A sequential, test-control methodology for conducting functional analyses of self-injurious behavior.
When multielement FA data are muddy, switch to a sequential test-control (pairwise) design to get clearer differentiation.
01Research in Context
What this study did
The authors compared two ways to run a functional analysis. One way flips quickly between conditions each day. The other way tests one condition for several days, then tests the control for several days.
They called the second way a sequential, test-control, or pairwise design. They wanted to see which design gave clearer answers about why self-injury happened.
What they found
In four cases they could read, the pairwise design showed a clear function twice. The fast-flip multielement design stayed muddy for those same two cases.
The other two cases looked the same in both designs. Bottom line: when your multielement graph is flat and messy, running conditions in long pairs can clean it up.
How this fits with other research
Rose et al. (2000) kept the fast-flip format but added color cues and different therapists. Clear functions popped out for half their kids, showing you can also sharpen data without slowing down.
Kahng et al. (1999) took a different route. They stayed in the same multielement schedule but zoomed in on minute-by-minute shifts. Tracking when reinforcement started or stopped also rescued unclear graphs.
These papers do not fight each other. Davis et al. (1994) says slow pairs help; Rose et al. (2000) says salient cues help; Kahng et al. (1999) says micro-pattern help. You can mix any of these tools when your FA looks like a plate of spaghetti.
Why it matters
You do not have to guess when an FA is undifferentiated. Try the pairwise trick first: run attention for three days, then control for three days, then escape for three days, and so on. You will spot the winning condition faster and protect your client from extra sessions. If time is tight, add colored posters or swap therapists instead. Either path beats writing 'no clear function' in the report.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Pick one undifferentiated FA on your caseload and rerun it in pairwise blocks of three sessions per condition
02At a glance
03Original abstract
Multielement and reversal designs used to identify maintaining variables for behavior disorders such as self-injury have several potential limitations, including interaction effects (multielement), inefficiency (reversal), and lack of a continuous control (reversal). This article describes a methodology that minimizes these problems yet captures the best features of both designs. This design consists of several phases implemented in a sequential (A-B-C) fashion, as in the reversal design. However, each phase consists of two conditions, a test and a control, presented concurrently in a multielement format. Five subjects' self-injury was assessed using both the multielement design and the sequential, test-control (or pairwise) design. Results for two subjects indicated that the multielement design produced clear assessment outcomes, and similar findings were obtained using the pairwise design. For two other subjects, the multielement assessments were somewhat undifferentiated, and clearer results were obtained using the pairwise design. The fifth subject's self-injury showed cyclical patterns using both assessment techniques. Benefits and limitations of the sequential assessment methodology are discussed.
Behavior modification, 1994 · doi:10.1177/01454455940183003