplines in different universities in Afghanistan participated in the study. In addition to testing the factor structure of the STARS, a multivariate multilevel analysis that incorporates the correlation in the data was carried out on the aggregated subscales of the STARS scores. Results showed that the original 6-factor structure did not fit the Afghanistan data well. Exploratory factor analysis identified 5-factor constructs to best fit the data and was confirmed by the fit indices as well as a likelihood ratio test. Male students showed more positive attitudes toward statistics and a higher level of statistics anxiety than their female counterparts. Female students experienced higher levels of fear of asking for help and less anxiety in computation. Students who had taken at least a previous statistics course had lower statistics anxiety than those taking the course for the first time.