運用教學反應結果作為國中閱讀障礙學生鑑定研判標準之試探性研究

Respose-To-Intervention as a Criterion for Identifying Reading Disability in Secondary Schools: A Pilot Study from 7th -Grade Remediation

本研究主要目的在評估將教學反應模式（Respose-To-Intervention Model， RTI） 的第二層級介入納入閱讀障礙鑑定系統的可行性。臺灣學障鑑定基準「介入難有 效改善」的要求即是以第二層級的補救教學反應結果作為鑑定研判標準的概念， 如何合理地訂定「無教學反應者」（Non-Responder， NR），評量標準是相當重要 的議題。研究者先比較採用不同測量工具、界定方法及切截點所組成之各種教學 反應標準對於鑑別出「無教學反應者」的差異情形，再透過現行閱讀障礙綜合研 判結果為效標，檢驗所設定教學反應標準的臨床效度。研究樣本為臺灣北區兩縣 市70 名接受一年閱讀理解補救教學方案並接受五次語文能力追蹤的國中七年級學 生，其中53 人於方案結束後同意接受閱讀障礙鑑定。研究者運用KAPPA 一致性 分析、二元邏輯斯迴歸、分類分析等方法進行資料分析。研究結果顯示，無論是 使用詞彙或閱讀理解測驗，均建議採用「最終水準」及「成長斜率」之雙重差距 標準，較能穩定選出NR 出現比例，且因測驗間研判一致性低，故應考慮同時使 用兩項測驗。迴歸及分類預測結果指出，在控制智力及前測後，納入詞彙及閱讀 理解測驗成長斜率或最終水準之模式能小幅提升模式解釋力，且獲致較佳的敏感 度（86%），惟受限於樣本數少、國中生語文能力成長幅度較小，故整體而言研 判效能提升效果並不顯著。研究者最後針對結果提出討論及相關建議。

Purpose: This study was aimed to appply the respose-to-intervention (RTI) model into the process of identifying students with reading disabilities (RD) in Taiwan. The main purpose was to compare the different measurements and methods in order to establish adequate responses to identify nonresponders (NRs)， and then to determine the types of data used as a criterion of RTI. Method: A total of 70 seventh graders from northern Taiwan participated in a 1-year Chinese literacy remedial program. They were assessed five times during the year at 2-month intervals through both word reading (WR) and passage reading (PR) tests. After the remediation， 53 of the participants were refered to RD identification with consent. Kappa consistency analysis， logistic regression statistics， and classification accuracy were analyzed. Results: The main findings were as follows: 1) The results of the WR and PR tests both suggested that the dual disapancy criterion， which combines the use of the final status and growth slope of the students’ performance， was the most stable means for screening a reasonable percentage of NRs. Howewer， of a possible 91 combinations from the different data， less than half the comparisons yielded a kappa value of ≧ 0.40; moreover， the percentage decreased with a more stringent cut-point. Consistency among the methods varied considerably: higher kappa values were driven by methods derived from the same test， as well as by similar cut-points. These consequences of comparison demonstrated the heterogeneous reading development of students at the secondary school level， so both the WR and PR assessments simultaneously were recommended for the decision of RTI. 2)The results of the logistic regression analysis and classification analysis indicate that the two proposed models that added the growth slope or posttest data of WR and PR respectively and controlled the index scores of the IQ tests and pretests of themselves， both showed adequate sensitivity (SEN) of classification. A SEN value of 86% indicated that only 14% of NR who were identified as RD could not be detected after the remedial program. Therefore， WR and PR were appropriate indicators at the screening stage of identification. However， the outcome could not be comfirmed as statistically significant predictors because of the small sample and limitation of the growth of reading abilities in these students. Conclusions: Therefore， the overall results yielded supporting evidence to RTI. The final status of students’ performance was recommended as the criteria of RTI compared with the growth slope. Other recommendations for further applications and studies are made.