How Should ABA Assessment Graphs Be Explained to Families?

ABA Assessment Graphs

Read the title and legend first. Check whether the denominator changes, zero differs from no opportunity, and data from different people or settings were combined. Mark every assessment or support change on the graph and inspect the underlying counts.

Start with what one point represents

A point might represent a session count, percentage, duration, rate, score, daily average, or summary across several observations. Ask for the unit and time window. “Four” could mean four requests, four minutes, or four events per hour.

The title should name the measure and context. Axes need units and readable scales. The legend should identify people, settings, phases, and conditions without relying only on color. If a point combines data, ask what was combined and how.

Always inspect the denominator

Percentages can move when opportunity counts change. A result of 4/5 is 80%, while 8/20 is 40%. The second observation contains more successful responses but a lower percentage. Neither should be interpreted without the opportunity definition and conditions.

Ask the graph or companion table to show raw numerator and denominator. Zero successful responses, zero eligible opportunities, missing observation, and unrecorded data need different symbols. Treating them all as zero can create a false decline.

Check whether conditions are comparable

Compare the setting, activity, people, communication and AAC access, prompts, partner behavior, wait time, health context, response definition, opportunity definition, observation duration, and collector. If these change, label them. A phase line can show when an assessment procedure, support, or definition changed, but the text still needs to explain what happened.

Describe patterns before explaining them

A family can first describe level, direction, variability, and missing data. “The four clinic points range from 30% to 80%, and the two highest points occurred with AAC and a familiar partner” is more defensible than saying the support caused improvement.

Evidence for cause requires a design and comparison suited to that question. A line rising after a change may be compatible with an effect while time, practice, setting, health, measurement, or other concurrent changes remain possible. The qualified clinician should state what the graph can and cannot support.

Make graphs accessible

Use high contrast, direct labels, patterns or shapes in addition to color, readable text, alt text, and a table of values. Explain the graph verbally and allow AAC, interpreter support, breaks, and questions. A client may prefer a simplified view first and the complete graph later.

Accessibility should preserve uncertainty and denominators. A simplified graph can retain the exact values, dates, and phase labels while reducing visual clutter.

Connect the graph to a decision carefully

Ask which assessment question the graph addresses and which additional evidence informs the decision. One graph may show performance in one condition without establishing generalization, preference, social importance, or future benefit.

If a recommendation depends heavily on the graph, request the underlying definition, raw table, exclusions, and analysis. The family can ask for a correction if a date, phase, or value is wrong and for a clinical explanation if the concern is interpretation.

Watch for scale and aggregation effects

A truncated vertical axis can make a small change look dramatic. A very broad axis can make meaningful variation look flat. Ask why the scale was chosen and view the actual values. Dual axes, stacked measures, and smoothed trend lines need especially clear explanation.

Weekly averages can hide day-level variation, while session-level points can overemphasize noise. If data are aggregated, ask for the rule, the number of observations in each summary, and whether missing sessions changed the average. Do not compare a weekly mean based on five sessions with one based on a single session without labeling the exposure.

Ask who checked the graph against the source data

The graph should reconcile with the underlying record. Check a few points against dates, raw counts, and phase changes. Confirm whether late entries, corrected data, or excluded observations are visible. A corrected point should preserve the reason and version rather than silently changing the displayed history.

Families do not need to recalculate every value. They can ask the clinician to walk through one representative point and one transition. If the explanation cannot connect the graph to the definition and raw count, the display is not ready to support a decision.

When several observers collect data, ask how definitions were trained and whether important disagreements appeared. Agreement measures can inform confidence, but they do not repair an inaccessible procedure or a poorly defined opportunity.

Before the feedback meeting ends, ask for an accessible copy of the graph and its value table. Record which version was discussed, any correction requested, and the specific assessment decision the display informed. If a later correction changes a plotted value or phase boundary, the revised graph should show its date and replace downstream copies deliberately.

Keep assessment authority and evidence clear

The CASP public summary places assessment and planning within its autism-treatment scope. The BACB Ethics Code addresses competence, understandable communication, client involvement, consent and assent when applicable, assessment, documentation, and evaluation for covered behavior analysts.

Make feedback accessible

The ASHA AAC portal supports continuous AAC access. For covered private practices, DOJ Title III guidance addresses effective communication and reasonable modifications, subject to scope and defenses.

Use the correct record route

For a HIPAA covered entity, HHS access guidance describes access to protected health information in a designated record set, subject to the rule's scope and procedures.

A feedback request, draft review, final report, and formal access request are different events.

A practical example

A graph shows 80% on Monday and 40% on Thursday. The raw values are 4/5 during a short clinic activity and 8/20 during a longer community visit. Monday included a familiar communication partner and open AAC vocabulary; Thursday used an unfamiliar partner and a different device page.

The family asks for raw counts, settings, partner, AAC access, and opportunity definitions to appear with each point. The clinician separates the conditions and states that the two percentages do not demonstrate a decline. A later matched observation uses the same definition and supports, while the clinician still describes the small sample and avoids a causal claim.

Questions families can use

Ask what each point, line, axis, and phase represents; what the raw numerator, denominator, unit, and time window are; how no opportunity and missing data are shown; which supports were present; whether definitions changed; which observations are comparable; and what the graph cannot establish.