Overview
Experimental Design is Section D of the BCBA outline, with 9 task statements. It covers independent and dependent variables, internal and external validity, threats to internal validity, the features of single-case designs, the strengths of single-case and group designs, critique of single-case data, the four main design types, comparative, component and parametric analyses, and applying designs in practice.
Questions focus on the logic of demonstrating a functional relation. You will see short descriptions of studies or clinical evaluations and choose the design that fits, identify what a phase contributes, or judge whether the data show experimental control. The ethical side matters too: you should not choose a reversal design for dangerous behavior or for skills that will not reverse. Section D builds directly on the measurement and visual analysis skills in Section C.
How it's tested
Section D items usually describe a client, a target behavior and a practical or ethical constraint, then ask which design or analysis fits. Others present the results of a design and ask what they show. Answers that ignore reversibility, safety or the need for staggered introduction are common distractors.
-
Pick the design
Constraints such as irreversible skills, dangerous behavior or multiple settings are given and you choose the design that fits them.
-
Name the threat
An outside event or measurement change coincides with the intervention and you identify the threat to internal validity.
-
What does this phase do
A design is described and you identify whether a phase provides prediction, verification or replication.
A clinician wants to evaluate a teaching procedure for reading sight words with one learner, across three word sets. Options are a reversal design, a multiple baseline across word sets, a changing-criterion design and a group comparison. Reading skills will not reverse once learned, so withdrawal cannot show control. Staggering the procedure across word sets demonstrates control without withdrawal, so a multiple baseline across behaviors is best.
Task statements 9
Official task statements from the Behavior Analyst Certification Board outline, in outline order.
D.1 Distinguish between dependent and independent variables.What is measured versus what is manipulated.
- What it means
- The dependent variable is the behavior being measured. The independent variable is the condition, procedure or treatment the researcher or practitioner manipulates to see whether it changes the dependent variable.
- What to understand
- In applied work the dependent variable is a socially significant target behavior and the independent variable is the intervention. Extraneous or confounding variables are uncontrolled factors that could also affect the dependent variable.
- How it shows up
- Items describe a study and ask which element is the independent or the dependent variable.
- Common confusion
- Calling the client or setting the independent variable.
D.2 Distinguish between internal and external validity.Confidence in a functional relation versus confidence in its generality.
- What it means
- Internal validity is the degree to which changes in the dependent variable can be attributed to the independent variable rather than to other factors. External validity is the degree to which a functional relation holds across people, settings, behaviors and time.
- What to understand
- Single-case designs build internal validity through repeated measurement and replication within the study. External validity in behavior analysis comes mainly from direct and systematic replication across studies, not from large samples.
- How it shows up
- A vignette describes a well-controlled study in one setting and asks what limits its conclusions.
- Common confusion
- Assuming a single-case study cannot have external validity at all.
D.3 Identify threats to internal validity (e.g., history, maturation).Factors other than the intervention that could explain change.
- What it means
- Threats include history (outside events), maturation (changes due to growth or time), testing effects, instrumentation changes such as observer drift, regression, sequence effects and treatment drift. Each offers an alternative explanation for a change in behavior.
- What to understand
- Designs control these threats through stable baselines, staggered introduction of treatment and replication. Integrity data rule out treatment drift as an explanation.
- How it shows up
- Items describe an outside event, such as a new medication starting with the intervention, and ask which threat it represents.
- Common confusion
- Confusing maturation with history when the change comes from an outside event.
D.4 Identify the defining features of single-case experimental designs (e.g., individuals serve as their own controls, repeated measures, prediction, verification, replication).The logic of baseline, prediction, verification and replication.
- What it means
- In single-case designs each participant serves as their own control and behavior is measured repeatedly. Baseline logic uses steady-state data to predict future responding, verification shows that behavior would have stayed the same without treatment, and replication repeats the effect.
- What to understand
- Prediction comes from a stable baseline. Verification comes from returning to baseline or from untreated tiers that stay stable. Replication strengthens the demonstration of a functional relation within a study.
- How it shows up
- Items ask which element of baseline logic a particular phase provides.
- Common confusion
- Confusing verification (showing baseline would have continued) with replication (repeating the effect).
D.5 Identify the relative strengths of single-case experimental designs and group designs.When each research approach is the better fit.
- What it means
- Single-case designs show individual effects, allow repeated measurement, are flexible and fit clinical practice. Group designs compare averages across many participants and can estimate how an intervention performs in a population.
- What to understand
- Group averages can hide individuals who do not respond or who get worse. Single-case designs reveal these patterns but rely on replication for generality. Each approach answers different questions.
- How it shows up
- A scenario asks which approach suits a question about an individual client or about a population.
- Common confusion
- Assuming large sample sizes automatically provide better evidence for an individual client.
D.6 Critique and interpret data from single-case experimental designs.Judging whether the data show a functional relation.
- What it means
- This task asks you to evaluate single-case data for stability, immediacy and size of effect, overlap, consistency across replications and threats to validity, and to decide whether experimental control was demonstrated.
- What to understand
- Look at each phase change for a clear change in level or trend that is replicated. Unstable baselines, too few data points or inconsistent replications weaken conclusions.
- How it shows up
- Data from a design are described and you judge whether a functional relation was shown.
- Common confusion
- Treating any improvement after the intervention as a functional relation.
D.7 Distinguish among reversal, multiple-baseline, multielement, and changing-criterion designs.Core single-case designs and their logic.
- What it means
- Reversal (A-B-A-B) designs withdraw and reintroduce treatment. Multiple-baseline designs stagger the intervention across behaviors, settings or participants. Multielement (alternating treatments) designs rapidly alternate conditions. Changing-criterion designs step the performance criterion and look for behavior to follow it.
- What to understand
- Choose reversal only when behavior is reversible and withdrawal is ethical. Multiple baseline suits irreversible skills. Multielement compares treatments quickly but risks multiple-treatment interference. Changing criterion suits gradual change in a behavior already in the repertoire.
- How it shows up
- A clinical situation is described and you select the design that fits its ethical and practical limits.
- Common confusion
- Recommending a reversal design for a skill that will not reverse or for dangerous behavior.
D.8 Identify rationales for conducting comparative, component, and parametric analyses.Answering different research questions about interventions.
- What it means
- Comparative analyses compare two or more interventions. Component analyses identify which parts of a package are active, using dropout or add-in approaches. Parametric analyses examine different values of a variable, such as different reinforcement magnitudes or schedule values.
- What to understand
- Choose the analysis by the question: which intervention, which component, or how much. Component analyses help make packages more efficient.
- How it shows up
- Items describe a study's purpose and ask which analysis it represents.
- Common confusion
- Calling a comparison of reinforcer amounts a comparative analysis instead of parametric.
D.9 Apply single-case experimental designs.Using designs in practice to evaluate client interventions.
- What it means
- This task asks you to design and conduct single-case evaluations in practice, choosing a design that fits the behavior, setting and ethical limits, and planning phase changes based on data.
- What to understand
- Practitioners can often embed a multiple baseline or brief reversal within services. Phase changes should depend on stable data, not on a calendar. Ethics and client welfare limit design choices.
- How it shows up
- A scenario gives constraints such as dangerous behavior or multiple settings, and you choose and apply a design.
- Common confusion
- Changing phases on a fixed schedule regardless of data stability.
High-yield concepts
The terms, rules and formulas this domain turns into questions most often.
-
Baseline logic
Prediction, verification and replication.
- Remember
- Stable baseline supports prediction.
- Tested as
- Phase-function items.
- Trap
- Confusing verification with replication.
-
Reversal design
A-B-A-B withdrawal and reintroduction of treatment.
- Remember
- Requires reversible behavior and ethical withdrawal.
- Tested as
- Design selection items.
- Trap
- Using it for dangerous or irreversible behavior.
-
Multiple-baseline design
Staggered introduction across behaviors, settings or people.
- Remember
- Untreated tiers must stay stable until treated.
- Tested as
- Common choice for skills.
- Trap
- Starting all tiers at once.
-
Multielement design
Rapid alternation of conditions.
- Remember
- Compares treatments quickly; risk of multiple-treatment interference.
- Tested as
- Treatment comparison and functional analysis items.
- Trap
- Ignoring carryover effects.
-
Changing-criterion design
Stepwise criteria; behavior follows each step.
- Remember
- Suits gradual change of a behavior already in repertoire.
- Tested as
- Shaping-style goals such as increasing exercise.
- Trap
- Using it for skills not yet learned.
-
Internal validity
Confidence that the independent variable caused the change.
- Remember
- Threatened by history, maturation, instrumentation and others.
- Tested as
- Threat identification items.
- Trap
- Confusing with external validity.
-
External validity
Generality across people, settings and time.
- Remember
- Built through replication across studies.
- Tested as
- Questions on limits of a single study.
- Trap
- Thinking only group designs provide it.
-
Multiple-treatment interference
One treatment's effects influence another.
- Remember
- A concern in multielement designs.
- Tested as
- Design critique items.
- Trap
- Ignoring sequence and carryover.
-
Component analysis
Identifies the active parts of a package.
- Remember
- Dropout or add-in methods.
- Tested as
- Analysis classification items.
- Trap
- Calling it comparative.
-
Parametric analysis
Evaluates different values of one variable.
- Remember
- How much, not which.
- Tested as
- Analysis classification items.
- Trap
- Confusing it with comparative analysis.
Side by side
Concepts candidates mix up, compared on the point that decides the answer.
Single-case designs
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Reversal | Withdraw and reintroduce treatment. | Needs reversibility. | Using it for dangerous behavior. |
| Multiple baseline | Stagger treatment across tiers. | No withdrawal needed. | Introducing treatment simultaneously. |
| Multielement | Rapidly alternate conditions. | Compares conditions within sessions or days. | Ignoring interference. |
| Changing criterion | Step criteria and track behavior. | Behavior must follow each criterion. | Using it to teach new skills. |
Types of analysis
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Comparative | Which of two or more interventions works better. | Different interventions. | Mixing with component analysis. |
| Component | Which parts of a package are needed. | Parts of one package. | Calling it parametric. |
| Parametric | Which value of a variable is best. | Different amounts of one variable. | Calling it comparative. |
Validity
| Concept | Meaning | Key difference | Common trap |
|---|---|---|---|
| Internal validity | Change attributable to the intervention. | Within the study. | Equating with generality. |
| External validity | Change holds elsewhere and with others. | Beyond the study. | Thinking sample size alone provides it. |
Exam traps
Each trap is an answer that looks right. Learn the clue that should change your mind.
-
Reversal for dangerous behavior
- What candidates pick
- Reversal design
- Why it looks right
- It gives the clearest demonstration of control.
- Why it's wrong
- Withdrawing an effective treatment for self-injury or aggression risks harm; a multiple baseline or other design is safer.
Clue to watch for The behavior could cause injury.
-
Reversal for skills
- What candidates pick
- Reversal design
- Why it looks right
- It is the classic design.
- Why it's wrong
- Once a skill is learned it usually does not return to baseline, so withdrawal cannot demonstrate control.
Clue to watch for The target is an academic or communication skill.
-
Simultaneous start in multiple baseline
- What candidates pick
- Begin treatment in all tiers
- Why it looks right
- It seems faster and fairer.
- Why it's wrong
- Without staggering, there is no verification that untreated tiers would have stayed stable.
Clue to watch for Look for staggered introduction.
-
History vs maturation
- What candidates pick
- Maturation
- Why it looks right
- Both concern changes over time.
- Why it's wrong
- An outside event such as a new medication or classroom change is history; maturation is change due to growth.
Clue to watch for The stem names a specific outside event.
-
Comparative vs parametric
- What candidates pick
- Comparative analysis
- Why it looks right
- Both compare conditions.
- Why it's wrong
- Comparing amounts of one variable is parametric; comparative analysis compares different interventions.
Clue to watch for The conditions differ only in amount or intensity.
-
Group design for one client
- What candidates pick
- A group design
- Why it looks right
- Group designs are seen as the stronger evidence.
- Why it's wrong
- Questions about whether an intervention works for one client are best answered with a single-case design.
Clue to watch for The question concerns a specific individual.
Answer framework
Use this process when choosing a design or judging whether data show experimental control.
- Identify the independent and dependent variables, and confirm that the dependent variable is measured repeatedly within every phase of the evaluation.
- Check whether the target behavior can reverse and whether withdrawing treatment would be safe and ethical for this particular client and setting.
- If withdrawal is unsuitable, consider staggering treatment across behaviors, settings or people, or stepping performance criteria when gradual change is the goal.
- Ensure baseline data are stable before each phase change, so the design supports prediction and later verification of the effect.
- Judge experimental control by looking for replicated changes in level or trend that occur only when the independent variable changes.
Practice question
Want more like this? The BCBA practice test has questions for every domain.
Checklist
0 of 10
Tick each one you can do without notes. Your ticks stay in this browser.
Study strategy
Allocate 15 to 20 hours to Section D. Start with baseline logic, because every design question depends on it. Draw each design by hand with hypothetical data and label which phases provide prediction, verification and replication. Then list the conditions that rule each design in or out, such as reversibility, safety and the number of available behaviors or settings.
Next, read three or four short single-case studies from the Journal of Applied Behavior Analysis. For each, name the design, the threats it controls and whether the data convinced you. This builds the critique skill in D.6 and links to visual analysis from Section C.
In the final phase, practise scenario items that include ethical constraints. Ask yourself whether withdrawing treatment would harm the client and whether a skill could truly reverse. Review comparative, component and parametric analyses with one example of each so the terms do not blur together on exam day.
FAQ
How many Experimental Design questions are on the BCBA exam?
Experimental Design is Section D and has 13 of the 175 scored questions, about 7% of the exam.
When should I avoid a reversal design?
Avoid it when the behavior is unlikely to reverse, such as a learned skill, or when withdrawing treatment would put the client or others at risk.
What makes a multiple-baseline design convincing?
Stable baselines in all tiers, staggered introduction of treatment, and changes that occur in each tier only when treatment starts there.
Do I need to know group design statistics?
No detailed statistics are listed. You should know the relative strengths of single-case and group designs and why group averages can hide individual responses.
How are multielement designs used in practice?
They compare two or more conditions quickly, as in functional analyses or treatment comparisons. Watch for multiple-treatment interference.
What is the difference between verification and replication?
Verification shows behavior would have stayed at baseline without treatment. Replication repeats the effect of the treatment, strengthening the demonstration of control.
Test your Domain 4 knowledge
Practise choosing single-case designs that fit ethical and practical constraints, and judging whether described data show experimental control.