中華心理學刊
民95 ， 48 卷， 1 期， 89-114

Shun-Wen Chang（National Taiwan Normal University）

Abstract

This study examined the effects on the BCTEST assessment of employing various formats of examinees’ raw scores in constructing scale scores using the arcsine transformation procedure. The three types, the  numbercorrect, z-score, and proportion-correct score formats were investigated. The resulting scale scores were evaluated with respect to the rawto- scale score conversions, summary statistics, and measurement properties computed based on the strong true score theory. The contributions of the individual tests to the BCTEST Composite, the gaps resulting from the conversions, and the effects of rounding and truncation were also  assessed. The findings indicated that each score format had its advantages and disadvantages with respect to
some of the criteria, but no one format possessed all desired characteristics. Regarding the effective  contributions of the tests to the Composite, the z-score type seemed to be the most promising. But, in terms of the criterion of constant error variance along the scale, the number-correct type, and to a lesser extent,
the proportion-correct type were the choices. Results from this research have offered a better understanding of the effects of the various score formats on the formation of the BCTEST Composite, and have provided valuable
insights into forming composite scores to convey appropriate meanings about examinees’ performance over different tests.

Keywords: scale score, score format, arcsine transformation, composite score, BCTEST