干预-应答模式鉴别学习障碍的有效性及其调节因素:20年研究的元分析

The Validity of the RTI Model for Identifying Learning Disabilities and the Moderators: A Meta-analysis of the Past Two Decades' Studies

本研究采用元分析方法探讨在学习障碍的鉴别中兴起的干预-应答(Response to Intervention,RTI)模式鉴别学习障碍内部亚组的有效性及其调节因素。通过系统检索1996-2015年的文献,获得了34项研究,包括6127名学生的45个样本、261个效应值。元分析结果表明,RTI模式可有效区分学习障碍风险儿童的内部亚组,对干预无应答和有应答的学生在学业成就、认知技能、行为等多方面存在系统、显著和持久的差异,但仍存在个体应答状态的进一步分化。RTI模式对于学习障碍风险儿童内部变异的区分效果受到干预对象、干预层次、干预时间、应答指标选择、测量方法和切分点等因素的影响。本元分析结果不仅为认识RTI模式鉴别学习障碍的有效性提供了进一步证据,更重要的是通过系列调节效应分析,为合理实施RTI模式鉴别学习障碍提供了直接依据。

Learning disability is one of common neurocognitive developmental disorders among children and adolescents. The mode of Response to intervention （RTI） has been recommended to identify learning disabilities since 2004. However, the validity of the RTI model for identifying learning disabilities has been questioned. This study aimed to exam the validity of the RTI model and the moderating effects of the major variables being involved in the application of RTI model on the basis of the past two decades＇ of studies. Articles were obtained using keywords, i. e. , ＂RTI（ response to intervention） ＂and ＂Dyslexia＂ or ＂Dyscalculia＂ or ＂Specific learning difficulties^disorders＂ or ＂Reading disorder （s） difficulties＂ or ＂Mathematical disorder （s） / difficulties＂ from Web of science, Psychoin- form and CNKI. Thirty-four studies published within 1996 to 2015 met the criteria and were included into the metaanalysis, yielding 45 samples （N = 6,127） and 261 weighted effect sizes （ESs）. Results indicated ： （ 1 ） The non- responders performed significantly poorer on academic achievement, cognitive skills and behaviors than the responders, and the ESs of all the variables were above 0.5. （2） Evidences from six longitudinal studies indicated that the gap between the groups of responders and non-responders were persistent across time on academic achievement and cognitive skills, but some responders may perform similarly to the non-responders at the follow-up assessments. （3） Moderating effects were found significant for the severity of learning difficulty, intervention tier and duration. Significantly larger ESs emerged from the studies about at-risk students, using short-term intensive Tier 2 intervention rather than from those students with severe learning difficulties, using long-term multi-tiered intervention. （4）Moderating effects were also found significant for measures and criteria of responsive status. Academic achievement rather than cognitive skills were more reliable as the indicators of responsiveness. Compared with the growth or the dual discrepancy criteria, the achievement status was more reliable in differentiating responders and non-responders. When standardized, norm-referenced instruments were used for measuring academic achievement, the 16th per- centile would be recommended as the cut-off points. And when non-standardized instruments were used, the 25th percentile would be preferred. Findings from this meta-analysis support the RTI model valid for identifying the subgroups of at risk students for learning disabilities. Guidelines for implementing RTI model in research and practices were recommended. Future directions were also discussed.