Background: The purpose of the study is to develop a standardized diagnostic tool designed to predict the level of the tested floorball skills in young school-age children that is necessary for future game performance. Material and methods: For the construction of the Guttman-type scale, the Rasch model was applied.
The methodology employed the procedures for standardization by Stochl & Musalek, fit functions to determine the fit of the data model, KR-20 coefficient for the reliability calculation, Fleiss' kappa coefficient to determine the inter-rater agreement, and PCA of residuals to determine the unidimensionality. Results: Only 9 items out of a total of 30 were selected and retained in the developed rating scale.
However, the items covered the continuity of the diagnosed feature very well, and the standardization procedure has been successful - the Rasch model fit the data, three criteria of unidimensionality were met, the reliability value of the rating scale was 0.81 and the inter-rater agreement reached 98.5%. Conclusions: The developed rating scale includes 9 items suited to assess ball handling, ball controlling and passing techniques.
Unfortunately, items containing shooting were not selected; they were too difficult and misfit the Rasch model.