PC – ABAB Reversal Graph

In an ABAB Reversal design, an experimenter rotates two or more conditions and has a participant complete several consecutive sessions in each condition. Typically, an experimenter rotates baseline and intervention conditions. This design is useful for demonstrating functional relations with performance behaviors. To demonstrate control in this design, one must observe different levels of behavior in the respective conditions.

SCROLL DOWN TO LEARN MORE ABOUT THE DESIGN

Video Instruction

Need to know how to set up the instruction?

 

More Information about Reversal Designs

 

This design is used to demonstrate all aspects of baseline logic (predictability, verification, replication) within the same participant. The experimenter begins with baseline and continues observing behavior until there is steady state responding, which allows one to predict that the behavior should remain the same if there is no change to the environment.  Now, the experimenter can implement the intervention and if behavior changes, affirm the consequent – that something other than behavior remaining the same occurred.  Once behavior becomes stable during intervention, the experimenter can repeat a baseline condition.  If behavior returns to the same level as the initial baseline, the experimenter has demonstrated verification of the baseline prediction.  That is, the experimenter gathered evidence that the initial prediction was a good prediction.  Last, the experimenter repeats intervention.  If behavior changes again, the experimenter has replicated the change in behavior (dependent behavior) when changing from baseline to intervention (independent variable).

One should not use a reversal design with behaviors that are not reversible.  For instance, if you teach a person to fish, that individual is not likely to forget how to fish, making a reversal design a poor choice for evaluating your fishing intervention. Control is demonstrated in this design when behavior returns to baseline levels and when one replicates the intervention effects.  If either of these do not occur, once has not demonstrated a functional relation.

Examples

Porterfield, Herbet-Jackson, and Risley (1976) used a reversal design to compare the effects of contingent observation to the effects of redirection with preschoolers.  The researchers found that redirection produced higher rates of disruption than contingent observation.

Krentz, Miltenberger, and Valbuena (2016) compared baseline and token-reinforcement conditions and effects on distance walking with adults with intellectual disabilities. The researchers found that the participants walked considerably more laps around a track when each lab resulted in a token that could be exchanged for a backup reinforcer.