What's tested on Domain D
- Single-subject (single-case) experimental designs: ABAB / reversal, multiple baseline (across behaviors, settings, or participants), multielement / alternating treatments, changing criterion
- When to use each design, and when reversal is unethical (irreversible behaviors, danger to client)
- Threats to internal validity: history, maturation, instrumentation, multiple-treatment interference
- Replication: direct, systematic, and across subjects
- Component analysis and parametric analysis
- How to identify experimental control in a graph
Why this domain matters
On the exam, Domain D shows up as design-recognition questions ("which design is depicted?") and design-selection scenarios ("the BCBA wants to evaluate X, which design is most appropriate?"). In practice, picking the wrong design means you can't tell whether a behavior change was caused by your intervention or by something else that happened to coincide.
How to study Domain D for the BCBA exam
- Memorize the visual signature of each design. Reversal has clear ABAB phases; multiple baseline has staggered onset across tiers; alternating treatments has rapidly alternating data paths.
- Know which designs require reversal of treatment and which don't; that's the biggest deciding factor in clinical contexts.
- Practice naming the threat to internal validity that each design controls for.
- Don't confuse multielement design with a multiple-baseline design; they have different logic, different graph.
Frequently asked questions
How many BCBA exam questions are on experimental design?
Thirteen questions out of 175, or 7% of the exam. Smaller than the application domains but conceptually dense.
When is a reversal design ethically problematic?
When the target behavior is dangerous to the client or others (self-injury, aggression), when the skill being taught is socially important enough that withdrawal would be harmful, or when the behavior change is irreversible (e.g., a skill that, once acquired, can't be "un-learned"). In those cases, multiple-baseline designs are usually the right alternative.