Exploring method effects in the six-factor structure of the survey of attitudes toward statistics (SATS-36) 4

Abstract

We use ordinal confirmatory factor analysis techniques to investigate the six-factor structure of the Survey of Attitudes Toward Statistics (SATS-36) and to estimate method effects associated with its items and factors. We extend previous confirmatory research to include posttest, as well as pretest, item-level data. We also investigate method effects by adding a common method factor to the original six-factor model. Interestingly, results reveal noticeable proportions of common variance associated with the Difficulty construct only at pretest and with the Value and Interest constructs only at posttest. We examine the characteristics of the SATS items as a possible source for these method effects.