Showing that an intervention works is only the first answer. Once you have a functional relation, better questions follow. Is this treatment better than another one? Which parts of it are doing the work? How much of it does the person need? Comparative, component, and parametric analyses are the designs that answer those three questions.
Why it matters
Every intervention costs something: staff time, materials, training, and the client's time. A package with five parts may work, but if two parts do all the work, the other three add effort without benefit. A reinforcer may work at 5 minutes and at 20 minutes, but the shorter value lets the learner get back to instruction sooner. These analyses help you choose treatments that are effective, efficient, and easier for caregivers and staff to keep doing.
This task feeds domain D, Experimental Design. Practice items on it describe a study or a clinical question and ask you to name the type of analysis or pick the reason it was run. Learn the question each analysis answers and you can sort most items quickly.
Key ideas
- Comparative analysis asks "which one?" It compares two or more distinct interventions to see which produces more change, faster change, or change with less effort.
- Component analysis asks "which parts?" It examines a treatment package to find which components are active, which are necessary, and which are sufficient on their own.
- Parametric analysis asks "how much?" It compares a range of values of one independent variable, such as reinforcer magnitude, schedule value, or session length, to see how behavior changes across those values.
- All three usually build on an effect you already trust. Their main job is to refine an intervention, not to show for the first time that something works.
- The design depends on the question. Comparative and parametric questions are often tested with an alternating treatments (multielement) design or a sequence of phases such as A-B-C-B-C. Component analyses usually add or remove components across phases.
Definitions
- Comparative analysis: an experiment that compares the effects of two or more different treatments on the same behavior.
- Treatment package: an intervention made of several components delivered together, such as a visual schedule, a token system, and a brief warning before transitions.
- Component analysis: an experiment designed to identify the active elements of a treatment package and their relative contributions.
- Drop-out analysis: a component analysis that starts with the full package and then removes components one at a time, watching whether behavior changes.
- Add-in analysis: a component analysis that tests components alone or in combinations before (or without) the full package.
- Parametric analysis: an experiment that compares the effects of different values of the same independent variable.
Examples from a session
Three questions, three analyses
- Clinic, comparative: Malik, 6, engages in aggression maintained by escape from tasks. His BCBA, Rosa, alternates sessions of functional communication training (requesting a break) with sessions of noncontingent breaks. Aggression drops faster and further during the communication sessions, so Rosa recommends that treatment.
- School, component: A fourth-grade teacher uses a package of a visual schedule, a 2-minute warning, and tokens to support transitions for Chloe, 10. To save time, the BCBA runs a drop-out analysis. Removing the warning has no effect, but removing either the visual schedule or the tokens brings back long transition times. The team keeps the schedule and tokens and drops the warning.
- Day program, parametric: Henry, 34, earns music time for completing work steps. His BCBA compares 2, 5, and 10 minutes of music. Work completion is about the same at 5 and 10 minutes and lower at 2. She uses 5 minutes, which keeps the effect and gives Henry more time in his work routine.
Notice that each example asked a question beyond "does it work?" The analysis improved the plan rather than proving it.
Common mistakes / exam traps
- Mixing up comparative and parametric. Two different treatments (differential reinforcement vs. noncontingent reinforcement) is comparative. Different amounts or values of the same treatment (1, 3, or 5 tokens) is parametric.
- Mixing up comparative and component. Comparing a whole package with a different treatment is comparative. Comparing a package with parts of itself is component.
- Ignoring sequence and interaction effects. In an add-in analysis, a component tested later may look stronger or weaker because of what came before. Some components work only in combination, so testing them alone can understate them. In a parametric analysis run as phases, values tested in only one order (1, then 3, then 5 tokens) are confounded with time and practice; repeat or reverse the order before crediting the value.
- Forgetting multiple-treatment interference. When treatments alternate rapidly in a comparative or parametric study, exposure to one condition can influence responding in another. Clear, distinct stimuli for each condition help.
Exam trap
Ask what the conditions have in common. Different interventions means comparative. Parts of one package means component. Different values of one variable means parametric.
Exam trap
Drop-out vs. add-in: check the direction. A drop-out analysis starts with the full package and removes parts; an add-in analysis starts with parts and builds up. Then check what the result shows. No change when a part is dropped means it was not necessary alongside the others, not that it has no effect alone; the remaining parts or natural contingencies may be holding the gain. No change when a part is tested alone means it is not sufficient by itself, not that it adds nothing in combination.
Quick check
Daniela, a BCBA, has shown that losing a point for each call-out reduces call-outs for Jae, 13. She now runs phases in which each call-out costs 1, 3, or 5 points. Which kind of analysis is she conducting?