非語文學習障礙篩選量表編製研究

Development of a Screening Checklist for Nonverbal Learning Disabilities

非語文學習障礙（nonverbal learning disabilities, NLD）是學習障礙的一種亞型，由於缺乏相關的評估工具，因而限制了NLD 的研究以及鑑定實務工作。本研究之目的為編製一份「非語文學習障礙篩選量表」（Screening Checklist for Nonverbal Learning Disabilities, SCNLD），並進行信度與效度之檢驗，以期能篩選出NLD 之學生。165 位國小普通班四至六年級學生以及7 位NLD 臨床模擬個案參與SCNLD 量表之編製，953 位普通班學生參與SCNLD 先驗常模之建立，以及32 位經鑑輔會鑑定的學習障礙個案（17 位語文類型與15 位非語文類型）參與SCNLD 效度之分析。SCNLD 原始題目為60 題，由家長填答，經項目分析、因素分析、以及題目間的相關分析後，刪減題目為52 題，包括社會人際、動作協調、以及知覺組織等三個分量表。在信度方面，SCLND 分量表以及全量表之Cronbach α 係數在.88~ .96 之間，在間隔兩個月的重測信度中，SCLND 分量表以及全量表前後測之相關係數位於.74~ .83 之間，內在等級相關係數位於.85~ .90 之間。在效度方面，無論是在SCNLD全量表或各分量表分數的比較，或是達到切截分數的人數比較，非語文類型學障學生均明顯高於語文類型學障學生，同時在「接受者操作特徵」（receiver operating characteristic, ROC）曲線下面積結果也顯示SCLND 對於NLD 具有良好的診斷價值。本研究編製的SCLND 具有滿意的信度與效度，ROC 曲線也支持SCLND 的診斷敏感度以及特異度。

Nonverbal Learning Disabilities (NLD) are one of the subtypes of learning disabilities. The purpose of this study was to develop a “Screening Checklist for Nonverbal Learning Disabilities” (SCNLD) in order to find which students have NLDs. The checklist was designed by analyzing 165 Grade 4-6 normal students and 7 analogue NLD clinical cases. Based on item analysis, factor analysis and correlation analysis, 52 items were selected for a checklist with 3 subscales or dimensions: social-interpersonal, motor-coordination, and perceptual-organization dimensions. In the reliability test, the Cronbach α coefficient was .88 to .96 and the test-retest coefficient was .74 to .83. As for test validity in terms of cut-off score, the normative data was established for 953 Grade 4-6 students, and the cut-off score was 41. 32 students with learning disabilities (17 verbal-type and 15 nonverbal-type) then participated in validation of the SCNLD. The mean score on the SCNLD, as well as the number of students who reached the SCNLD cut-off score, were higher for nonverbal-type learning-disabled students, and the receiver-operated characteristic (ROC) curve indicated that this SCNLD can be a good diagnostic tool. The reliability and validity of the SCNLD were satisfied, and the ROC curve supported the sensitivity and specificity of this Screening Checklist.