Abstract
Data of item responses to the Hong Kong Specific Learning Difficulties Behaviour Checklist from 673 Chinese primary grade students were analyzed using the dichotomous Rasch measurement model. Rasch scaling suggested that the data fit the model adequately with a latent dimension of global dyslexic dysfunctioning. Estimates of item attributes and person functioning on the same logit scale allowed the locations of person functioning to be interpreted in terms of item content. The substantial correlation between person estimates and external criterion measures (literacy and cognitive skills) provided support for the validity of the checklist in identifying children at risk for dyslexia. A common cutoff score based on person estimates across grades is suggested to replace the different cutoff scores based on conventional test scores for different grades. Implications of the findings for the use and the future development or refinement of the checklist are discussed.
Acknowledgements
The study was supported in part by a grant from The Hong Kong Jockey Club Charities Trust as a research project of the ‘Read and Write: A Jockey Club Learning Support Network’.