運動能力 테스트의 信賴度 檢査

Abstract

The purpose of this paper was to scrutinize the meaning of the relialility of a motor ability test, especially a norm-referenced test, and the mathmatical theories and statistical methods for estimating reliability. Reliability is a important criterion of a measurement, a number obtained by a test, as known a reliable test measures whatever it measures cousistently. Thus, reliability is defined as the dependability of a measure or the consistency of an individual in performing a test. The theory of true and error scores(classical theory) and the domain smapling theory (generalizability) are the theoretical models to understand relialility. However the classical theory that assumes any measurement on a continuous scale contains an inherent component of test. a true and error score is used to assess the reliability of a motor ability In this concept, the reliability is the ratio of the true score variance to the obtained score variance. There are many methods for estimating reliality based on the characteristics, of a test such as stability reliability coefficient, pearson product momentum correlation, from the test-retest method and internal-consistency reliability, usually for a written test, from the split-half method. Intraclass correlation coefficient is the better method then interclass correlalion coefficient when there is more than 2 scores per person. Also Spearman Brown phrophecy formula is used when the legth of a test, the number of trials in a motor ability test, is increased. Many studies for estimating relialiity of a motor ability test by using various ANOVA models and the other statistical techniques like canonical analysis have been conducted in physical echecation and sports science.