Are Students Statistics “STARS”? Psychometric Properties of the Statistics Anxiety Rating Scale (STARS)

Many undergraduate psychology students resent having to take statistics and research methods courses, yet these classes are important elements of a psychology degree. Thus, poor performance in these subjects can be vexing for both instructors and students. One factor predicting achievement in these courses is statistics anxiety. Cruise, Cash, and Bolton (1985) developed the statistics anxiety rating scale (STARS) to measure this important influence on performance. I measured statistics anxiety using the STARS as part of a larger study. Here, I report on an assessment of the reliability, validity, and factor structure of the STARS that I conducted to ensure that the measurement instrument I used was sound. An exploratory factor analysis (N = 112) reduced the original six-subscale measure to a three-factor solution that had clearer, simpler factor structure. Fifteen of the original 51 items cross-loaded on multiple subscales, so I dropped them from the model. Confirmatory factor analysis (N = 98) on the remaining 36 items found that fit was not appreciably different in the three- versus six-factor model, both of which fit poorly, but I kept the three-factor model for its simplicity and theoretical soundness. I then subjected the items on each of these subscales to a further factor analysis in an iterative process to remove items that did not support clear unidimensional subscale factor structure. Finally, I will report on results of a path analysis that tested whether the STARS was related to statistics self-efficacy and class grade in the expected manner, accounting for prior GPA