Knowing what each single-case design looks like is one skill. Picking the right one for a client, and running it well, is another: fitting the design to the question, the behavior, and the setting.
Why it matters
As a BCBA you are expected to make treatment decisions based on data, and the quality of those decisions depends on how the data were collected. A design that fits the situation lets you say with some confidence that your intervention, not something else, changed behavior. A poorly chosen design can be unethical (withdrawing treatment for dangerous behavior), impossible (trying to "unlearn" a skill), or simply uninformative.
Practice items for this task give a short case: pick the design that fits, or spot what is wrong with the one chosen. A few standard questions settle most cases.
Key ideas
Before choosing a design, ask:
- What is the question? Do you want to show that one intervention works, compare two or more interventions, or move behavior gradually toward a goal?
- Will the behavior reverse? Many behaviors maintained by ongoing contingencies return toward baseline when treatment is withdrawn. Newly learned skills, such as reading words or tying shoes, often do not.
- Is withdrawal acceptable? Removing an effective treatment for severe self-injury or aggression, even briefly, raises ethical concerns.
- How many targets do you have? Several behaviors, settings, or people that could receive the same intervention at different times open the door to a multiple baseline.
- Does the behavior change in steps? Behavior that is already in the repertoire and can be increased or decreased toward a goal in stages fits a changing-criterion design.
A quick matching guide:
- Reversal (A-B-A-B): strongest demonstration of control when the behavior is reversible and withdrawal is safe.
- Multiple baseline (across behaviors, settings, or participants): no withdrawal needed; staggered, concurrent baselines rule out many rival explanations. Tiers should be independent of one another but likely to respond to the same intervention.
- Multiple probe: a multiple baseline with intermittent baseline probes, useful when continuous baseline measurement would be impractical or could itself change behavior.
- Multielement (alternating treatments): rapidly alternate two or more conditions to compare them, without a withdrawal phase or a long stable baseline.
- Changing criterion: step the performance criterion up or down; control is shown when behavior changes to match each new criterion.
Running the design well matters as much as choosing it. Change phases based on the data, not the calendar: wait for stable responding (or a clear trend away from the expected effect) before introducing or changing a condition. Keep measurement, setting, and staff consistent, and check procedural fidelity so the independent variable is what you think it is.
Definitions
- Reversal design: baseline and treatment conditions are introduced, withdrawn, and reintroduced, so effects can be verified and replicated within one person.
- Multiple baseline design: baselines start at the same time for two or more behaviors, settings, or participants, and the intervention begins in each tier at a different, staggered time.
- Multiple probe design: a variation of the multiple baseline in which baseline data for untreated tiers are collected intermittently rather than continuously.
- Multielement design: two or more conditions alternate rapidly, often session by session, each paired with a distinct stimulus. Also called an alternating treatments design.
- Changing-criterion design: treatment stays in place while the performance criterion changes in a series of steps.
- Multiple treatment interference: when behavior in one condition is affected by exposure to another condition, such as carryover or order effects.
Examples from a session
Matching design to case
- Home, reversible behavior: Ji-woo, 6, often leaves the dinner table. Praise and a token for staying seated work well, and leaving the table is not dangerous. A brief return to baseline, then reinstating the tokens, gives a clear A-B-A-B demonstration.
- School, skills that will not reverse: Ms. Okafor, a school-based BCBA, teaches Andre, 9, three independent conversation skills. Because learned skills are unlikely to return to baseline, she uses a multiple baseline across behaviors and starts teaching each skill at a different time.
- Clinic, comparing treatments: Lucia, 4, shows disruptive behavior during work. Her BCBA alternates two break procedures across sessions, each signaled by a different colored card, and compares the two data paths in a multielement design.
- Adult services, gradual change: Sam, 38, already rides a stationary bike but only for a few minutes. Reinforcement is delivered when he meets a minutes criterion that rises from 5 to 10 to 15 and beyond. Riding time that tracks each step shows control in a changing-criterion design.
Common mistakes / exam traps
- Withdrawing treatment for dangerous behavior without strong justification. When safety is at stake, a design that keeps treatment in place is usually the better choice.
- Tiers that are not independent. If teaching one behavior also changes the next, untreated tiers move early and verification is lost. If the tiers are too different, they may not respond to the same intervention.
- Ignoring carryover in a multielement design. Counterbalance the order of conditions and make each condition clearly discriminable. When comparing teaching procedures, give each its own equally difficult targets, since a skill learned under one also shows up under the other.
- Changing criteria in identical steps. Varying the size and length of criterion phases makes it more convincing that behavior is following the criterion, not just drifting.
Exam trap
Reversal vs. multiple baseline: the deciding feature is whether the behavior can reverse and whether withdrawal is safe. A learned skill will not reverse, so a withdrawal phase verifies nothing; stagger the intervention across tiers in a multiple baseline or multiple probe instead.
Exam trap
Multielement vs. reversal: a multielement design compares conditions without a stable baseline or a withdrawal phase, but its price is multiple treatment interference: responding in one condition can carry over into the other, so check for carryover before trusting the comparison.
Quick check
Rafael, a BCBA, has reduced Nia's severe hand biting with a function-based treatment at home. He wants to show the same treatment works at school and at her grandmother's house. Which plan best demonstrates control while avoiding a withdrawal of treatment?