BCBA Experimental Design: Single-Case Designs and How to Read Them

What Domain D of the BCBA exam covers

The BCBA Test Content Outline (6th ed.) gives Experimental Design 13 of the 175 scored questions (7%), across 9 tasks. Only Behaviorism and Philosophical Foundations is smaller. The outline stays in force for 2027: the BACB's 2027 BCBA Requirements keep the exam on the 6th edition, in effect since January 1, 2025. The tasks move from vocabulary to judgment:

Domain D leans on Domain C: you read the same graphs, but the question shifts from "what happened?" to "can we say the intervention caused it?" The task numbers and topics are the BACB's; the explanations below follow the published sources listed with this guide.

Variables, validity and the threats the exam names

The dependent variable (D.1) is what you measure, usually the target behavior. The independent variable is what you manipulate, usually the intervention.

Internal validity (D.2) is how confident you can be that the change in the dependent variable came from the independent variable and not from something else. A study with strong internal validity demonstrates experimental control. External validity is how far the effect generalizes to other people, settings or behaviors. Single-case research builds external validity through replication rather than large samples.

The outline gives history and maturation as examples of threats to internal validity (D.3). These and related threats appear in scenario items:

Threat names and design features are recall material, and 41 flashcards on Experimental Design are planned for launch.

Baseline logic: prediction, verification and replication

Task D.4 lists the features that make single-case designs work. Each person serves as their own control, and behavior is measured repeatedly, so every phase gives a picture of the behavior over time. Three steps then build the argument:

• Prediction. A stable baseline lets you predict what the behavior would keep doing without intervention.
• Verification. Showing that the prediction would have held, for example by returning to baseline conditions and seeing the behavior return to baseline levels, or by showing that untreated baselines stay unchanged.
• Replication. Producing the same effect again, in the same person or across people, settings or behaviors.

A simple A-B design, one baseline and one intervention phase, shows a change but has no verification or replication, so it gives little basis for ruling out history or maturation. When an item asks whether an A-B study shows a functional relation, that missing step is what to look for.

Single-case versus group designs

Task D.5 asks for the relative strengths of each, not a winner:

Each has a cost: group averages can hide individuals for whom the treatment didn't work, and a single-case result needs replication before it can be generalized.

The four single-case designs the outline names

Task D.7 asks you to tell these apart, usually from a description or a graph:

Practicing the designs from unlabeled graphs helps this table stick, and 63 planned questions on Experimental Design are part of full prep at launch.

Comparative, component and parametric analyses

Task D.8 asks why a study would be set up each way:

• Comparative analysis. Compares the relative effects of two or more different interventions to see which works better for this person.
• Component analysis. Identifies which parts of a treatment package produce the effect, either by removing components (dropout) or adding them one at a time (add-in), often so that ineffective or unnecessary parts can be dropped.
• Parametric analysis. Compares different values of the same independent variable, such as three different durations of the same procedure, to find an effective or efficient level.

A quick way to sort an item: different treatments means comparative, parts of one treatment means component, and amounts of one variable means parametric.

Judging whether a graph shows experimental control

Tasks D.6 and D.9 ask you to critique data and apply designs. A checklist for any single-case graph:

1. Is the baseline stable, or at least not already moving in the direction the intervention is supposed to produce?
2. Are there enough data points in each phase to see level, trend and variability?
3. Does the behavior change when, and only when, the condition changes?
4. Is the effect replicated, across phases or tiers?
5. In a multiple-baseline design, do untreated tiers stay put until the intervention reaches them?
6. In a multielement design, are the data paths clearly separated, without much overlap?
7. Were measurement and implementation checked, with interobserver agreement and procedural integrity data?

For applying designs, match the constraint in the stem to the design: if the behavior can't or shouldn't be reversed, a multiple-baseline design; if two treatments need a quick comparison, multielement; if the goal moves in steps, changing-criterion; if the behavior is reversible and withdrawal is acceptable, reversal. The BCBA measurement and data display guide covers how to describe each phase before you judge.

A worked Domain D example written by Passy

This item was written by Passy for practice. It is not a BACB question.

A BCBA wants to evaluate a token system meant to increase work completion for three students in three different classrooms. The school will not allow the system to be removed once a student starts it. Which design fits best?

• A. Reversal design
• B. Multiple-baseline design across participants
• C. Changing-criterion design
• D. Multielement design

Keyed answer: B. Starting the token system with one student at a time, while the others stay in baseline, shows control without ever removing it. A requires withdrawing the system, which the school won't allow. C needs a criterion that changes in steps, which the stem doesn't describe. D needs conditions to alternate rapidly, which again means switching the system off and on.

Studying Experimental Design for the BCBA exam

• Sketch each design from memory: one reversal, one multiple-baseline with three tiers, one multielement and one changing-criterion.
• Name the design from unlabeled graphs in published articles you read for coursework.
• Run the checklist above on every graph, including ones that look convincing.
• Study it with Measurement. The BCBA study plan guide pairs Domains C and D in the same week.

Your next step for Domain D

Draw the four designs today without notes, then check them against the table. Try the 20 free BCBA practice questions, which include an Experimental Design item, and if Domain D stays your weakest, Weak Spot drills planned for launch work on your lowest-accuracy domain five questions at a time.

Passy is not affiliated with or endorsed by the BACB. The example item is original and isn't a BACB exam question.