Rigour in quantitative analysis: The promise of Rasch analysis techniques

The article argues for greater rigour in quantitative analysis through the use of the Rasch model, by comparing this model to traditional analysis techniques. It demonstrates how this model can be used to analyse and improve a test instrument and secondly, how it is able to generate linear (interval strength) scores, thus meeting the assumptions of parametric statistical tests such as t-tests and ANOVA. The article first presents a brief overview of the Rasch model followed by the re-analysis of the data presented in the previous article in this issue by Hattingh et al. It then demonstrates how the Rasch model is able to “unconfound” learner and item analysis, thus providing separate measures of reliability for each. Since items and learners can be assigned scores on the same metric, informative visuals can be generated in which the two are compared.