身心障礙學生功能本位行為介入方案成效之後設分析

The Effects of Function-Based Behavioral Interventions for Students with Disabilities: A Meta-Analysis

本研究分析2001 至2012 年間，國內39 篇身心障礙學生的功能本位行為介入 方案之單一受試研究現況，並以超越中數比率（percentage of data points wxceeding the median of baseline， PEM）後設分析行為功能、介入策略種類與策略內容對不同 行為問題種類與正向行為之介入成效。研究結果顯示，使用功能本位行為介入方 案，對減少身心障礙學生的行為問題及提升正向行為的整體介入成效介於稍有效 果到高度有效之間，各研究間的成效差異落差頗大。分析國內研究介入之行為問 題種類發現，應用功能本位之行為介入方案能夠改善各類別的行為問題，其中又 以改善固著行為與提升可取代固著行為的正向行為介入成果最佳。在行為功能方 面，結果顯示，不同功能的行為問題與正向行為的介入為中度到高度有效，且不 因行為功能的不同而有顯著差異。在介入策略種類方面，發現不同類別的介入策 略對行為問題的介入成果介於中度到高度，而對正向行為的介入介於稍有到高度 有效，各組成效落差頗大。就介入策略內容而論，不同於前事策略對行為問題與 正向行為的介入皆有中度效果，教導策略與後果策略的介入則介於稍有到高度有效，各種前事策略與後果策略的介入成效並未達顯著差異。

Purpose: Function-based behavioral interventions have become widely used for decreasing challenging behaviors and increasing positive behaviors of students with disabilities. The purpose of this study is to meta-analyze the effects of function-based behavioral interventions on behaviors of students with disabilities in Taiwan. Method: To identify studies， the authors searched electronic databases such as the Index to Taiwan Periodical Literature System and the Chinese Electronic Periodical Services (CEPS)， and manually searched the journals listed on the Special Education Transmit Net and the reference lists of the studies obtained from the aforementioned sources. The inclusion criteria used to select studies include: (a) using single-subject designs， (b) providing data in graphs and/or conducting visual analysis， (c) including functional behavior assessment and/or function-based interventions as independent variables， (d) involving students with disabilities. A total of 39 single-subject research studies between 2001 and 2012 were identified， producing 27 to 59 effect sizes according to the dependent variable. The metaanalytic study utilized the percentage of data points exceeding the median of baseline (PEM) approach to calculate the effect sizes of each study. The PEM effect sizes were analyzed across a variety of variables， including types of behavior functions and intervention strategies. H test and U test were then applied to compare the differences of intention effects among the subgroups under each variable. Results/Findings: Results indicated that function-based behavioral interventions overall had moderate to large effects on decreasing challenging behavior and increasing positive behavior of students with disabili ties. Function-based behavioral interventions had positive effects on various types of challenging and positive behaviors， among which the effects were greater for stereotypical behaviors and those positive behaviors replacing stereotypical behaviors. Functionbased behavioral interventions also showed medium to large effects on various types of behavioral functions; however， no significant difference was shown among the effect sizes on different types of behavioral functions. Regarding different types of intervention strategies， the results of PEM analysis indicated that medium to large effects were yielded for challenging behavior， while mild to large effects were found for positive behavior. Finally， all types of antecedent strategies had medium to large effects on both challenging and positive behaviors while instructional and consequence strategies showed mild to large effects. Differences among effects of different types of antecedent strategies and of consequence strategies were not significant. Conclusions/Implications: While the results showed that function-based behavior interventions had promising effects on both challenging and positive behaviors of students with disabilities， about half of the studies did not report the changes of positive behaviors of their participants. It is suggested that future studies report the changes of both challenging and positive behaviors of the participants and document the effects of intervention across students with various disabilities， different types of intervention strategies， and different functional behavior assessment methods. Future meta-analytic studies can consider to use a different method to calculate the effect sizes and to group the studies under each variable.