How can families review ABA data? Start with the exact definition and why the measure matters. Check raw counts, eligible opportunities, exclusions, prompts, settings, ordinary supports, dates, missing data, observers, and every intervention change before interpreting a graph or percentage. Ask how the person's own feedback and daily-life outcomes relate to the pattern, and what decision the data can or cannot support.
Begin with the question and definition
Ask what decision the measure is intended to inform and how an observer identifies an event. Request examples and nonexamples. A label such as engagement, independence, or distress can hide different definitions across people and settings.
The CASP public guideline summary places data within individualized assessment, treatment, and evaluation.
Rebuild the denominator
For opportunity measures, ask who was eligible, when the opportunity began and ended, what counted, which events were excluded, and why. Keep client withdrawal, missing AAC, unavailable materials, prompts, and environmental failures visible. For durations, identify the start and stop events.
Calculate a few rows from raw counts to confirm the reported percentage.
Read graphs with context
Check axis labels, scale, dates, condition lines, intervention changes, missing periods, setting, observers, and ordinary supports. A truncated axis or pooled settings can exaggerate or hide a pattern. Compare similar conditions before drawing a conclusion.
Ask whether observer agreement, calibration, or other quality checks are relevant when the measure drives a major decision.
Include communication and client meaning
The person's report can differ from an adult observation. Keep both sources attributable. The ASHA AAC portal supports continuous AAC access, including during measurement and review.
The BACB Ethics Code addresses data, documentation, client involvement, assessment, intervention, risk, and evaluation for covered behavior analysts.
A fictional raw-data check
A family's worksheet contains 20 recorded opportunities. Three lack needed materials and two lack AAC, leaving 15 valid client-skill opportunities. The person uses the defined message in 11: 11 of 15, or 73.3%. System readiness is reported separately as 15 of 20, or 75%.
Pooling all 20 would mislabel access failures as client performance. The figures still cannot establish causality.
Tie evidence to a bounded decision
Ask whether the pattern supports continue, change, gather evidence, refer, pause, transition, or close. Record uncertainty and what new observation would matter. Avoid using one metric to answer safety, quality, preference, and outcome questions at once.
Review ABA data as a conversation among definition, context, direct client input, and clinical judgment. Clean arithmetic matters, while the decision still requires qualified interpretation.
Sources
Finni resources
