Single -subject statistical analysis in an applied setting: A comparative investigation of statistical process control and conservative dual criteria procedures.

摘要

Behavior analysis utilizes single-subject research design as the primary mechanism in its experimental methodology. Visual inspection is the method of choice for behavior analysis. Behavior variability is more difficult to control and more evident in the real world compared to an experimental setting. There has been extensive debate regarding the incorporation of statistical methods in evaluating single-subject graphs with visual inspection as the preferred approach in applied behavior analysis. Two judgmental aids that have been proposed are Statistical Process Control (SPC) and Conservative Dual Criteria procedures (CDC). SPC has a lengthier history and arose out of industry's attempt to increase productivity and control variation. CDC arose out of behavior analytic literature and compared with other methods it has been shown as a potentially useful judgmental aid to complement visual inspection. This dissertation compared SPC and CDC procedures applied to the same data sets. The percent agreement obtained between SPC and CDC procedures in this study was 54%. The CDC procedure is a more conservative approach and the results of this study support Fisher and colleagues' claims for the use of CDC. Results also suggest that SPC and CDC procedures could be considered valuable tools for clinicians. These may be useful compliments to visual analysis when conveying data to consumers such as third party funding sources, clients, families, direct care staff, and agency administrators. The implications for the practicing psychologist are discussed.