Single-Case and Intensive Within-Subject Designs

Core Idea

Single-case designs (ABAB, multiple-baseline, changing-criterion) intensively study one individual across repeated measurements and condition changes. They establish causality by showing that behavior changes when treatment is introduced and reverts when withdrawn. Suitable for clinical practice, organizational consultancy, and understanding individual mechanisms when group designs are impractical.

How It's Best Learned

Study a published single-case design and trace how repeated measurements reveal treatment effects. Design your own single-case study specifying baseline, treatment, and reversal phases. Discuss when individual-level evidence is clinically or theoretically crucial.

Common Misconceptions

• Single-case designs are anecdotal; - Causality requires large group studies; - Single-case findings cannot be generalized; - Multiple-baseline designs require reversal.

Explainer

From your study of research design selection, you know that choosing a design depends on matching the research question to the appropriate unit of analysis, the available sample, and the causal claims you want to make. Most experimental designs achieve causal inference by averaging across many participants — random assignment distributes individual differences across conditions, so any systematic outcome difference is attributable to the manipulation. Single-case designs take a different route to the same destination: instead of averaging across people, they accumulate many observations *within* one person over time, using the person's own stable baseline as the counterfactual.

The foundational logic is reversal. In the simplest ABAB design, the researcher establishes a baseline phase (A) by measuring the target behavior repeatedly until it is stable. Then treatment is introduced (B), and the behavior is measured repeatedly again. If the behavior changes with the introduction of treatment, that's suggestive — but it could be coincidence or a natural trend. The critical move is withdrawal: the treatment is removed (returning to A), and if the behavior reverts toward baseline, the pattern strongly implicates treatment as the cause. Reintroducing treatment (second B) and seeing the behavior change again creates a replication within the single case, making chance explanations increasingly implausible. Four phase changes within one participant, each moving in the predicted direction, provide convincing causal evidence.

The multiple-baseline design is used when reversal is impractical or unethical — you wouldn't want to withdraw a successful intervention for self-injury just to prove a point. Multiple-baseline designs introduce treatment to different behaviors, settings, or individuals at staggered time points while keeping the others at baseline. If each target behavior changes only when treatment is introduced to it (and not before), this staggered replication rules out history, maturation, and other time-based confounds. The logic is "if it only moves when I push it, I must be doing the pushing." No reversal is needed; temporal coincidence of change and intervention across multiple baselines makes the causal argument.

Operational measurement — your other prerequisite — matters enormously here because inference depends entirely on the reliability and precision of the dependent variable across hundreds of repeated observations. Unlike group designs where measurement error averages out across participants, in single-case work each data point must be trustworthy. Poor operationalization introduces noise that masks true treatment effects; inconsistent measurement creates artifactual trends that mimic treatment responses. The design's causal power is proportional to measurement quality. This is why single-case researchers invest heavily in behavioral coding procedures, inter-rater reliability checks, and observer calibration.

Single-case designs are not purely local — they can support generalization through systematic replication: applying the same treatment protocol with the same design across a series of individuals, settings, and clinicians. When a therapeutic technique works in ABAB replications with 20 individuals across five different clinicians in three different settings, the accumulation of evidence rivals the external validity of many randomized controlled trials, while also providing mechanistic insight into how treatment works at the individual level. This is why single-case methodology dominates applied behavior analysis, clinical psychology for rare conditions, and educational interventions for individual students — contexts where understanding and helping the individual is both the practical and scientific goal.