Assessing the Characteristics of Published ABAB Single-Subject Research Graphs

A key components of single-subject research (SSR) designs (see a recent blog post for an overview of common single-subject designs used in ABA and education research) is the visual presentation of data. Visual analysis of behavioral changes within and between conditions are the key analytic procedure when inspecting data presented in time-series or line graphs.

Visual analyses should focus on four key areas: (a) the number of data points per condition (3 or more minimum, but more is always better); (b) level, typically reported as a number (mean or median) and words; (c) trend, reported via a qualitative descriptor (e.g., increasing, positive, zero or flat, etc.); and (d) variability, or the amount of change within a data path, also typically described via a word (e.g., high, medium, moderately-low, etc.). While many SSR designs include requisite components, few empirical guidelines on graph structure and design exist.

To better understand how ABA researchers presented data in experiments relying on a reversal (or ABAB) design, our research team conducted a scoping review of the 2007 and 2012 volumes of 6 ABA journals (Behavior Modification, Journal of Applied Behavior Analysis, Education and Treatment of Children, Journal of Behavioral Education, The Analysis of Verbal Behavior, and European Journal of Behavior Analysis). The research team reviewed 176 graphs in 45 published behavior analytic articles against a set of visual analysis components (e.g., how presented/described number of data point, level, variability, and trend).

Results from our review were rather bleak – in all, manuscripts contained just over 7% of the possible analyses. Less that 1% of the reviewed graphs/articles included description of the number of data points within conditions. Almost 3% presented numbers and/or words regarding trend and less (2%) described data variability. Numeral and/or word descriptions of data level were only provided for nearly 23% of possible opportunities.

In all, a clear majority of graphs failed to contain even 1 within condition analysis. Manuscript authors failed to conduct a within condition analysis tactic 93% of the time. Additionally, when those within condition analyses were conducted, there was variability from existing guiding literature and/or ‘best practice.’ Without a proper (i.e., excluded and/or incorrect) visual analysis, researchers dramatically increase the likelihood of making type 1 (false positive result) and type 2 (false negative result) errors.

You can access the slides for our presentation at the Applied Behavior Analysis International Conference HERE.

More information about this topic and professional support is available – reach out to Justin with questions and requests! He looks forward to talking with you.

Dr. Douglas E. Kostewicz is an Associate Professor of Special Education and a Board Certified Behavior Analyst-D at the University of Pittsburgh. He teaches courses in special education methods, applied behavior analysis, and single case research. Dr. Kostewicz’s research focuses on the effective use of teacher attention in classroom, graphical construction, and the effective use of academic and behavioral intervention for students with disabilities. You can find more information about Dr. Kostewicz, including his CV, HERE.

Dr. Kubina is a Professor of Special Education at the Pennsylvania State University. Dr. Kubina is a Board Certified Behavior Analyst, Doctoral level and serves on a number of editorial boards for behavioral and special education journals. He was the editor of the Journal of Precision Teaching and Celeration. Currently, Dr. Kubina teaches courses on methods for teaching reading, behavior analysis, and single-case design. Dr. Kubina conducts wide-ranging research in the area of Applied Behavior Analysis, Precision Teaching, and technology. You can find more information about Dr. Kubina HERE.