本研究旨在估計台灣魏氏兒童智力量表第三版(Wechsler Intelligence Scale for Children, the 3rd edition, WISC-Ⅲ)各差異分數之信度,並且分析相關計量限制。本研究根據WISC-Ⅲ,100位6至16歲台灣常模樣本為分析對象,主要有下列四點發現:(1)WISC-Ⅲ各差異分數中,以智商指數間差異分數的信度最高,平均在.80上下;(2)兩兩分測驗差異分數的信度多在.60-.80之間;(3)單一分測驗與一組分測驗平均差異分數的信度相對最低,若組合包含8個以上分測驗,信度中數在.60以上,若組合包含4-6個分測驗,則信度中數介於.45-.60之間;(4)就個別分數言,包含語文智商、作業智商、語文理解指數、知覺組織指數、符號替代、記憶廣度或詞彙等分數的差異值普遍有相對較高的信度,但包含專心注意指數或物型配置分測驗之差異分數信度則普遍較低。文中討論相關限制並對後續應用提出建議。
The purpose of this study was to investigate the reliabilities of various WISC-Ⅲ discrepancy scores, also discussed were the psychometric limitations built in these reliability estimations. The sample used was the WISC-Ⅲ Taiwan standardization sample. A total of 1, 100 children, ages from 6 to 16, were included in this study. Four major findings are: (1) IQ and index discrepancy scores shows the best reliabilities, roughtly with an average of .80; (2) majority of the reliabilities of subtest discrepancy scores are around .60-.80; (3) when considering the discrepancies between a subtest and the mean of a group of subtests, discrepancy reliabilities increased with the number of subtests in the group. When there are 8 or above subtests in this combination, reliabilities roughly are above .60; however, when the number of subtests is less than 6, reliabilities go down to .45-.60; (4) Discrepancy scores which relate to VIQ, PTQ, verbal comprehension index, perceptual organization index, coding, digit span, or vocabulary subtest relatively show higher reliabilities; on the other hand, reliabilities related to freedom from distracbility index and object assembly are lower, comparatively. Suggestions for future researches are discussed.