An ABA data correction log should make a change understandable without pretending the earlier record never existed. A later reviewer should be able to identify the original entry, see what changed and why, locate the evidence, and determine who reviewed it. The same trail should identify every graph, decision, export, or disclosure that may have used the earlier value.

Clinicians & ABA Professionals / Data, Outcomes and Clinical Decision-Making.

Use this worksheet while the original record remains intact. It is not a universal amendment policy or permission to change a surprising result. Qualified clinical, health-information, privacy, compliance, legal, and payer reviewers must determine the applicable process.

Start with the record, not the preferred answer

The current BCBA Test Content Outline, Sixth Edition includes observable definitions, measurement-system evaluation, data interpretation, and procedural-integrity measurement. The BACB's test-content outline page identifies the current outline. These materials do not prescribe a correction form or decide whether an amendment is justified.

Before editing a display or recomputing a percentage, hold the current view steady and define the review question. Name the record, definition, observation window, storage location, author or system, and review trigger. Separate a suspected error from a confirmed discrepancy. Memory that differs from documentation is a review signal, not proof that the stored value is wrong.

Correction-review contractEntryClient or authorized case codeRecord type and stable record identifierService, observation, or reporting periodMeasurement definition and versionOriginal author or system, timestamp, and time zoneCurrent storage location and access ownerReview trigger and date discoveredSuspected field, value, or linkageEvidence expected before dispositionAuthorized reviewer and escalation routePrivacy, security, retention, disclosure, billing, payer, legal, or incident route

If the original record cannot be located, stop the amendment workflow and escalate the provenance problem. Reconstructing a value from memory and labeling it a correction can create a second unsupported record.

Preserve an immutable view of the original

The BACB's Ethics Codes page provides the current code resources, and the Ethics Code for Behavior Analysts addresses accurate work, documentation, confidentiality, records, and accountability within its scope. It does not determine an organization's record architecture or the law governing a specific record.

Capture the original value and context before proposing an amendment. Preserve the existing audit event, version identifier, or export when available. Avoid shadow copies containing more protected information than the review requires. The goal is a traceable link.

Original-record snapshotEntryStable record ID and versionExact field or row under reviewOriginal value exactly as storedOriginal unit and denominatorOriginal timestamp, author/system, and definition versionLinked note, raw observation, or device recordExisting graph, report, export, or clinical decisionExisting disclosure, claim, or payer submissionPreservation method and authorized locationSnapshot completed by and timestamp

A screenshot may omit values, version history, or context. A copied row may omit how the application calculated a display. Preserve the least expansive evidence that still reproduces the original state.

Classify the proposed change

Several very different problems can produce an apparently wrong value. Naming the candidate category helps reviewers ask the right questions and prevents a numerical correction from silently becoming a clinical reinterpretation.

  • Transcription or entry error: Does source evidence show that the stored code differs from the contemporaneous source?
  • Wrong identity or linkage: What establishes the client, goal, session, observer, or timestamp, and did the error expose another person's information?
  • Wrong unit or denominator: Which definition and eligibility rule applied during collection?
  • Duplicate record: Are two entries the same event or only similar events?
  • Transformation error: Can the source payload, transformation version, and destination value be reconciled?
  • Late entry or addendum: Is new information being added rather than an earlier value replaced?
  • Clinical reinterpretation: Is the team changing the definition or conclusion rather than fixing an entry?
  • Unresolved discrepancy: What evidence is missing, and how will uncertainty remain visible?

For guidance focused on the documentation distinctions, see Late Entries, Addenda and Corrections in ABA Records: What Clinicians Should Do. This page supplies the row-level working log; it does not replace that broader policy discussion.

Use an item-level amendment ledger

Record the proposed value beside the original. Keep reason, evidence, authority, and downstream impact separate. “Corrected” is not a sufficient reason, and approval without reviewer identity and time is not a useful audit trail.

ItemStable record and fieldOriginal valueProposed valueCategoryReason and evidenceProposed by / timeReviewer disposition / timeEffective display ruleDownstream action1Approve / decline / unresolved2Approve / decline / unresolved3Approve / decline / unresolved4Approve / decline / unresolved

Evidence may support an amendment, leave the discrepancy unresolved, or show that the original was reasonable under the definition then in force. The reviewer should not convert uncertainty into certainty simply to close the task.

The CASP ABA Practice Guidelines Version 3 access page identifies the public guideline version, intended audience, and licensing conditions. This worksheet does not reproduce licensed guideline text and should not be described as a CASP form.

Review the downstream footprint

A one-cell amendment can affect many outputs. Trace graphs, session summaries, progress reports, treatment reviews, communications, quality reports, exports, authorization packets, and claims-related artifacts. An earlier output may need an annotation, replacement, or documented no-action decision.

Downstream-impact mapUsed original?Material effect?Required actionOwnerCompleted / evidenceRaw-data viewYes / no / unknownYes / no / unknownClinical graph or dashboardYes / no / unknownYes / no / unknownSession or progress noteYes / no / unknownYes / no / unknownTreatment-plan review or clinical decisionYes / no / unknownYes / no / unknownClient or caregiver communicationYes / no / unknownYes / no / unknownSupervisor or quality reviewYes / no / unknownYes / no / unknownAuthorization or utilization-review packetYes / no / unknownYes / no / unknownClaim, encounter, invoice, or payer reportYes / no / unknownYes / no / unknownExternal export or analytics setYes / no / unknownYes / no / unknown

If the original value influenced a decision, use How to Handle Corrected ABA Data After a Clinical Decision. A changed datum neither invalidates nor preserves the earlier decision automatically.

A fictional worked correction

The following arithmetic is synthetic. It is not a client record, performance target, clinical threshold, payer rule, or statement that one type of response is preferable.

A fictional teaching record contains eight eligible break-request opportunities. The version frozen after the observation shows five independent responses and three prompted responses.

Original percentage: 5 independent / 8 eligible opportunities × 100 = 62.5%.

During a scheduled source-record check, the reviewer finds contemporaneous evidence that opportunity 6 was independently correct but was transcribed into the prompted column. The evidence fits the same definition version and observation window. The proposed amendment is to reclassify that one entry; it does not add or remove an opportunity.

OpportunityOriginal codeSource evidence reviewedDispositionAmended code1IndependentNot in disputePreserveIndependent2PromptedNot in disputePreservePrompted3IndependentNot in disputePreserveIndependent4IndependentNot in disputePreserveIndependent5PromptedNot in disputePreservePrompted6PromptedContemporaneous source supports independent response under the same definitionApprove amendmentIndependent7IndependentNot in disputePreserveIndependent8IndependentNot in disputePreserveIndependent

The amended view contains six independent and two prompted responses.

Amended percentage: 6 independent / 8 eligible opportunities × 100 = 75.0%.

Change in displayed percentage: 75.0% − 62.5% = +12.5 percentage points.

The denominator remains eight: 6 + 2 = 8. The log preserves the original 5 + 3 = 8 composition, changed row, evidence, approver, timestamps, and new display. It records whether 62.5% appeared in a graph or decision note. The mathematical increase does not prove improvement, treatment effectiveness, observer accuracy, or client benefit.

Worked-example audit entryValueStable fictional recordEXAMPLE-ONLY-08Original state5 independent; 3 prompted; 8 eligibleOriginal display5 / 8 = 62.5%Changed itemOpportunity 6: prompted → independentAmendment basisContemporaneous evidence under unchanged definitionAmended state6 independent; 2 prompted; 8 eligibleAmended display6 / 8 = 75.0%Display delta+12.5 percentage pointsPrior output checkGraph and decision note reviewed; disposition recordedClinical conclusionNone from this arithmetic alone

Amendment and privacy boundaries

The HHS HIPAA Privacy Rule overview explains the federal privacy framework for covered entities and business associates. The current 45 CFR 164.526 addresses an individual's right to request amendment of protected health information and, when an amendment is accepted, identifies appending or linking the amendment as part of the regulated process. An HHS health-information correction discussion describes the importance of preserving and communicating corrections in health-information exchange.

Not every numerical QA correction is a formal HIPAA amendment request, and this worksheet cannot determine whether HIPAA applies. Designated record-set, amendment, denial, notification, retention, and disclosure procedures need qualified review. HHS explains that business associates may have amendment-related obligations under the rule and their contracts; reviewers still must examine the actual relationship.

Limit access to what the review requires. HHS Security Rule risk-analysis guidance provides context but does not approve a spreadsheet, storage location, email route, or retention period. Keep protected information out of unmanaged copies.

Review measurement quality without overstating it

A correction may reveal an entry, definition, workflow, or reliability problem. The peer-reviewed discussion of data reliability and treatment integrity in applied behavior analysis explains why data quality and implementation information matter. It supplies neither a universal threshold nor authority for retrospective alteration.

Ask what the discrepancy teaches the team:

  • Was the original definition observable and available at the point of collection?
  • Did the interface make mutually exclusive codes clear?
  • Did the observer have the access or accommodation needed to record accurately?
  • Did a transformation, import, or export change units or labels?
  • Did one row change, or does the same pattern appear elsewhere?
  • Can the problem be prevented without increasing burden or surveillance unnecessarily?

Record any prospective improvement separately from the amendment. Updating a definition or interface for future data does not rewrite past records collected under an earlier version.

Reviewer disposition

  • Is the original record preserved and linked?
  • Are the changed field and old and new values explicit?
  • Is the reason supported by contemporaneous evidence?
  • Does the same measurement definition apply?
  • Are authorship and both timestamps preserved?
  • Is the amendment approved, declined, or unresolved?
  • Are graph and summary effects calculated transparently?
  • Are prior decisions and communications reviewed?
  • Are payer, billing, privacy, legal, or incident routes implicated?
  • Were required people informed through an authorized route?
  • Is preventive follow-up separate from record amendment?

Possible dispositions include approval, reasoned decline, unresolved status, an evidence request, or escalation to the formal amendment or incident process. A documented “no change” disposition still preserves the review trail.

Close the trail, not the question

  • Original remains accessible under the authorized retention process.
  • The amendment or linked addendum is visible as required.
  • Author, reviewer, dates, and time zones are recorded.
  • Affected calculations were independently recomputed.
  • Graphs and reports show the appropriate version or annotation.
  • Earlier decisions were reviewed without hindsight rewriting.
  • Required notifications or disclosures were completed.
  • Privacy, security, payer, billing, and legal escalations were resolved.
  • Preventive action has an owner and prospective effective date.
  • Uncertainty and unresolved items remain visible.

The log is complete when a reviewer can follow the original record through disposition and downstream action. Completion does not certify clinical validity, treatment success, payment, or compliance.

Related resources

Sources