Based on a collapsing procedure and Rasch analysis, a Rasch-based optimal categorization procedure has been introduced for the determination of the categorization of a test or scale. As a result of the Rasch-based optimal categorization, the number of response categories could be reduced, which may be a threat to the internal structure of a measure. Using three available data sets (two data sets from one study with N = 480 and third set from another study with N = 219), this study examined the differences among the structures of exercise barrier scale constructs when a different number of response categories is used. Specifically, two models of exercise barrier constructs were compared using the structural equation modeling. The results suggest that the collapsing of categories has no effect on the structure of the latent variables. In addition, the results suggest that the collapsed num- ber of categories provides a slightly better model-data fit statistics. Two consequences for the no-impact finding are: (a) a better categorization may help eliminate systematic error related to response categories and (b) the range of ability, or between-subject variance, was still maintained. More studies are needed to determine these possible explanations’ contributions. The analysis of the internal structure illustrated in this study should be a part of the Rasch-based optimal categorization procedure.

Charles University Prague Faculty of Physical Education and Sport Department of Kinanthropology

University of Illinois at Urbana Champaign Department of Kinesiology and Community Health

Vliv katetrizace na vnitřní strukturu dotazníku exercise barrier inventory

Obsahuje 4 tabulky

Literatura