Clinical importance versus statistical significance in ABA compares two different judgments. Statistical significance reflects a result under a specified model, design, statistic, and assumptions. Clinical importance asks whether the observed change matters to the person in daily life, given size, context, durability, burden, risk, and alternatives. A small p value cannot make a trivial outcome meaningful, and a valued change can deserve attention without one.

State the statistical question

Name the design, statistic, reference distribution or model, assumptions, missingness, and exact result.

State the clinical question

Describe what change would improve daily life, for whom, in which setting, and with what retained supports.

Report magnitude in original units

Show counts, minutes, probabilities, steps, or another interpretable measure with denominators and variation.

Assess burden and unwanted effects

Include effort, time, distress, opportunity cost, access, adverse outcomes, and the person's preference.

Keep conclusions separate

Record statistical evidence, causal inference, clinical importance, generalization, and durability in distinct fields.

Build Ben's clinical-and-statistical interpretation table

For the clinical importance versus statistical significance ABA question, create a versioned clinical-and-statistical interpretation table. Preserve the client priority, exact question, original unit, eligible opportunity or exposure, numerator, denominator, source identity, communication access, context, direct observation, self-report, proxy report, burden, unwanted effects, generalization, maintenance, decision owner, and uncertainty. Another qualified reviewer should be able to reconstruct Ben's outcome without replacing lived experience with a summary score.

Work through Ben's outcome example

Ben's randomization analysis returns p = .03 for an average increase of 0.4 independent choices per session. The calculation may indicate the pattern is unusual under that design's null assignment distribution. Ben says he notices no practical difference, and the change does not appear in community sessions. Statistical evidence and clinical importance therefore receive different entries. Show each source, count, denominator, setting, and timeframe before summaries. This fictional clinic choice-making study example illustrates one outcome-review pattern and does not establish a universal meaningful-change threshold, instrument, treatment recommendation, social-validity conclusion, or causal claim.

Audit Ben's outcome evidence

Ben's table records design, assignment mechanism, statistic, p value, original-unit change, opportunity denominator, context, client report, generalization, burden, and uncertainty. It prevents p = .03 from being copied into a clinically meaningful field. The audit also checks definition and instrument versions, source validity, administration access, privacy, missing and declined states, correction history, context coverage, graph or calculation precision, burden, adverse effects, and client review. Unresolved evidence remains visible and narrows the exact claim it affects.

Address Ben's main interpretation risk

Threshold language can turn p values into pass or fail labels and conceal effect size or design limits. Ben's report gives the exact result and avoids universal cutoffs. A measure can be reliable and still miss the person's priority. A client-valued outcome can be important even when the design cannot attribute it. Reviewers preserve both distinctions and avoid turning one person's result into a population promise.

Choose Ben's next action

The team can refine the measure, gather meaningful-context evidence, or continue because Ben values another part of the plan. No statistical result overrides his access, safety, or preferences. Record the action, rationale, qualified owner, client response, support, due date, and review trigger. Keep technical change, client value, burden, harm, generalization, maintenance, and causal evidence as separate fields.

Protect Ben's access and voice

Keep Ben's AAC, interpreters, mobility, food, water, bathroom use, prescribed care, health support, rest, relationships, and emergency help available. Offer accessible and private ways to accept, decline, pause, correct, or withdraw from outcome collection. A caregiver or clinician can contribute evidence without authoring Ben's report. Immediate safety or medical concerns follow the applicable route.

Apply current sources to Ben's review

Ben's page applies evidence-based ABA principles by combining research evidence, local data, context, and client values without merging their roles. The BACB ethics hub and CASP public summary provide professional context, while the BCBA Test Content Outline identifies examination content on client goals, measurement, and evaluation. An evidence-based ABA framework, contemporary social-validity analysis, and review of social validity describe client values and repeated evaluation. Research on measuring what matters, family-centered care, and an accessible autistic-adult measurement toolkit highlights priority and access issues with scoped validation. ASHA supports continuous AAC access.

Rehearse Ben's outcome workflow

Test the clinical-and-statistical interpretation table with fictional client and proxy disagreement, a declined rating, AAC failure, small denominator, large percentage with trivial magnitude, benefit with burden, adverse event, response substitution, missing community context, and incomplete maintenance. Confirm that states, denominators, source labels, urgent routes, claim holds, and qualified review behave as intended. Store expected outputs, reviewer notes, software version, and corrections before live use.

Close Ben's outcome review

Review the clinical-and-statistical interpretation table with Ben, the responsible clinician, and specialists required by the outcome. Preserve raw data, direct voice, access method, proxy evidence, original-unit magnitude, contexts, burden, adverse effects, generalization, maintenance, claim wording, disagreements, and limits. Keep the page draft and noindex until every manifest-named review is complete.

Related resources

Sources