Article

OBJECTIVE: To review the development of Rasch analysis by examining the history
of its application to the Functional Independence Measure (FIMTM), and
highlighting current issues in the approach. METHODS: All Rasch-based papers
concerning the FIMTM were reviewed for their analytical strategy and results.
Four analytical pathways were identified that accommodated the majority of these
strategies. Data derived from secondary analysis of 340 in-patients undergoing
rehabilitation following stroke, measured on the FIMTM Motor Scale, were fitted
to the Rasch measurement model according to these 4 pathways, with 2 additional
pathways to accommodate recent developments. RESULTS: In the analytical pathway,
where items are not re-scored, the fit to the Partial Credit parameterization was
better than the Rating Scale version. Fit improved following re-scoring of
disordered thresholds. When local dependency was accommodated by 4 testlets, the
Partial Credit, re-scored testlet version achieved adequate summary fit with no
misfit among items, and unidimensionality. All other pathways required item
deletion. CONCLUSION: The current study has shown that the FIMTM Motor Scale, as
applied to a stroke rehabilitation sample, satisfies Rasch model expectations and
the unidimensionality assumptions, having accommodated local dependency issues,
and by using the partial credit parameterization with re-scored categories. Other
analytical pathways gave less ideal solutions, and are consistent with the wide
range of solutions found for the scale over the years. Consequently, the
development of the Rasch approach in health outcomes can be traced in the history
of analysis of the FIMTM, and that development continues to this day.

Langue : ANGLAIS