Richman David M, Barnard-Brak Lucy, Grubb Laura, Bosch Amanda, Abby Layla
Texas Tech University.
J Appl Behav Anal. 2015 Spring;48(1):131-52. doi: 10.1002/jaba.189.
A meta-analysis of noncontingent reinforcement (NCR) outcomes was conducted using hierarchical linear modeling (a) to document the effect size for decreasing problem behavior, (b) to compare effect sizes for NCR using functional reinforcers and nonfunctional reinforcers, and (c) to document the influence of schedule thinning on effect size. Analyses were conducted with data from 55 studies and 91 participants. Results indicate that NCR was associated with a very strong effect size (d =-1.58) for reduction of problem behavior, functional reinforcers were slightly more effective than nonfunctional reinforcers, and schedule thinning resulted in minor degradation of effect size. Meta-analysis of single-case design data provides a method to quantitatively estimate effect sizes of interventions across participants. Therefore, it allows one to identify important variables that are not otherwise evident in single-case data, helps to disseminate findings to the broader scientific community, and contributes to the documentation of empirically supported interventions.
使用分层线性模型对非偶然性强化（NCR）结果进行了一项荟萃分析，目的如下：（a）记录减少问题行为的效应大小；（b）比较使用功能性强化物和非功能性强化物的NCR的效应大小；（c）记录时间表稀疏对效应大小的影响。分析采用了来自55项研究和91名参与者的数据。结果表明，NCR与减少问题行为的非常强的效应大小（d = -1.58）相关，功能性强化物比非功能性强化物略有效，时间表稀疏导致效应大小略有下降。单案例设计数据的荟萃分析提供了一种定量估计跨参与者干预效应大小的方法。因此，它使人们能够识别单案例数据中不明显的重要变量，有助于将研究结果传播给更广泛的科学界，并有助于记录经实证支持的干预措施。
