ABSTRACT
The primary function of a correlation coefficient in measurement is to describe test reliability and test validity. The three most common types of reliability coefficients are those that are associated with the coefficients of stability, equivalence, and internal consistency. The difference between objective measurement and subjective evaluation for a rating scale is that in objective measurement, each dimension includes at least two, preferably more, clearly stated continuous criteria. When evaluation is used in place of measurement, the reliability and the validity of the rating scale are put in doubt. Except for a rating scale designed for measuring students' performance achievement, a well-designed multiple choice test is the best type of test for measuring students' music achievement as it relates to both discrimination and inference learning. If a criterion-referenced test could make provision for adapting instruction to students' individual musical differences, there would be less practical difference between a criterion-referenced test and a norms-referenced test.
