Parametric analysis

One feature changes by planned amounts

A parametric analysis might compare a 2-second, 5-second, and 10-second prompt delay. Another might compare two, four, and six practice opportunities per session. The values are selected prospectively and applied according to a clear schedule.

The independent variable is the quantitative feature. The dependent variable is the measured outcome, such as independent responses, errors, latency, preference, or implementation burden. Other procedure components stay stable enough for the value comparison to make sense.

A simple “low versus high” comparison can be parametric when the values are clearly defined, although three or more values often reveal shape better. A dose-response relation may be increasing, decreasing, flat, threshold-like, or curvilinear.

Parametric and component analyses differ

A component analysis adds, removes, or separates parts of a treatment package. A parametric analysis keeps the feature and changes its amount or value.

Comparing a visual schedule alone with a visual schedule plus modeling is a component question. Comparing two, four, and six modeled practice trials is a parametric question. A study can include both, but each inference needs its own planned comparison.

The BACB BCBA Test Content Outline identifies rationales for comparative, component, and parametric analyses within its experimental-design domain. The outline describes examination content rather than supplying a clinical protocol.

Choose values that answer a useful question

Begin with the real decision. A team may want the shortest delay that supports independence, the fewest trials associated with steady acquisition, or a schedule that balances performance and feasibility.

Select values that are:

• safe and clinically reasonable
• distinguishable in practice
• feasible to implement accurately
• broad enough to show a pattern
• compatible with client preference and access
• measurable through adequate opportunities

Avoid values whose purpose is to provoke distress, exhaust the person, restrict communication, or remove essential health and safety support. Parametric logic never supplies authority for a risky procedure.

Control order and exposure

Value order can affect the result. Repeated practice may improve performance over time, so an ascending sequence can make the highest value look best. Fatigue can create the opposite pattern.

Counterbalance or randomize order when appropriate. Repeat each value across enough occasions to examine variability. Use distinct condition cues when rapid switching could confuse the participant or implementer. Record carryover, missed opportunities, prompts, and procedural integrity.

If changing values produces durable learning, later values begin with a different history. Report that limitation and use a design suited to the question.

A fictional opportunity analysis

Nia chooses a goal of independently selecting an activity using speech, AAC, or gesture. The team compares sessions with two, four, or six arranged choice opportunities. Each value appears in six sessions, and all recognizable choices are honored.

The raw count rises across values, yet the proportions are 75.0%, 79.2%, and 69.4%. Six opportunities produce more total practice while the independent percentage falls. The team also finds longer sessions and more signs of fatigue at six.

Four opportunities may offer the best observed balance for this narrow sample. The data do not prove a universal optimum. Order, practice, day-to-day variation, and the small number of sessions remain plausible influences.

Interpret the whole response curve

Graph each value and its order of presentation. Examine level, trend, variability, immediacy, overlap, and consistency. Include raw opportunities and fidelity beside percentages.

Look for a region where added intensity produces little added benefit or greater burden. A single highest point can be noise. Replication around the candidate range is often more informative than adding an extreme value.

The WWC Single-Case Design Technical Documentation describes repeated measurement, internal-validity concerns, visual analysis, and evidence of a relation. Its standards are research-review guidance and do not set a clinical dose.

Report enough detail to reproduce the values

For every condition, report the unit, exact value, scheduling rule, exposure, opportunity definition, exclusions, implementation, and relevant context. Define how a partial interval or interrupted session is handled.

Separate effectiveness from feasibility. Record staff time, client experience, errors, adverse effects, and preference. A value that produces a marginal performance increase at heavy cost may be a poor operating choice.

Retain data from every tested value, including values stopped under a safety or assent rule. Report why exposure ended and which observations remain interpretable.