Baseline

Baseline describes performance under named conditions

A baseline answers what is happening now and under which conditions. Examples include accessible communication frequency, completed routine steps, participation duration, activity-start latency, and the percentage of eligible opportunities with a defined response. Name the behavior, measure, dates, settings, people, ordinary supports, and observation or opportunity rules.

The CASP ABA Practice Guidelines place assessment and treatment planning within standards of care for ABA behavioral health treatment of people diagnosed with autism. This article uses only that public scope; the licensed guideline was not used.

Several numbers can describe one baseline

An average can hide trend, range, or setting differences. Graph observations in order and retain raw counts, durations, and opportunities. For a percentage, report numerator and denominator; for a rate, count and observation time. Score zero only when a valid opportunity occurred and no defined event was observed. Code no opportunity, unavailable AAC, absent observer, and lost data separately.

Build a representative baseline

The BACB BCBA Test Content Outline, 6th edition covers operational definitions, representative measurement, graphs and interpretation, procedural-integrity data, assessment of strengths and needs, client-informed goals, and single-case-design features. It is examination content, not a clinical protocol. Apply those distinctions:

1. Define the response in observable, accessible terms.
2. Choose a decision-relevant dimension: count, duration, latency, steps, or opportunities.
3. Define observation boundaries and eligible opportunities.
4. Sample people, settings, times, activities, and support conditions.
5. Record prompts, AAC access, health or environmental events, missing data, and exclusions.
6. Train observers and sample agreement or integrity when needed.
7. Graph promptly to detect errors, risk, or a trend.

Baseline length depends on the decision, measure, variability, trend, risk, feasibility, and design. “Stable” should mean sufficiently predictable for that decision, not necessarily flat, and no fixed point count establishes it. “Three data points” is not a universal treatment-start rule. WWC Version 5.0 generally requires six initial-baseline points for its highest rating in multiple-baseline/probe, treatment-reversal, and changing-criterion designs, with a three-point exception for zero within-phase variance. These are education-research review thresholds, not clinical prescriptions. More observations cannot repair an ambiguous definition, biased sample, inaccessible communication, or delayed needed care.

A clinical reference differs from experimental control

In treatment planning, baseline is a starting picture. Later change does not prove the plan caused it. Time, practice, health, staffing, season, setting, measurement drift, or simultaneous events can confound a before-and-after comparison.

The BACB Test Content Outline identifies repeated measurement, prediction, verification, and replication as defining single-case experimental-design features used to evaluate a functional relation. The WWC Handbooks page identifies Version 5.0 as current. Its handbook applies phase, data-point, design, and visual-analysis standards when reviewing education research. Those standards do not determine whether clinical service is research or when care should start.

Safety and access set the boundary

Baseline does not justify prolonging danger, distress, pain, medical risk, lost communication, or another preventable barrier. Set observable stop and action criteria, and provide urgent or required care immediately. Use records, a brief safe sample, concurrent assessment, or another design when extended observation would be unsafe or burdensome. Do not delay medically necessary care or a safety support to obtain a cleaner baseline.

An experimental design does not by itself make an activity research. For behavior analysts covered by the BACB Ethics Code, research requires approval by a formal research review committee; research conducted during service delivery must prioritize client services and welfare and follow applicable research-consent and assent requirements. Other legal and institutional requirements may also apply.

Name ordinary supports in the baseline condition. Do not remove augmentative and alternative communication (AAC), mobility tools, prescribed care, protective equipment, sensory supports, or a safety plan for an “unsupported” level. ASHA's AAC practice portal says AAC users should always have access to their communication tools or devices.

For covered BCBA and BCaBA certificants and people who have completed an application for either credential, the current BACB Ethics Code addresses client involvement, informed consent and assent when applicable, assessment, risk minimization, accurate data and documentation, and continual evaluation. BACB has no separate jurisdiction over organizations or corporations. Other laws, professions, schools, payers, and settings may add duties.

A fictional baseline example

Mika, a fictional 14-year-old AAC user, wants a reliable way to pause crowded group activities. The team defines an eligible opportunity as a scheduled group transition when Mika is present, awake, and has AAC within reach. An independent pause message is speech, AAC, sign, or a pre-agreed gesture before a prompt. Any other clear withdrawal, refusal, or distress signal is honored and documented even if it does not meet that scoring definition. A partner pause within ten seconds counts as timely; an immediate safety action is coded separately.

An access audit covers 12 scheduled transitions and finds AAC within reach for 10 of 12. The two transitions without AAC are environmental-access failures, not zero communication responses. After device placement is fixed, 18 transitions across six days meet all eligibility rules. Mika independently communicates pause during 5 of 18 eligible opportunities, or 27.8%. No safety exception occurs for those five messages; partners pause within ten seconds after 4 of 5, or 80%.

The report keeps both measures: Mika's message count and the adults' response. It also shows daily values of 0/3, 1/3, 0/3, 1/3, 1/3, and 2/3, so readers can see an upward trend with day-to-day variation. The clinician begins teaching without waiting for a flat line and marks the preexisting trend when interpreting later change. Mika helps select the goal, and the plan includes partner training because adult response is part of usable communication.

Later data can be compared with this baseline under the same definitions. Any change in opportunities, partners, support, prompts, or measurement is marked on the graph. The comparison can describe progress; a claim that teaching caused it would require a design capable of supporting that conclusion.

Questions families can ask

• Which goal or decision will this baseline inform?
• How were the response and eligible opportunity defined?
• Which settings, people, and support conditions are represented?
• Were access failures separated from the person's performance?
• What do the level, trend, variability, and raw data show?
• Which safety event or dissent signal ends observation and triggers action?
• How will my family and the person receiving care review or question the interpretation?
• What would lead the team to change the measure or reassess?