Can ABA Data From Different Settings Be Combined?

Settings can change the opportunity itself

A help request at home, a clinic table, a noisy store, and a telehealth call may involve different partners, materials, delays, sensory conditions, risks, and consequences. Ten opportunities in one setting may not represent the same task as ten in another.

Ask for the target definition, eligibility rule, ordinary supports, prompt plan, observation duration, and staff role for each setting.

Run a comparability check before pooling

Pooling can be reasonable when the measure intentionally represents a broader class and each setting uses the same valid rule. It can also be useful to report an overall total beside setting-specific results. Pooling becomes misleading when one setting supplies most opportunities or uses different supports, definitions, or observation windows.

The BCBA Test Content Outline covers measurement, graphing, generalization, and data interpretation as examination content rather than a universal pooling rule.

Keep raw counts and denominators visible

Suppose a client responds in 8 of 10 clinic opportunities and 1 of 2 home opportunities. The combined result is 9 of 12, or 75%. That total hides the small home sample. Report all three facts and avoid calling home performance stable from two opportunities.

If one setting lacks AAC, has different prompts, or records only successful trials, repair the condition before comparing outcomes.

Clinical interpretation stays qualified

The BACB Ethics Code addresses assessment, client-informed goals, intervention, data evaluation, documentation, and generalization for covered behavior analysts. The clinician should predeclare when setting data are pooled, explain weights, and mark changes.

Client and family feedback can identify which setting matters most or feels least workable.

A practical example

Layla's team measures an accessible break request at home, clinic, and art class. The same response forms and five-second window apply, but the art-class partner response differs. The report shows each setting separately. It uses a combined total only for the broad question of whether a recognizable request occurred, while partner-response outcomes remain setting-specific.

Build a setting comparability matrix

Place each setting in a row and compare the target definition, eligible opportunity, response window, observation length, partner, ordinary supports, prompt plan, data collector, phase, and purpose. Mark each dimension as aligned, different but acceptable for the broad question, or incompatible.

This makes the pooling decision reviewable. “Same goal” is not enough when the actual opportunity differs. A request during a planned clinic trial and a spontaneous request during a crowded event may both be important while requiring separate interpretation.

Choose weighting deliberately

Pooling raw opportunities gives settings with more opportunities greater influence. Averaging setting percentages gives each setting equal influence regardless of sample size. Neither rule is universally correct.

Suppose clinic performance is 18 of 20 and home performance is 2 of 4. The pooled result is 20 of 24, or 83.3%. The unweighted mean of 90% and 50% is 70%. A family should see both setting values before the team selects a summary. The chosen weighting should match the decision.

Keep generalization questions visible

Setting-specific data often answer whether a skill transfers beyond teaching conditions. A combined total can hide that evidence. If performance is strong in the clinic and unobserved in the community, the overall result should not imply community use.

The team can report a matrix of trained, probed, supported, and unobserved settings. Generalization claims should name the people, places, materials, and supports actually sampled.

Repair access differences before outcome comparison

If AAC, visual supports, mobility access, language support, or another ordinary aid is missing in one setting, address the access problem. Poor performance under an inaccessible condition should not be used as evidence that the person lacks the skill.

The record can still preserve what occurred, including the system gap. The qualified clinician can decide whether the observation is valid for the clinical question and whether a new observation is needed after access is restored.

A second example with unequal observation time

Omar uses a coping routine twice during a 30-minute clinic session and four times during a three-hour community outing. Counts alone suggest more use in the community. Rates are four per hour in clinic and 1.33 per hour in the community.

Those rates may still answer different questions because the opportunities and activities differ. The team should define whether it is measuring routine use per hour, use per eligible stressor, or successful access when Omar chooses the routine. Combining counts before defining the denominator would obscure the decision.

Know when to keep settings separate

Separate displays are usually clearer when definitions, supports, prompt levels, observers, opportunity rules, or phases differ materially. They are also useful when one setting is central to the person's priorities or when a safety or access issue occurs only there.

A combined summary can appear beside these displays when it adds a legitimate broad view. It should never be the only view if it conceals a meaningful difference.

Document the pooling rule prospectively

The plan or measurement note can state which settings may be combined, the formula, minimum sample, missing-setting rule, and triggers for separate review. Mark changes in that rule. This reduces the risk that staff pool data only when the combined result looks favorable.

At the family meeting, ask the clinician to explain what the pooled number changes. If no decision needs it, the setting-specific evidence may be the more useful report.

Review combined data after a setting changes

A move, new classroom, telehealth switch, staffing change, or facility disruption can make an earlier pooling rule obsolete. Mark the date and compare conditions before carrying the old combined trend forward.

The new setting may need its own baseline. If service continues during transition, label observations as transition data rather than forcing them into an established setting average. Ask the person which supports and partner responses need to transfer.

Preserve minority settings in the report

One location may produce only a few opportunities while remaining highly important. A rare medical visit, community activity, or family routine should not vanish because clinic data dominate the denominator.

Show the small sample with its limits and the person's experience. The team can decide whether more observation is useful or burdensome. A low-volume setting deserves an explicit disposition, even when it cannot support a stable percentage.

Questions families can use

Ask which settings are included, whether definitions and supports match, how opportunities are weighted, whether one setting dominates the total, what missing settings mean, how client preference is represented, and which clinical decision needs a pooled result.