/ Single-Case Design Graphs

Overview › Design guide › Reversal

Design guide

Reversal (ABAB) design

Introduce the intervention, withdraw it, and introduce it again. If behavior changes every time, the intervention is the most likely cause.

You will be able to

  • Explain how the four phases of an ABAB design give prediction, affirmation, verification, and replication.
  • Check a reversal design against the minimum phases and data points per phase in the WWC pilot standards.
  • Decide whether reversal graphs show experimental control, and say why when they don't, reaching 8 of your last 10 in the practice.
Question it answersDoes behavior change each time the intervention is introduced and withdrawn?
Use it whenBehavior is likely to return to baseline when the intervention stops, and withdrawing it is safe, ethical, and acceptable.
Avoid it whenThe behavior won't reverse (such as a learned skill), or withdrawing treatment would put someone at risk.

How it shows control

Each phase plays a role in baseline logic. The first baseline predicts what would happen without intervention. The first intervention phase shows behavior departing from that prediction. That is affirmation of the consequent: if the intervention controls the behavior, behavior should change when it starts, and it did. On its own, that proves little, because something else could have changed at the same time. Returning to baseline verifies the prediction: if behavior goes back, the first baseline was an accurate forecast. Reintroducing the intervention replicates the effect. The result is three demonstrations of the effect at three points in time.

Reversal graph of percentage of intervals on task. Baseline is near 30 percent; a dashed predicted line continues at 30 percent through the first intervention phase, where data rise to 60 to 85 percent. Data return to 30 to 45 percent in the second baseline and rise to 80 to 92 percent when the intervention is reintroduced. Labels mark prediction, affirmation, verification, and replication.
Baseline logic. The dashed line shows the level baseline predicted for the first intervention phase. Behavior changed instead (affirmation of the consequent), went back when the intervention was withdrawn (verification), and changed again when it was reintroduced (replication).

The logic of this design in more depth, its judgment calls and exceptions, and a knowledge check: Visual analysis Lesson 3, Part B.

Minimum standards

What experimental control looks like

Experimental control Reversal graph of percentage of intervals on task. Data are 25 to 35 percent in the first baseline, 60 to 85 percent in the first intervention phase, 30 to 45 percent in the second baseline, and 80 to 92 percent in the second intervention phase.
Three clear demonstrations. The level changes immediately at every phase change, there is no overlap between adjacent phases, and similar phases look alike (consistency). Every phase has at least five data points.

What it doesn't look like

No reversal Reversal graph of percentage of intervals on task. Data rise from 25 to 35 percent in baseline to 60 to 85 percent in the first intervention phase, then stay at 80 to 86 percent after the intervention is withdrawn and at 82 to 88 percent when it is reintroduced.
Behavior didn't return. In the second baseline, behavior stayed at intervention levels. The design is left with one demonstration, and history or maturation could explain the first change.
Not replicated Reversal graph of percentage of intervals on task. Data are 25 to 35 percent in the first baseline, rise to 60 to 85 percent in the first intervention phase, fall to 30 to 45 percent in the second baseline, and stay at 29 to 34 percent when the intervention is reintroduced.
Behavior didn't change again. Behavior rose when the intervention began (mean 30% to 75%) and fell when it was withdrawn (mean 35%), but stayed at the baseline level (mean 31%) when it was reintroduced. That is two demonstrations, not three, so the graph doesn't show experimental control.
Baseline trend Reversal graph where percentage of intervals on task rises steadily from 20 to 84 percent across all four phases, about 3 to 4 points per session, with no jump at any phase change.
The trend was already there. Behavior rose steadily from the first session, and phase changes made no difference to the data path. The first baseline predicted the later values. A baseline trend weakens the case only when it points toward the level the intervention should produce. If baseline was trending away from that level and behavior still changed at each phase change, the case is stronger, not weaker.
Too variable Reversal graph where percentage of intervals on task swings between about 20 and 80 percent in every phase, so the four phases overlap almost completely.
No difference in level. The data swing widely and overlap almost completely across phases. With this much variability, a longer baseline or tighter measurement is needed before changing phases. Variability alone doesn't rule out control: data that bounce around but stay clear of the adjacent phases can still show it.

Variations. An A-B-A-B-A-B design adds more replications. To test a second treatment, give it its own reversal, for example A-B-A-B-A-C-A-C, so each treatment is compared with baseline three times. Treatments in adjacent phases, as in A-B-A-B-C-B-C, can be compared with each other directly, but the earlier treatment can change how behavior responds to the later one (multiple-treatment interference), so counterbalance the order across participants when you can. A B-A-B design starts with treatment, so it lacks a true baseline and, like A-B-A, gives only two demonstrations.

Practice

Decide whether each graph shows experimental control. When it doesn't, also choose why. An item counts only when both answers are right.

Graph it