Rasch Models in the Analysis of Repgrid Data

The present work shows how useful clinical insights can derive from a Rasch analysis of repgrids. Rasch models allow meaningful comparisons not only within elements and within constructs, but also between them. In addition, they allow the identification of unexpected evaluations that should be further explored with the respondent. Large values of infit or outfit may suggest a lack of attention in completing the grid, uncertainties in the evaluation of elements on constructs, or critical issues regarding certain elements or constructs. Classical analysis procedures and a Rasch model are applied on real repgrid data. The results indicate that the Rasch model provides additional information beyond that resulting from classical analysis procedures.