An ABA paired stimulus preference assessment data sheet should show more than which option was selected. A useful record also shows which pair was available, where each option appeared, how the person communicated, whether the trial was valid, and whether promised access occurred.

Use this copyable ABA paired stimulus preference assessment data sheet to prepare pairs, record direct choices, and interpret item-specific opportunity counts. The worksheet protects no selection, simultaneous selection, help, decline, changing preference, and ordinary access. It does not call a selected option a reinforcer.

Clinicians & ABA Professionals / Assessment and Treatment Planning.

Clinical and access boundary: A qualified team must decide whether paired stimulus fits the assessment question, person, setting, and available evidence. Preserve AAC, communication, sensory, motor, visual, hearing, literacy, language, mobility, and other ordinary access. Honor help, break, stop, refusal, distress, pain, fatigue, and withdrawal under the person’s plan. This worksheet does not authorize forced selection, blocked communication, deprivation, physical guidance, access restriction, or a treatment change.

Start with the question, not the array

A paired-stimulus assessment presents two defined candidates at a time and records the person’s selection under an agreed arrangement. Across a balanced schedule, each candidate can appear with every other candidate. The resulting pattern describes relative selection within that candidate set, procedure version, and context.

The result is narrower than “what the person likes.” It does not establish a permanent preference, a reason for choosing, or a reinforcing effect. Preference may change with time, health, access history, effort, setting, people, quality, duration, or the competing option. That possibility belongs in the interpretation.

The Sixth Edition BCBA Test Content Outline includes designing and evaluating preference assessments, representative measurement, and procedural-integrity measurement. Those competencies support careful planning. They do not prescribe paired stimulus for a particular person.

Assessment purpose, candidates, and access card

Begin with direct client input. A candidate list built only from staff convenience can produce orderly numbers that answer the wrong question. Rejected options matter too.

FieldCase-specific entryClient-informed assessment question and meaningful context.How the person nominated, rejected, or changed candidates.Candidate code, exact item or activity, quality, duration, and version.Ordinary access that remains available throughout.Accepted selection forms, including AAC or partner-assisted scanning.Familiarization and demonstrated contact with each candidate.Equal or intentionally different access after selection.Health, allergy, swallowing, sensory, motor, privacy, or safety review.Consent requirements and assent or dissent process.Qualified clinical owner and interdisciplinary collaborators.Effective date, review date, and superseded version.

The ASHA AAC Practice Portal describes multimodal systems and direct or indirect selection methods. The ASHA communication-access framework emphasizes asking about preferences and supporting participation. AAC, adapted positioning, extra processing time, or a familiar scanning method is access, not an artificial assessment prompt.

Define observable trial states before collecting data

Write definitions another trained observer could apply. “Chose A” is incomplete when the record does not say what counted as a choice or what happened after an ambiguous response.

ElementObservable definitionExampleInvalid, pause, or stop conditionPair available.Candidate quality and representation.Left and right position.Selection response.Simultaneous or multiple selection.No selection or unclear response.Help request.Decline, break, or stop.Access delivered as arranged.Valid trial.Invalid trial.

A missing device, inaccessible reach, unplanned cue, unavailable option, quality change, early staff recommendation, unequal promised access, or safety stop may make a trial invalid under the prospective rule. Keep the row and reason. Do not turn an invalid trial into a low-preference vote.

Build the pair and position schedule

For n candidates, the number of unordered pairs is n x (n - 1) / 2. Six candidates yield 6 x 5 / 2 = 15 pairs. Showing each pair twice can balance planned left and right exposure, creating 30 planned trials. That arithmetic is an example, not a required dose.

Pair IDCandidate left, run 1Candidate right, run 1Candidate left, run 2Candidate right, run 2Planned context or timeCompleted?ABABBAACACCAADADDAContinue for the approved candidate set.

Randomization does not repair unequal candidates. Check whether representations are equally understandable, items are similarly visible and reachable, and promised access is actually available. Equity comes before scheduling. If repeated trials are likely to change satiation, fatigue, or burden, record that concern rather than hiding it inside a rank.

Paired-stimulus trial ledger

One row represents one planned pair presentation. Record the direct selection before interpreting it.

Date/timePair ID/versionLeft optionRight optionAccess and choice methodValid? reasonSelectionSimultaneous/no selection/declineAccess deliveredEngagement or useLatencyBurden or feedbackObserveryes / noleft / rightyes / no / not applicable

Do not “resolve” an unclear or simultaneous response by choosing for the person. Apply the written repeat, clarification, invalid, pause, or stop rule. If a person changes a selection before access begins, preserve both events and the applicable rule.

The original Fisher and colleagues paired-stimulus comparison reported differentiated selections and better reinforcer prediction than its comparator for four children in a specific arrangement. A later paired-stimulus position-bias study shows that position-related responding can compromise interpretation for some participants. Neither study supplies a universal pair schedule or a client-specific conclusion.

Summarize item opportunities before percentages

Invalid trials can make exposure unequal. Count how often each item was actually available in a valid pair before dividing selections.

Candidate and versionPlanned pair appearancesInvalid appearancesValid item opportunitiesLeft validRight validValid selectionsSelection percentageAccess mismatchesNotesABCContinue.

Review questionNumeratorDenominatorBounded meaningTrial coveragevalid trialsplanned trialsInterpretable presentations under the written rulesItem selectionselections of one itemvalid opportunities containing that itemRelative selection in this set and versionLeft-position selectionvalid left selectionsvalid trials with one defined selectionPosition review signal onlyAccess deliveryarranged access deliveredvalid selections requiring accessImplementation, not preference strengthFidelitycorrectly implemented componentsobservable component opportunitiesCorrespondence with the written procedure

A raw selection count is misleading when one candidate appeared more often. Keep each item-specific denominator visible. A left/right selection difference is also not proof of bias. Option identities, pair difficulty, accessibility, and limited data may explain the pattern.

Fictional community-art-studio example

Riley is a fictional adult choosing among six community-art-studio activities. The team used codes A through F after Riley reviewed accessible samples and rejected two other candidates. A speech-generating device, adapted pointer, and ordinary break card remained available. The example tests the worksheet’s arithmetic, not the option set or clinical method.

Six candidates create 15 unordered pairs. Each pair was planned twice with positions reversed, for 30 planned trials. Three trials were invalid: AB began without the communication display, CD used a damaged sample, and EF included an early staff recommendation. Valid coverage was 30 - 3 = 27 trials, or 27 / 30 x 100 = 90.0%.

Each valid trial contributed two item opportunities, so there were 27 x 2 = 54 valid item opportunities. Because the three invalid pairs together contained every candidate once, each candidate had nine valid opportunities.

CandidateValid opportunitiesSelectionsItem-specific percentageA977 / 9 x 100 = 77.8%B966 / 9 x 100 = 66.7%C955 / 9 x 100 = 55.6%D944 / 9 x 100 = 44.4%E933 / 9 x 100 = 33.3%F922 / 9 x 100 = 22.2%Total5427One selection per valid trial

The selections reconcile: 7 + 6 + 5 + 4 + 3 + 2 = 27. Riley selected the left option 15 times and the right option 12 times. Those descriptive shares are 15 / 27 x 100 = 55.6% and 12 / 27 x 100 = 44.4%. The difference invites a position and pair review; it does not prove bias.

Access was delivered as arranged after 25 of 27 selections. The implementation result was 25 / 27 x 100 = 92.6%. The two mismatches require review because unequal access can change later responding. They cannot be ignored because the item totals look orderly.

No item is called a reinforcer. A preference assessment predicts or describes selection under its arrangement. Demonstrating a reinforcing effect requires a separate, qualified question and design.

Fidelity, burden, and interpretation record

ObservationCorrect versionOptions equivalent and accessiblePosition schedule followedNo steeringResponse window honoredAmbiguity rule followedDecline/stop honoredSelected access deliveredTrial code accurateBurden recordedCorrectOpportunities

The fictional fidelity sample reviewed 12 observations across 10 components. That produced 12 x 10 = 120 component opportunities. Reviewers scored 113 correct and seven mismatches, with 113 + 7 = 120. Descriptive fidelity was 113 / 120 x 100 = 94.2%.

Inspect each mismatch. Missing AAC, steering, inaccessible placement, or failure to honor withdrawal is not repaired by a high aggregate. Fidelity says whether the written arrangement occurred. It does not establish appropriateness, preference, validity, benefit, or safety.

Interpretation domainDirect evidenceAlternative explanation or limitNeeded follow-upOwnerCandidate relevance and changing preferencePosition and pair effectsFamiliarization and representationAccess delivery and engagementCommunication, sensory, motor, or health factorsBurden, fatigue, distress, or privacyGenerality to the intended settingSeparate reinforcer-effect question, if warranted

The SPADS decision model discusses practitioner considerations when choosing among preference-assessment formats. A video-based preference-assessment study highlights the relevance of representation and discrimination skills. A recent paired assessment of social interaction found task-specific limits in two participants. Use these sources as bounded evidence, not as a universal prediction.

Decision and version record

Review fieldEntryAssessment question, candidate set, representations, and version.Client nomination, rejection, communication, help, and withdrawal evidence.Pair schedule, valid coverage, and item-specific opportunities.Position, access-delivery, engagement, and burden findings.Fidelity mismatches and validity limits.What the results do and do not support.Retain / revise candidates / change format / pause / stop question.Qualified owner, rationale, effective date, and next review.

Preserve the prior version and raw rows. A changed candidate set, representation, access duration, communication method, setting, or pair schedule creates a different condition. Do not retroactively rewrite earlier selections.

The BACB Ethics Code and CASP ABA Practice Guidelines access page provide professional context. An implementation-training study illustrates that step-level implementation can be measured. None replaces competence, consent, assent, privacy, interdisciplinary review, or applicable local requirements.

Related resources

Sources

The studies use small samples, specific stimuli, particular access arrangements, and defined contexts. They support direct measurement and careful qualification. They do not establish one best format, candidate set, rank, cutoff, reinforcer, treatment, or expected result.