Single-Case Design Logic: Prediction, Verification, Replication
Key Takeaways
- Prediction is the anticipated future pattern of behavior if conditions remain unchanged.
- Verification occurs when a condition change tests whether the predicted baseline pattern would have continued.
- Replication occurs when the effect is repeated across phases, tiers, behaviors, settings, or participants.
- Repeated measurement and visual analysis are central to evaluating level, trend, variability, immediacy, overlap, and consistency.
The logic behind the graph
Single-case designs evaluate behavior repeatedly within and across conditions. The analyst first predicts what behavior will do if the current condition continues. Then the analyst changes the condition to test that prediction. If behavior changes and the effect is replicated, confidence in a functional relation increases.
| Logic step | Question to ask | Graph cue |
|---|---|---|
| Prediction | What would happen if nothing changed? | Stable or interpretable baseline |
| Verification | Did the old pattern return or remain absent when tested? | Reversal, staggered untreated tier, or continued baseline |
| Replication | Did the effect happen again? | Repeated change across phases or tiers |
| Experimental control | Are alternatives less plausible? | Consistent behavior change tied to the IV |
Visual analysis evaluates six common features: level, trend, variability, immediacy of effect, overlap, and consistency across similar phases. No single feature is enough. A design with high overlap, delayed change, and unstable baseline is harder to defend than one with clear, immediate, repeated changes.
Prediction requires enough data to see a pattern. A baseline does not need a fixed number of points, but it should be long enough to support a reasonable forecast unless ethical or practical constraints require action. An increasing baseline for severe behavior may justify rapid treatment but weakens a simple treatment-effect claim.
Verification can occur differently by design. In a reversal, returning to baseline tests whether behavior changes back. In a multiple baseline, untreated tiers verify the prediction by remaining unchanged until the intervention reaches them. In a changing-criterion design, behavior should track each new criterion.
Replication is the core protection against coincidence. A single behavior change after treatment might be due to history. Repeated changes at planned times make that explanation less likely.
In a multiple-baseline design, two untreated settings remain at baseline levels while intervention changes behavior in the first setting. What design logic is this best illustrating?
Which visual pattern most strongly supports a functional relation?
A baseline has a clear decreasing trend before treatment begins, and the target behavior continues decreasing after treatment starts. What is the best interpretation?