教學反應模式在國中閱讀障礙鑑定系統之建構

Abstract

以教學反應模式（RTI）做為學習障礙鑑定工具的議題下，第二層次的介入尤其重要，因為必須兼具補救與診斷的功能，除了選擇適當的教學模式進行重要教學成分之有效教學之外，更重要的是必須能決定適當的教學反應標準做為資格鑑定的參考。 本研究主要研究目的有二，係透過執行國語文低成就學生在接受第二層級的語文補救教學方案過程時定期監控學生在語文相關能力的變化情形，先探究第二層級教學反應標準對於鑑別「無教學反應者」（Non-Responder，NR）的可行性，再檢驗第二層級教學反應標準對於研判「閱讀障礙」（Reading Disabilities，RD）的有效性。 研究樣本資料取自國立臺灣師範大學教育評鑑與研究中心語文科補救教學課程之部份資料，並依研究目的不同分別為完整參與補救教學課程的70名實驗組學生及經過家長同意接受閱讀障礙綜合研判之53名學生等兩群人。實驗組學生接受研究者提供為期一年小組形式、每週2-3次、共計44節之閱讀理解補救教學方案，該方案屬於標準計畫取向之層次二補救教學，教學者全程接受課程督導以控制教學執行度。全體學生在一年中接受5次已建有複本之詞彙成長測驗、閱讀理解成長測驗，所有測驗分數以Embreston’s multidimentsional trait model 估計出個人能力值(θ，theta)，研究者得以比較及監控群體語文相關能力改變情形。 本研究對於無教學反應者的參考標準係根據過去研究文獻建議之閱讀成就表現低落程度，各種教學反應標準之設計亦是採用研究常用的「能力參照標準」、「界定教學反應的方法」和「切截點」等操作原則，閱讀障礙資格之確認則是採用國內學者柯華葳、洪儷瑜等(2007)建議的研判標準，由受過專業訓練且得到認證之心評人員實施，並取得專家審查通過。主要分析方法為KAPPA一致性分析、分類分析法等相關考驗。 主要研究發現如下：（一）國中階段的教學反應標準，應該同時考慮「詞彙」及「閱讀理解」兩種語文能力做為參照指標，採用同時評估「最後表現水準」及「成長速率」的「雙重差距標準」，為求方便解釋及使用，建議可優先採用常模參照取向之標準；（二）根據整體分類正確率、敏感性、特異性、陽性預測率及陰性預測率等指標之效度檢驗結果，建議「較佳之教學反應標準」有二：其一為採用「詞彙能力」測驗，並使用「最後表現水準（常模參照-1.0SD）+成長速率（平均成長率-1.0SD）」的雙重差距標準（簡稱標準A）；其二為採用「閱讀理解能力」測驗，並使用「最後表現水準（常模參照-0.5SD）+成長速率（平均成長率-0.5SD）」的雙重差距標準（簡稱標準B）；（三）根據臨床研判效能檢驗結果，研究者建議應綜合運用「較佳的教學反應標準」（也就是考量標準A或標準B落入NR即可），其整體研判正確率為62.26%，敏感性指標為67.86%，特異性指標為56.00%，尚稱理想，且相較於過去之轉介流程設計而言，能夠降低轉介學障鑑定之人數達將近一半的比例，且相對有較高的確認閱讀障礙比例；（四）針對此一階段之偽陽性及偽陰性問題，應可透過轉介機制的後續階段提供其他評估工具來輔助研判及排除其他可能的障礙類別。 基於研究結果和發現，研究者進一步提出在教育實務及未來研究之相關建議。

In the Response to intervention model, which has been conceptualized as a prevention and remediation framework, progress data of students who are receiving Tier 2 interventions is monitored in order to inform instructional practice as well as make decisions about student’s movement between tiers of intervention, and serves as a more valid method of identification of disabilities. The main purposes of this study were to compare different measurements and methods of establishing adequate criteria used to identify non-responders (NR), and to analyze the validity of the results of NR by verifying that the results were consistent with the students who had been determined to have RD using a comprehensive evaluation. The study, conducted from 2011-2012, was comprised of seventh-grade students from eleven schools in two counties in northern Taiwan. These students included 70 disadvantaged, low-achieving students who attended a one-year literacy remedial program, qualified as a standard protocol approach Tier 2 intervention, with more intensive sessions in small groups. Out of the 70, 53 finished the comprehensive evaluation with parents’ consent followed Ko& Hung’s (2007) suggestion for RD identification. Every student was assessed 5 times during the year at 2-month intervals using both word knowledge test (CWKPM, Hung, et al., 2011) and passage reading test (CPRPM, Hung, et al., 2011), and all test scores were transformed by Embreston's multidimensional trait model to estimate personal abilities (θ). Therefore the researcher was able to compare and monitor changes in growth among the students. All possible combinations of cut-points, methods andmeasures surveyed from previous studies were compared, and the analytic approach included Kappa statistic and classification analysis etc. The main findings were as follows: 1) As determining a suitable criteria of response to intervention in secondary school level, selecting dual discrepancy both from word knowledge and reading comprehension abilities were preferred because of its stability and coverage in NR. For an easier understanding and application, norm reference approach could be a nice choice. 2) From the result of classification agreement analysis, two better criteria of response to intervention were suggested. The first one was using dual discrepancy of CWKPM with -1.0SD cutoff (referred to as criteria A), and the second one was using dual discrepancy of CPRPM with -0.5SD cutoff (referred to as criteria B). 3) Based on the clinical validation, the researcher recommended the combined usage of criteria A and B, since this combination produced an acceptable overall success rate (62.26%), and sensitivity of 67.86% was associated with specificity of 56.00%. Furthermore, adding this stage could reduce almost half of the students for RD evaluation compared to the prior referral procedure. 4) Using RTI as a pre-referral stage of RD identification, should be followed by the diagnostic and exclusive criteria to solve the false positive and false negative problems. Based on research findings, the researcher also makes some suggestions for educational practices and future researches.