Date of Award

12-2019

Degree Name

Doctor of Philosophy

Department

Psychology

First Advisor

Dr. Bradley E. Huitema

Second Advisor

Dr. Alyce M. Dickinson

Third Advisor

Dr. Heather M. McGee

Fourth Advisor

Dr. Joseph W. McKean

Keywords

Single-case design, statistical analysis, visual analysis

Abstract

Visual analysis is the favored method for analyzing single-subject research data in more traditional behavior analysis programs of study; however, some critics of the practice argue that it should not be the only method used to analyze data given the previous research demonstrating significant inconsistencies in conclusions drawn based on the practice (Bulté & Onghena, 2012; Campbell & Herzinger, 2010; Kazdin, 2011). Supporters of statistical analysis have argued that although statistical analyses are not free from subjectivity and bias, the decision-making processes and criteria are more straightforward than those offered by the complexities involved in visual analysis (Kazdin, 2011). It is therefore of interest to identify a statistical analysis that parallels the procedural steps of visual analysis to add to the credibility and generalizability of single-case design research results. The key to complementing visual analysis results with a statistical analysis is in choosing an appropriate analysis that intends to answer the same questions as that of the visual analysis (Huitema, 2004; Huitema, McKean, & Laraway, 2008). The purpose of the current research study was to identify ways to improve upon the consistency of results produced through visual analysis through training and the use of a checklist, identify specific characteristics of the data and study designs that may influence consistency in results produced through visual analysis, and determine the relationship between the results of a specific type of statistical analysis and results provided by trained visual analysts. A randomized group design was used to compare the consistency of visual analysis results produced by participants randomly assigned to one the following three conditions: (a) training and assessment with checklist; (b) training and assessment with no checklist; and (c) control group (no training and no checklist). The results of the study indicate that training increases consistency of visual analysis results produced by novice raters, but that no additional benefit was demonstrated by adding the checklist. The results also demonstrate that graph type significantly affects consistency of ratings, and that the H-M regression analysis supports the results produced through visual analysis. Despite the limitations associated with the current research study, the findings shed light on many future areas of research that will continue to aid in increasing the credibility and generalizability of single-subject research results and visual analysis as a way to draw sound conclusions.

Access Setting

Dissertation-Campus Only

Restricted to Campus until

12-2020

Share

COinS