以「魏氏兒童智力量表」(第四版)的四因素為基礎之簡式版本分析：臨床應用指引

Development of Factor-Based WISC-IV Tetrads: A Guild to Clinical Practice

本研究根據臺灣「魏氏兒童智力量表」（第四版）之標準化樣本，共968位6至16歲兒童，檢視以四因素為基礎的簡式版本。文中考量心理計量特性、施測省時便易性，以及臨床意義性等多項因素，並以線性轉換法分析36組估計智商之適用性。結果顯示，「類同─矩陣推理─數字序列─符號尋找」與「類同─矩陣推理─記憶廣度─符號尋找」二組合是相對較佳選擇，施測時間約在21至23分鐘間。此外，「詞彙─矩陣推理─記憶廣度─符號尋找」、「詞彙─圖形設計─數字序列─符號尋找」，以及「類同─圖形設計─記憶廣度─符號尋找」亦是可選用的組合。此五組合與實際智商的整組平均差異在0.01分內，相關在.92至.93間，適用於團體平均智商研究。唯結果亦發現，僅約68～69%兒童估計智商位於真實智商的95%信賴區間內，對智商120分以上兒童更普遍有低估傾向。此外，根據實際智商或估計智商進行的排名亦有顯著差異，排名差異絕對值之中位數約在60至70名間，即使是相對較佳的簡式組合仍存有相當之估計誤差。臨床工作者務必注意簡式版本應僅作為篩選或團體研究用途。如何在省時與效度間達到平衡，及如何正確使用和解釋簡式版本估計智商，將持續是使用者的中心考量。

Factor-based WISC-IV tetrads were investigated with the Taiwan WISC-IV standardization sample of 968 children， ages 6 to 16 years old. Various psychometric characteristics， clinical utilities， as well as time constraints were compared among 36 tetrads. By linear equating procedure， current results identified five tetrads with relatively better overall quality. Both the ＂Similarities-Matrix Reasoning-Letter Number Sequencing-Symbol Search＂ and ＂Similarities-Matrix Reasoning-Digit Span-Symbol Search＂ forms exceeded others with respect to validity， content representation， and time saving. The testing time estimates were between 21 to 23 minutes. Three other forms， ＂Vocabulary-Matrix Reasoning-Digit Span-Symbol Search＂; ＂Vocabulary-Block Design-Letter Number Sequencing-Symbol Search＂， and ＂Similarities-Block Design-Digit Span - Symbol Search＂ were also considered as optimal options. For all the above five selected tetrads， the mean differences between estimated and actual FSIQ were within 0.01 IQ points， and the correlation were between .92 to .93. They were appropriate for group study. Nonetheless， results also revealed that merely 68-69% of the sample getting their estimated FSIQ within the 95% confidence interval of the true FSIQ. Besides， rankings based on estimated FSIQ were found differ significantly from empirical rankings. The median for ranking difference was around 60 to 70. These factor-based tetrads tend to show under-estimation for children with FSIQ higher than 120. Users are cautioned to use these short forms for screening， and group research purposes only.