An Algorithm for Testing Unidimensionality and Clustering Items in Rasch Measurement
Rudolf Debelak and Martin Arendasy
Educational and Psychological Measurement
The proposed method allowing the identification of item scales is based on hierarchical cluster analysis and constructs clusters of items which show a good fit to the unidimensional Rasch model in a multidimensional item set. The simulated and real data were used to evaluate the model. The structure of the model and the statistic of model fit were introduced in a simple way.
The described method could serve as an alternative to other approaches for the assessment of unidimensionality in the context of Rasch measurement. If an item set fits the Rasch model, it is to be expected that the suggested procedure assigns all items to a single cluster.
The method is also compared with the application of a principal components analysis and parallel analysis based on tetrachoric correlations.
The results show that the model can provide practically usable results under many circumstances but generally requires large person samples.
It seems that the method is benefit for researchers to test unidimensionality except some limitions.
Questions:
1 The procedure of the method the new method is not very clear. Why O3? Is it for Q3 or just an assumption? How can we use this procedure to conduct our own study?
2 There is too few explanation of the results in the two tables. How to understand.