Regression to the mean in a single-case phase change can make an extreme observation tend to be followed by a value closer to the person's usual range, even without a phase effect. Reduce this risk by avoiding transitions triggered solely by a peak or trough, preserving the full series, defining timing rules prospectively, checking context, and seeking replicated changes. A post-change return from an extreme point is weak causal evidence by itself.

Define an extreme point in context

Use the person's series, scale, opportunities, and setting rather than a universal numerical cutoff.

Inspect why the point occurred

Review measurement, health, schedule, access, partner, task, and environmental events without assuming error.

Avoid single-point release rules

Require a broader pattern or a timing mechanism that does not wait for an unusually favorable transition.

Preserve sensitivity results

Show how reasonable inclusion and exclusion rules affect summaries without selecting the version that produces the strongest claim.

Seek replication before causal language

Repeated changes at planned phase transitions provide more useful evidence than one movement away from an extreme value.

Build Zoe's extreme-point phase review

For the regression to the mean single-case phase change question, create a versioned extreme-point phase review. Record the target, measurement unit, eligible opportunities or observation time, scale, desired direction, raw series, missing and invalid states, context, graph version, timing mechanism, decision rule, reviewers, clinical gates, participation response, decision time, actual change, deviations, and follow-up. The record should let another qualified reviewer reconstruct why Zoe's phase decision occurred without relying on memory.

Work through Zoe's baseline example

Zoe's baseline counts are 2, 3, 2, 3, and 9. The team changes phase immediately after 9, and the next values are 3, 2, and 4. Those values return toward the earlier range. The sequence does not show that the new condition reduced the behavior because an unusual baseline peak determined the transition. Show the raw series and all denominators before adding summaries. This fictional outpatient coping-skills study example illustrates one decision pattern and does not create a universal stability percentage, point count, phase duration, treatment rule, or causal conclusion.

Audit the evidence available for Zoe

Zoe's review retains all eight values, the point-9 context, observer and opportunity checks, the timing rule, the decision time, and whether the rule existed before the peak. It compares conclusions with and without the extreme value as a labeled sensitivity check while preserving the original series. The audit also checks data-entry history, definition version, observer training, agreement when needed, graph axes, session spacing, exclusions, phase labels, source timestamps, and access to original records. Unresolved discrepancies remain visible and pause any claim that depends on them.

Address the main interpretation risk for Zoe

Deleting the 9 erases the decision context, while treating 9-to-3 as the treatment effect exaggerates the comparison. Zoe's report displays the point, investigates it, and gives the later phase enough evidence and replication to stand on more than one transition. The review describes the observed pattern in original units, names plausible alternatives, and separates a methods judgment from a clinical recommendation. Software may calculate, graph, and surface missing evidence; qualified people interpret the series and make decisions within their authority.

Use the later phase responsibly for Zoe

The team looks for repeated, temporally related changes across phases or tiers and considers level, trend, variability, immediacy, overlap, and clinical meaning. It states that ordinary fluctuation remains a plausible explanation where replication is absent. Visual analysis examines level, trend, variability, immediacy, overlap, and consistency across comparable phases or tiers. A supplemental statistic can summarize a defined feature. It cannot establish clinical importance, functional relation, consent, authorization, or treatment fit on its own.

Protect Zoe's access and participation

Keep Zoe's augmentative and alternative communication, interpreters, mobility, food, water, bathroom use, prescribed care, health support, rest, relationships, and emergency help available throughout the outpatient coping-skills study. Use accessible consent and assent processes when applicable and respond to withdrawal, dissent, or distress. Necessary safety or clinical action proceeds through qualified authority even when it changes the planned phase timing.

Apply current sources to Zoe's decision

Zoe's source trail emphasizes full-pattern visual analysis and makes the extreme-point timing risk explicit in the phase decision. The BACB ethics hub and CASP public summary provide professional context, while the BCBA Test Content Outline identifies examination content on measurement and single-case design. The WWC Version 5.0 handbook is a research-review standard. A single-case design review describes within-phase and between-phase visual analysis. Research on waiting for baseline stability examines response-guided timing, and work on systematic visual protocols, current analytic reflections, and visual-analysis software clarifies methods and limits. ASHA supports continuous AAC access.

Rehearse Zoe's review before live use

Run the extreme-point phase review with a fictional series before it governs Zoe's data. Test minimum and maximum observations, missing values, equal timestamps, trend in both directions, extreme points, floor and ceiling values, a failed access gate, assent withdrawal, delayed implementation, and an amended rule. Store expected decisions, reviewer rationale, screenshots or graph versions, software version, and correction history. The rehearsal tests mechanics while leaving the live judgment to qualified reviewers.

Close Zoe's phase-decision review

Review the extreme-point phase review with Zoe, the responsible clinician, and a methodologist familiar with the design. Preserve raw data, graph, rule, decision snapshot, reviewers, client input, access and safety evidence, phase implementation, deviations, sensitivity checks, later outcomes, and inference limits. Keep the page draft and noindex until the manifest-named reviews are complete.

Related resources

Sources