An ABA generalization probe data sheet should show which conditions changed, which stayed constant, and the first response observed before programmed help. The matrix also needs a place for unavailable, unpresented, and invalid probes. Otherwise a pooled percentage can conceal that certain people, settings, materials, examples, or response forms were barely sampled.

Clinicians & ABA Professionals / Assessment and Treatment Planning.

This worksheet records a defined probe plan; it does not create a generalization program. The responsible qualified clinician defines the socially meaningful target, selects safe and relevant contrasts, specifies the procedure, and interprets the record with the client and team.

The evidence question comes before the grid

The current BCBA Test Content Outline, Sixth Edition distinguishes stimulus and response generalization, representative measurement, procedures that promote generalization, maintenance, contextual fit, procedural integrity, and data-based decisions. The BACB's current test-content outline page identifies the Sixth Edition BCBA outline and the Third Edition RBT outline now in use. These certification materials do not select a probe for a client or endorse this worksheet.

Frame a bounded question before opening the matrix. “Does the skill generalize?” cannot be scored as written. A reviewable version names the trained condition, intended contrast, response definition, ordinary supports, initial-response window, and required evidence. Performance with a new person does not automatically describe performance in a new setting or with a new exemplar.

Probe contract and version

Open a new version when the target, condition map, response definition, initial-response window, consequence, prompt rule, or ordinary access condition changes materially.

Generalization probe contractEntryClient or authorized case codeClient-selected goal and practical reasonTarget response and measurement definitionTraining condition and last teaching phaseProbe question and condition contrastEligible opportunity and start/end boundaryInitial-response windowCorrect, incorrect/no-response, invalid, unavailable, and not-observed codesPrompt, feedback, correction, and consequence rule during a probeOrdinary AAC, communication, mobility, sensory, and environmental supportsPlanned people, settings, materials, examples, and response formsClient choice, assent, dissent, break, and stop processSafety and escalation conditionsObserver, integrity, and agreement planAuthorized record location and correction ruleResponsible clinician and review trigger

The probe contract should state what would invalidate an initial-response observation. If the planned probe requires no programmed prompt before the initial window ends, an early prompt belongs in the integrity record and the affected observation should not be presented as an independent probe result.

Condition map

List one deliberate contrast per row when possible. If several dimensions change together, name the combination and avoid claiming which change mattered.

CellPersonSettingMaterials or exemplarInstruction or naturally occurring cueResponse form and ordinary accessTrained, novel, or mixedPlanned opportunitiesReason this contrast mattersABCD

Define “novel” locally instead of relying on the label. Someone may not know a particular staff member but may know the role. Materials can look different while sharing critical features. The entry should let another reviewer reconstruct the contrast that was actually observed.

Trial-level probe ledger

Enter the initial outcome before any later teaching, correction, or support. Preserve a probe label only when the recorded event met the current contract.

Date/timeCellOpportunityPerson, setting, and exemplar actually presentInitial responseCorrect?Programmed prompt before initial outcome?Available?Valid probe?Later teaching or correctionClient communication and contextObserver1Yes / noYes / noYes / noYes / no2Yes / noYes / noYes / noYes / no3Yes / noYes / noYes / noYes / no4Yes / noYes / noYes / noYes / no5Yes / noYes / noYes / noYes / no6Yes / noYes / noYes / noYes / no

An opportunity can be unavailable because the relevant activity did not occur, a planned person was absent, safety conditions changed, or the client declined. It can be invalid because the cue, response window, consequence, or support did not match the contract. Those categories are not incorrect responding.

Coverage before outcome percentages

Count the observation frame before interpreting performance.

Review quantityCalculationBoundaryPresentation coverageOpportunities actually presented / opportunities planned x 100How much of the plan occurredValid probe coverageValid probes / opportunities planned x 100How much of the plan supports the intended probe questionCorrect responding within valid probesIndependently correct initial responses / valid probes x 100Performance only within the valid observed subsetCell-specific correct respondingCorrect initial responses in the cell / valid probes in the cell x 100Describes that recorded condition cell onlyInvalid or contaminated rateInvalid observations / presented opportunities x 100Integrity signal, not a client-performance rate

Display every numerator and denominator beside its percentage. The ABA generalization probe data sheet keeps the planned frame and valid subset visible for that reason. One valid probe and a well-sampled cell may both show 100%, but they do not offer the same coverage or stability.

Condition-by-condition review

CellPlannedPresentedValidCorrect initial responsesIncorrect/no responseInvalid or contaminatedCell percentageContext or integrity noteABCDTotal

Review the cells before the total. If multiple condition dimensions changed in one cell, describe the result as performance under that combination. Do not assign causation to the person, setting, exemplar, or support without an appropriate design and evidence.

Fictional completed example

The example is invented for arithmetic review. Case G is not a real client, and the skill, condition contrasts, probe schedule, and interpretation are not recommendations.

Case G selected a fictional library checkout routine as a meaningful goal. The current target is completing the defined checkout response using the person's ordinary communication system. Four condition cells were planned with six opportunities each, for 24 planned probes.

CellFictional contrastPlannedPresentedValidIndependently correctIncorrect/no responseWhat remained visibleAFamiliar person, familiar setting, novel eligible exemplars66651One initial incorrect responseBNovel person, familiar setting, familiar exemplars66532One early prompt invalidated the initial probeCFamiliar person, novel setting, familiar exemplars65541One planned activity did not occurDNovel person, novel setting, novel eligible exemplars65422One access condition differed from the contractTotalFour defined condition cells242220146Two unpresented and two invalid observations retained

Presentation coverage is 22 divided by 24, or 91.7%. Valid probe coverage is 20 divided by 24, or 83.3%. Correct responding within the valid probes is 14 divided by 20, or 70.0%.

The cell results are also necessary. Cell A is 5 divided by 6, or 83.3%. Cell B is 3 divided by 5, or 60.0%. Cell C is 4 divided by 5, or 80.0%. Cell D is 2 divided by 4, or 50.0%. The reconciliation check is 5 + 3 + 4 + 2 = 14 correct responses and 1 + 2 + 1 + 2 = 6 incorrect or no-response outcomes across 20 valid probes.

The 70.0% pooled value does not establish broad generalization. It cannot make the four cells interchangeable, restore unpresented opportunities, resolve invalid observations, or identify the influential dimension. Nor does this record establish mastery, maintenance, causation, treatment effect, social validity, or client benefit.

What the research examples show

A descriptive review of mastery criteria and maintenance reported that generalization was assessed across different dimensions in the sampled studies, including stimuli, environments, therapists, behaviors, and clients. The variation supports naming the tested dimension rather than treating generalization as a single undifferentiated outcome.

A study of concurrent mands for items and location information arranged generalization observations across novel instructors, settings, items, and natural situations under study-specific definitions. A study that programmed common stimuli evaluated a specific strategy and noted other possible approaches. A simultaneous-prompting study used different people and stimulus features in its generalization sessions.

Across these studies, the condition and probe arrangement are part of the evidence rather than background detail. The reported participants, skills, settings, designs, and findings do not validate this worksheet or supply a universal matrix, sample size, criterion, or expected outcome.

The CASP ABA Practice Guidelines access page identifies Version 3.0 and explains its licensing conditions. This page does not reproduce that licensed document or claim that this worksheet complies with it. Consult the authorized current document and applicable requirements when relevant.

Communication, access, and client choice

Current ASHA AAC practice guidance discusses multimodal communication and access. A probe across settings should not quietly remove a person's usual communication mode, interpreter, device position, vocabulary, mobility support, sensory accommodation, corrective lenses, or prescribed equipment. If an access condition is intentionally part of the question, qualified reviewers must justify, authorize, and document it without compromising communication or safety.

Record assent, dissent, requests to pause, and client preferences as context. Do not create a novel-setting probe by withholding food, water, bathroom access, prescribed care, emergency communication, or another ordinary right. Family and staff participation should fit their authorized roles and should not turn an informal encounter into an undisclosed clinical test.

Ethics, records, and interpretation boundaries

The BACB Ethics Codes hub directs certificants to current materials. The Ethics Code for Behavior Analysts addresses competence, client welfare, consent, confidentiality, records, assessment, behavior-change programs, public statements, and data accuracy. Application depends on role and circumstances. This page is not ethics or legal advice.

Use an authorized record system for identifiable probe data. The current HHS Security Rule risk-analysis guidance explains that regulated organizations assess risks to electronic protected health information without prescribing one device, worksheet, app, retention period, or disclosure method. Follow the applicable access, correction, audit, export, backup, retention, and disposal rules.

Never pool across incompatible target versions or condition definitions. Preserve late entries, integrity deviations, observer changes, ordinary-support changes, and later teaching. If a probe becomes a teaching opportunity after the initial response, keep the initial probe record and label the later event instead of overwriting it.

Revision and exception log

DateContract versionOriginal entry or conditionCorrection, exception, or new versionReason and source evidenceAuthorized byOpportunities affectedSummary affected

Pause collection or interpretation when the response definition, condition contrast, access arrangement, prompt rule, valid-probe rule, client authorization, safety plan, responsible clinician, or record destination is unclear. Route distress, assent withdrawal, inaccessible communication, acute risk, material integrity deviation, suspected data alteration, privacy exposure, or work beyond competence through the appropriate clinical, emergency, organizational, privacy, security, compliance, legal, payer, or regulatory process.

Related resources

Sources