The big idea
Once an intervention is known to work, three follow-up questions are common. Which of two treatments works better? Which parts of a treatment do the work? How much of a variable is needed? Each question has its own kind of analysis:
- A comparative analysis compares two or more different treatments.
- A component analysis tests the parts of a treatment package to find which ones are responsible for the effect.
- A parametric analysis tests different values of one independent variable.
All three serve the same practical goals: interventions that are more effective, more efficient and better understood.
Breaking it down
Comparative analysis
A comparative analysis pits distinct interventions against each other, such as functional communication training versus noncontingent reinforcement, or two prompting procedures. The rationale is choosing the treatment that produces the most change, the fastest change, or a good change for the least cost and effort. Comparisons usually use a multielement (alternating treatments) design or a reversal design with more than one treatment phase (A-B-A-C), so multiple-treatment interference is a concern (see D.3).
Component analysis
Many interventions are a treatment package: several procedures used together, such as the instructions, modeling, rehearsal and feedback of behavioral skills training. A component analysis finds which components are necessary, which are sufficient and which add little. There are two approaches (Ward-Horner & Sturmey, 2010):
- Drop-out: start with the full package, then remove components one at a time. If the effect weakens when a component is removed, that component mattered. A weakness is that once the package has worked, behavior may be maintained by other variables, so removing a component may show no loss even if it helped at first.
- Add-in: assess components one at a time, or in growing combinations, before the full package. This can show which components are sufficient alone, but sequence effects can make a later component look stronger than it would be by itself.
The rationale is efficiency and understanding. A package with unnecessary parts costs staff time, takes longer to train and may be harder for caregivers to carry out consistently. Knowing which part does the work also makes the intervention more conceptually systematic (see A.5).
Parametric analysis
A parametric analysis keeps the treatment the same and varies one of its values: reinforcer magnitude, session length, the interval in a DRO schedule, the number of tokens needed for an exchange or the amount of practice. The rationale is finding the value that produces the desired effect most efficiently, such as the shortest effective break or the smallest effective amount of reinforcement. It also shows how behavior changes across a range of values, not just whether the variable matters at all.
Seeing it in practice
| Question | Analysis | How it might run |
|---|---|---|
| Is picture prompting or video prompting faster for teaching Ji-woo to make a sandwich? | Comparative | The two procedures alternate across sessions, each with its own recipe. |
| Which parts of a caregiver-training package (written instructions, modeling, role-play, feedback) are needed? | Component (drop-out) | After the full package works, role-play is removed, then modeling, while caregiver accuracy is tracked. |
| Does feedback alone improve staff data accuracy, or is goal-setting also needed? | Component (add-in) | Feedback is introduced first, then goal-setting is added. |
| How long should breaks be to keep Malik working without problem behavior? | Parametric | Breaks of 30 seconds, 1 minute and 3 minutes are compared. |
| Does the number of tokens needed for an exchange change work rates? | Parametric | Exchange requirements of 3, 5 and 10 tokens are tested in turn. |
Common mix-ups
- Comparative vs. component. Comparative analyses compare different treatments. Component analyses compare parts of one treatment.
- Component vs. parametric. A component analysis asks which part matters. A parametric analysis asks how much of one variable works best.
- Drop-out vs. add-in. Drop-out starts with everything and subtracts. Add-in starts with less and adds.
- These are kinds of questions, not designs. Each is answered with the designs in D.7, most often multielement or reversal arrangements.
Exam tip: Look at what varies across conditions. Two different procedures: comparative. Parts of a package added or removed: component. Different amounts, durations or intensities of the same procedure: parametric.
Quick recap
- Comparative: which treatment is better (more effective, faster or cheaper).
- Component: which parts of a package are responsible; drop-out removes parts, add-in adds them.
- Parametric: which value of a variable works best.
- All three make interventions more efficient and better understood.
Further reading
- Ward-Horner, J., & Sturmey, P. (2010). Component analyses using single-subject experimental designs: A review. Journal of Applied Behavior Analysis, 43(4), 685-704.
- Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.), chapters 8 to 10.
Key terms
- Comparative analysis
- An analysis that compares the effects of two or more distinct interventions to find out which is more effective or efficient, usually with a multielement design or a reversal design that includes more than one treatment.
- Component analysis
- An analysis that identifies which parts of a treatment package are responsible for its effects. In a drop-out analysis, components are removed one at a time from the full package; in an add-in analysis, components are assessed alone or added in combinations.
- Parametric analysis
- An analysis that compares the effects of different values of one independent variable, such as reinforcer magnitude, session length or a schedule requirement, to find the value that works best or to describe how behavior changes across values.
- Treatment package
- An intervention made up of two or more distinct procedures used together. A component analysis can show which parts of a package produce its effect.
- Multielement design (Alternating treatments design, ATD)
- A single-case design in which two or more conditions alternate rapidly, often from session to session, each paired with a distinct stimulus and plotted as its own data path. Differences between the data paths show the effect of the conditions.
Check yourself
5 questions on D.8. Choose an answer to see whether you're right and why.
Show the answer
Answer: A. Starting with the full package and removing components one at a time is a drop-out component analysis. An add-in analysis would start with fewer components and add them. A comparative analysis would compare the package with a different treatment.
Show the answer
Answer: B. The treatment stays the same while one variable, the exchange requirement, takes different values. That is a parametric analysis, which looks for the most effective or efficient value. Testing parts of a package would be a component analysis.
Show the answer
Answer: C. Two distinct treatments are being compared, which makes it a comparative analysis. A multielement (alternating treatments) design is the usual choice because it compares conditions quickly within one person. Component analyses test the parts of a single package.
Show the answer
Answer: C. In an add-in analysis, components are introduced one after another, so a later component may look strong partly because of the earlier ones. Removing components from a package that has already worked describes a drop-out analysis, not an add-in.
Show the answer
Answer: C. A component analysis identifies which parts of a package produce the effect, so parts that add little can be dropped, saving staff time and effort. Finding the best number of tokens is a parametric question, and comparing with a different treatment is a comparative question.
Adds a step to your study record. Saved in this browser only.