Single-Case Research: Documenting Evidence-based Practice Rob Horner University of Oregon
Main Message For Applied Behavior Analysis to achieve wide spread adoption we need clear procedures by which single-case research can be used to document “evidence-based practices.” Acknowledgements: Dr. Edward Carr Dr. James Halle Dr. Gail McGee Dr. Sam Odom Dr. Mark Wolery “The use of single-subject research to identify evidence-based practices in special education” 2005 Exceptional Children, 71 (2)
The roles of Single-case research Using Single-case research to examine behavioral principles Using Single-case research to document effective interventions. Using Single-case research to document evidence-based practices
A Challenge Federal and state governments are being required/encouraged to allocate funding only for practices that are “evidence-based.” The current standards for documenting a practice as “evidence-based”: Are being managed by people who do not have experience with Single-case technology Require use of a randomized control trials (RCT)
The Need Build a set of unified, professional standards Clarity in defining a “practice” Criteria for documenting experimental control Criteria for designating a practice as “evidence- based” Meta-analysis technology Collectively promote those standards
Clarity in defining a “Practice” A “practice” is a procedure, or set of procedures, designed for use in a specific context, by individuals with certain skills/features, to produce specific outcomes for specific individuals.
Clarity in defining a “Practice” Define a Practice by stipulating: Procedures Who uses the procedures Contexts where procedures are appropriate Population(s) for whom the procedures are intended Specific outcomes expected
Clarity in defining a “Practice” Issue Size of practice (principle; single-component intervention; multi- component intervention) ABA Practices (?): Precision TeachingSelf-delivered reinforcement Time OutPeer tutoring Toilet trainingDiscrete Trial Training Non-contingent reinfDirect Instruction Positive reinforcementTime-Delay Functional Communication Training Acceptance and Commitment Therapy Applied Behavior AnalysisPositive Behavior Support
Criteria for documenting experimental control within Single-case designs To assess an “effect” between two phases compare: Level, Trend, Variability, Overlap, Immediacy of effect, similarity of effect for similar phases. -- Parsonson & Baer (1978) To assess if “effects” constitutes experimental control Three demonstrations of the effect at three points in time Need for Training Materials Wayne Fisher (Munroe-Meyer Institute) Fisher Model Examples Training Graphs
Criteria for Defining a “Practice” as Evidence-based At least 5 peer-reviewed studies Practice is defined Fidelity is documented Experimental control is documented Replication of experimental control across at least three contexts/researchers At least 20 participants documenting effect
Example of using Single-Case Research to document Evidence-based Practice Functional Communication Training Procedure Functional behavioral assessment/ functional analysis Instruction on a functionally equivalent skill Used in schools/home/community Used by teachers, psychologists, interventionists Used with individuals with typical abilities to severe disabilities who engage in problem behavior Outcomes: Reduction in problem behavior and increase in use of functionally equivalent skill.
Example of using Single-Case Research to document Evidence-based Practice Functional Communication Training At least Five Studies Bird, Dores, Moniz & Robinson (1989) Brown et al., (2000) Carr & Durand (1985) Durand & Carr (1987) Hagopian, Fisher, Sullivan, Acquisto & LeBlanc (1998) Mildon, Moore, & Dixon (2004) Wacker et al., (1990) At least three locations/scholars Six locations, Research Groups At least 20 participants 42 different participants document effect
Adequacy of Proposed Criteria Identify a “practice” that you believe meets these criteria? To what extent do these criteria meet the needs of the field?
Standards for Statistical Analysis Analysis of Individual Studies Time series models, HLM/SEM, non-parametric models Statistical models exist for assessing if there is an effect between two phases Current statistical models DO NOT account for whole design Similarity of effect in similar phases Three demonstrations of effect at three points in time Meta-Analysis of Multiple Single-Case Studies Documentation of effect size HLM Janet Marquis Will Shadish Tom Kratochwill
Summary Single-case designs will not be used to document evidence- based practices unless Behavior Analysts propose a set of “accepted professional standards” We need professional agreement We need to promote the standards across non-ABA audiences Federal agencies Mental Health, Psychology, Education Slides available at