An ABA schedule fairness audit tests whether written rules and actual scheduling outcomes differ across comparable cohorts. It examines access, wait time, offers, preferred windows, cancellations, geography, payer routes, communication supports, staff burden, exceptions, and unresolved records. The audit defines its population and comparison before reviewing results, then routes unexplained differences to qualified operational, clinical, access, workforce, payer, or legal review.

Choose a decision and cohort

Audit one process such as waitlist offers, staff assignment, cancellation backfill, preferred-window access, or leave coverage. Lock the period and eligible records before calculating outcomes. State which comparisons are lawful, relevant, and adequately supported.

Measure process and outcome

Compare response, review, offer, acceptance, release, wait days, cancellation, service loss, access fulfillment, and exceptions. Include open records and aging. A completed-case denominator can hide people still waiting unless the original cohort remains visible.

Review access separately from clinical fit

DOJ effective-communication guidance informs communication for covered entities. Language, disability, AAC, interpreter, mobility, or alternate-channel needs should appear as implementation work, not a negative fit or priority score.

Preserve authority and context

The BACB Ethics Code addresses nondiscrimination, client involvement, competence, resources, timeliness, referrals, and documentation for covered professionals. Qualified reviewers interpret differences in context and within scope.

A fictional offer audit

Maple Bend ABA reviews 60 eligible waitlist records across two time bands. Morning offers reach 18 of 24 eligible records, or 75%. After-school offers reach 20 of 36, or 55.6%. The gap triggers review of staffing, communication, access, geography, and rule application; the counts alone establish no cause.

Turn findings into controls

Document validated differences, likely mechanisms, missing data, owner, action, due date, and retest cohort. Review written criteria, staff discretion, inaccessible channels, capacity design, exceptions, and source fields. Preserve both improvement and unresolved disparity.

Write the audit protocol first

State the decision being audited, population, period, eligibility, outcome definitions, comparison groups, data sources, missing-data treatment, minimum cell handling, reviewers, privacy controls, and escalation. Freeze the protocol before viewing group results.

Document lawful purpose and qualified legal or access input where needed. The audit should answer a defined operating question without collecting unrelated sensitive information or creating a ranking of individual families or staff.

Validate the underlying records

Sample schedule states, offer dates, communication outcomes, access requests, staff assignments, exceptions, and reasons against source evidence. Measure missing, unknown, corrected, and inconsistent values by group. A difference driven by poor data should remain a data-quality finding.

Confirm that the same definitions and maturity windows apply to every comparison. Rebuild the result when a system or policy version changed during the period.

Choose a meaningful comparison

Compare people who were eligible for the same decision under the same policy version and had enough exposure time. Control operationally relevant factors such as service, time band, geography, staff configuration, and readiness without using them to erase the access barrier under review.

Report counts, rates, medians, ranges, and open records. Avoid interpreting a small or unstable difference as proof of unfair treatment.

Trace the decision path

When a gap appears, inspect the sequence from inquiry through review, contact, offer, acceptance, release, and delivery. Compare wait time at each state, discretion points, exception use, failed contacts, unavailable supports, and capacity constraints. Interview responsible roles and affected people through an appropriate process.

The analysis should identify plausible mechanisms and missing evidence. It should not assign unlawful intent or clinical cause from a percentage alone.

Protect sensitive audit data

Limit identifiable records to authorized reviewers, use coded analysis files, define retention, and control exports. Aggregate results only when the cell size and context protect people. Route privacy, employment, and legal questions to qualified owners.

Keep the operational source record intact. Corrections should preserve original value, corrected value, author, date, reason, and reports affected by the change.

Review exceptions and discretion

Compare exception requests, approvals, declines, conditions, and outcomes across similarly situated cohorts. Identify decisions that lack a source, use inconsistent criteria, or remain open without an owner. Test whether a neutral written rule has a different operational effect because access or capacity support is missing.

Revise authority, evidence, training, or escalation where discretion is necessary but poorly controlled. Preserve a route for case-specific qualified decisions.

Create and retest an action plan

For every validated issue, assign an owner, intervention, due date, affected process, baseline, success measure, and retest cohort. Actions may include accessible communication, capacity changes, clearer criteria, staff training, data corrections, or stronger review of exceptions.

Retest after enough records mature. Report improved, unchanged, worsened, and unresolved findings. A completed action item is different from evidence that scheduling outcomes changed.

Keep an audit trail

Archive the protocol, extracts, transformations, reviewers, decisions, actions, and retest result. Record who may access identifiable evidence and when working files should be removed.

Interpret differences without claiming a cause the audit cannot prove

Compare similarly situated records using factors defined before the review. A difference in offered times, wait, travel, cancellation, or staff changes may reflect policy, access barriers, configuration, clinical fit, payer state, family choice, geography, missing data, or inconsistent discretion. Report what the records support and name the additional evidence or qualified review needed. Avoid treating correlation as intent or a scheduling audit as a legal conclusion.

Use both process and outcome views. People may receive the same number of offers while one group gets shorter response windows or less usable options. Conversely, different outcomes may follow a documented, appropriate individual need. Trace samples through the full decision path, preserve exceptions, and include the people who never reached offer-ready status so readiness rules do not hide a barrier.

Repair a control and retest the same decision

Choose remediation that matches the finding: clearer criteria, accessible outreach, corrected data, supervisor review, monitored discretion, different allocation, staff training, or a redesigned operating constraint. Define the affected cohort, owner, effective date, interim safeguard, and test. Reapply the audit protocol after enough decisions mature and compare with the locked baseline.

Owner fairness-audit questions

  • Is the decision, cohort, comparison, period, policy version, and outcome defined before review?
  • Are source records accurate enough to support the comparison?
  • Do process measures include contact, response time, option usability, exceptions, and corrections?
  • Are legitimate clinical, payer, access, geography, and choice factors preserved without becoming blanket explanations?
  • Is sensitive information restricted while reviewers can still reproduce material findings?
  • Does each action have an owner and a same-method retest?

A fairness-audit example

An audit finds that families using an alternate communication channel receive fewer completed offers than portal users. The initial difference does not establish why. Reviewers trace reminder timing, delivery evidence, response windows, option completeness, schedule fit, access requests, and corrections. They discover that alternate-channel messages were sent later because the workflow required a manual handoff.

The practice adds an owned handoff deadline and monitoring queue, keeps the affected cohort visible, and retests the same measures after enough offers mature. It also checks whether the change improves option usability rather than merely increasing message count. Legal and accessibility conclusions remain with qualified reviewers, while the operational audit documents the supported process failure and its repair.

Related resources

Sources