The purpose of this study was to develop criterion-referenced measurement items for play analysis of
top players in the game of tennis. The sample size was 752 shots, with 157 points, at a five-set match in the
finals of the 2014 US Open Championship. The tactical play structure and the measurement items, with
criteria, were constructed using qualitative causal-effect analysis by the Delphi technique, with the
assistance of three tennis experts. The measurement item criteria were extracted using decision tree
analysis with a CART algorithm. The characteristics of the criterion-referenced measurement items and
test were statistically analyzed using item response theory (IRT) analysis with the two-parameter logistic
model (2PLM). The resulting 26 criterion-referenced measurement items of the three play phases of serve,
return, and rally were selected to be measured. In the three phases of tennis play, the play items were
found to have local independence, unidimensionality and conformity to the 2PLM, as well as invariance of
the item difficulty, item discrimination and ability values. The tennis play items constructed using the
qualitative analysis conformed to the ICC of the 2PLM and constituted an achievement rating scale for
play in the game of tennis. In conclusion, firstly, the tennis play of top players was divided into three main
phases: the serve phase, the return phase and the rally phase. Second, the test items for assessing
achievement in a tennis game were found to be unidimensional, fit to the 2PLM, and invariant to the
estimates in each phase of play. Third, the achievement assessment tests of play in the game of tennis were
found to be reliable, valid and fit for each phase of the test. Finally, the criterion for assessing the degree of
achievement in tennis, consisting of item difficulty, was found to be valid.
抄録全体を表示