Changing Criterion Designs

The changing criterion design is a type of singlesubject research design that may be used as either an experimental or a quasi-experimental design, depending on the amount of experimenter control. Typically, changing criterion designs are used to evaluate the effects of instructional programs in building fluency or affecting rate of responding over time. Fluency evaluation with academic facts, the speed with which a task is performed, and the number of tasks that are completed in a set period of time are examples of behaviors for which this design would be appropriate. By monitoring gradual changes in performance over time, experimental control is demonstrated as stepwise changes in criterion predict concurrent changes in performance. As a quasi-experimental research design, the changing criteria design may be considered as a series of repeated AB (A = baseline conditions, B = intervention manipulation) designs, replicated across time with a changing criterion.

Experimental control is established as each subphase serves as baseline for the next and controls for student learning. The power of proposed functional relationships between the intervention and behavior is determined by the number of replications present in the design, the temporal contiguity of observed changes in intervention and behavior, and the degree to which step changes predict consistent trend and level changes in performance. The experimental nature of the design may be enhanced by adding a reversal condition demonstrating lower criterion levels predicting lower performance levels, then raising (or lowering) the criteria and again demonstrating a predictable rise (or decline) in performance. However, the main advantage of the changing criterion design is as a teaching design that evaluates instructional strategies and programs for individual students. In addition, the changing criterion design is one of only two single-subject designs (the other is multiple baseline design) that involves an inherent control for learning such that what the subject learns during the intervention does not confound the experimenter's ability to maintain experimental control.

To use a changing criterion design, one must first measure baseline responding and use that performance to determine the first criterion for acceptable performance. At each step, the criterion should be obtainable with some level of effort, with the focus being on creating success. Next, the subject receives instruction and is introduced to both the criterion and contingencies for reinforcement. Performance is then measured during each trial until the criterion has been met and is stable. Generally, performance should be exhibited three or more consecutive times to ensure mastery and [Page 1217]ability to successfully move on. Each new criterion should be set at variable levels to keep the focus on performance rather than on the degree of change. As a general rule, experimental control may be demonstrated with no fewer than four step changes, with predictable changes in behavior. If stronger demonstration of experimental control is desired, a reversal to some lower (or higher) criteria may be reintroduced to demonstrate a corresponding drop in performance. Reversals should be followed by a return to the original step at which performance was last recorded and continue on until performance is at a desired level.

...