代幣制與增強物之效果:以`超越中數線率`法對個案實驗研究所作後設分析(II)

Abstract

本研究旨在以「超越中數線率」(PEM)計算各種增強物之平均效應量。本專題為兩年期計劃。它為應用「超越中數線率」的方法對代幣制及增強物之效果作後設分析。第一年將聚焦於代幣制之效果。第二年則集中在原始增強物之效果，並且比較代幣制與原始性增強物之相對效果。樣本論文之取得為透過電腦資料庫之查尋，包括EBSCOhost、 ERIC、及 ProQuest。 第一?研究所用之關鍵字為 \"token economy\" 或 \"token system\"。第二研究用之關鍵字為 \"reinforcement\" 或\"reinforcer\"。另外對紙本期刊，如 Journal of Applied Behavior Analysis; Behavior Disorders; Behavior Modification; Behavior Assessment; Behavior Therapy; Behavior, Research and Therapy, and Journal of Special Education等亦將作地?式搜尋。兩年之研究亦將比較「超越中數線量」與「不重疊率」之效度。效度為兩者與原作者對實驗處理效果之判斷之Spearman等級相關係數。

The present study intends to calculate the average effect sizes of different reinforcers by means of 「Percentage of data points exceeding the median」 (PEM). This proposal is a two-year project applying PEM approach to synthetically analyze the overall effectiveness of token and reinforcers. The research in the first year will focus on the meta-analysis of the effectiveness of token system (token economy). The study in the second year will concentrate on the effectiveness of primary reinforcers and then compare the relative effectiveness of token and primary reinforcers. The single-subject studies on token economy used in this synthesis will be obtained through a computer-assisted search of the relevant databases, including EBSCOhost, ERIC, and ProQuest. Descriptors will include token economy or token system for the first year and reinforcement or reinforcer for the second year. A hand search of relevant behavior analysis journals such as Journal of Applied Behavior Analysis; Behavior Disorders; Behavior Modification; Behavior Assessment; Behavior Therapy; Behavior, Research and Therapy, and Journal of Special Education will be conducted. Further articles will be traced from references listed in the collected articles. The Spearman』s rank correlation between PEM or PND with the decision of original author(s) over the effectiveness of treatment will be used as criterion of validity. The validity of PEM and 「the percentage of non-overlapping data」 (PND) approach will be calculated separately and compared.