MSWO Preference Assessment: Procedure, Variants, and Decision Logic for BCBAs and RBTs

Origins and the case for MSWO over MSW and paired stimulus

MSWO was developed as a modification of the multiple-stimulus-with-replacement (MSW) procedure that addressed two practical problems: MSW lets a learner repeatedly select the same dominant item, producing a hierarchy that identifies the top stimulus but poorly differentiates mid- and lower-ranked items, and full paired-stimulus assessments require n(n−1)/2 trials, which becomes prohibitive when arrays grow beyond five or six items Melanson et al. (2023). By removing each chosen item from the array and re-presenting the remainder, MSWO forces the learner to discriminate among an ever-shrinking set of options, yielding a full rank order from a small number of trials — typically a 5- to 7-item array, three to five rounds, in 5–15 minutes total Melanson et al. (2023) Conine et al. (2021). The procedure has since become the most heavily studied rank-order format in the applied literature, with format-comparison studies treating MSWO as the default reference against which newer variants are validated Conine et al. (2021) Brodhead et al. (2017).

Intra-session stability is weaker than practitioners assume

The single most operationally important finding in the recent MSWO literature is that within-session stability is far from perfect. Melanson and colleagues re-analyzed 40 MSWO assessments across 17 children with autism using Spearman rank correlations between consecutive stimulus-reset rounds and found that round-to-round preference correspondence held only about 60% of the time, with rankings shifting meaningfully across resets in roughly 40% of within-administration comparisons; edible stimuli were noticeably less stable than leisure items in this pediatric sample Melanson et al. (2023). The operational implication is direct: a single round is not a definitive hierarchy, and round one is not a reliable proxy for the assessment as a whole. Run at least two — preferably three — rounds before adopting a final hierarchy, and expect higher within-session shifts whenever the array contains edibles, particularly mid-session when satiation effects accumulate Melanson et al. (2023).

Castillo and colleagues add a complementary caution. In a small replication across four adults with intellectual and developmental disabilities, one participant systematically saved the best for last during both edible and leisure MSWOs — selecting a high-preference item in the final position — and a follow-up progressive-ratio reinforcer assessment confirmed that last-selected item as a strong reinforcer Castillo et al. (2022). This pattern is not the norm, but it is real and undetectable from rank order alone. Validate top items with an independent reinforcer assessment rather than treating "selected first" as definitionally equivalent to "most reinforcing" Castillo et al. (2022) (Kronfli et al., 2024).

Brief MSWO: how short is short enough?

The standard recommendation has long been three full sessions to produce a stable hierarchy. Conine and colleagues challenged that defaulting assumption directly: across 157 MSWOs administered to 49 children and adults with developmental disabilities, hierarchies generated by single-session and two-session MSWOs correlated strongly with the three-session reference, but neither abbreviated format produced identical preferred-item rankings Conine et al. (2021). The one-session brief MSWO is "reasonably accurate but not perfect" — useful for screening, mid-session reinforcer adjustment, or rapid in-the-moment validation, but not a substitute for the three-session reference when stakes are high (BIP reinforcer selection, skill-acquisition program design, or any context where a wrong rank cascades into ineffective treatment) Conine et al. (2021).

Brodhead and colleagues extended brief MSWO into a video format that withholds access to the chosen item — useful when delivering items contingent on selection is impractical (sanitation concerns, items that cannot be repeatedly accessed, video stimuli that would consume session time if played fully). Across four children with autism, a 5-trial video-based MSWO-NO produced item rankings with moderate-to-strong correspondence to the standard MSWO with access Brodhead et al. (2017). The 5-trial no-access video MSWO is therefore a defensible substitute when contamination, sanitation, or stimulus-delivery constraints make standard MSWO impractical — but, as with all brief variants, it should not be the default for high-stakes reinforcer selection without a downstream reinforcer probe Brodhead et al. (2017).

Tablet, web, and digital MSWO variants reproduce tangible accuracy

The MSWO procedure has been reproduced across multiple alternative-modality platforms with strong validity evidence when prerequisite skills are intact Hoffmann et al. (2019) Curiel et al. (2020). Hoffmann and colleagues compared a tablet MSWO using app-icon pictures to a traditional 3-D tangible MSWO across three boys aged 4–8 with ASD; rank-order correlations averaged 0.94, and both formats were rated as easy to implement Hoffmann et al. (2019). The catch, repeated across the alternative-modality literature, is that learners must first reliably match the icon picture to the corresponding app — without that prerequisite, picture-based MSWO produces silent invalid hierarchies Lill et al. (2021). Curiel and colleagues developed and described a free, browser-based MSWO tool that lets practitioners upload videos, set parameters, run the assessment via clickable thumbnails, and export selection data to spreadsheet — digitizing the entire MSWO workflow and standardizing implementation in the process Curiel et al. (2020). For digital reinforcers (videos, app icons, streaming content), tablet- or web-based MSWO is now the procedurally standard format; for tangible reinforcers, the choice is more about staff workflow than validity Hoffmann et al. (2019).

MSWO has also been validated in non-pediatric, non-clinical populations. Jarmolowicz and colleagues used a paper MSWO array of photographs with 41 college students to rank potential sexual partners and then validated those rankings against a hypothetical purchase task; participants' purchasing decisions mirrored the MSWO hierarchy, with highest valuations for MSWO-identified high-preference partners and lowest for low-preference Jarmolowicz et al. (2016). The methodological point: MSWO produces criterion-valid hierarchies in adult decision-making contexts when direct consumption is impractical or inappropriate, providing a brief, non-consumable alternative to traditional preference assessment Jarmolowicz et al. (2016).

MSWO with adults who have neurocognitive disorder: limits and modifications

MSWO does not transfer cleanly to all populations. Ford and colleagues compared single-stimulus and MSWO assessments with caregiver ratings in six community-residing older adults with dementia; rank-ordered MSWO results had limited correspondence with caregiver rank orders, and SS assessments were both faster to run and produced fewer implementation errors than MSWO Ford et al. (2022). The MSWO scanning-and-selection requirement defeats some clients with cognitive impairment; rank-order discrimination collapses when scanning capacity is limited, and the resulting hierarchy ceases to mean what the procedure assumes it means. For older adults with NCD and any client with limited choice repertoire, switch to single stimulus or use MSWO only with significant procedural modifications (small arrays, fewer rounds, paired with caregiver confirmation) Ford et al. (2022).

Staff training: brief packages reach mastery, but not all packages work equally

The procedural-fidelity research is largely encouraging but recently more nuanced. Arnal Wishnowski and colleagues showed that an online self-instruction package (manual plus video modeling) raised MSWO implementation accuracy to 94% for university students and 87% for developmental-services staff, with gains maintained 7–17 days later — supporting computer-based training as a cost-effective option when in-person supervision is unavailable (Wishnowski et al., 2018). O'Handley and colleagues replicated this with brief Behavioral Skills Training (BST) for three preservice school psychologists in a simulated classroom, improving accuracy from 9–19% in baseline to 73–93% after BST and maintaining ≥82% fidelity one week later O’Handley et al. (2021). Bartle and colleagues compared video-modeling-with-voice-over (VMVO) training using exemplars-only versus exemplars-plus-nonexemplars across three adult staff; exemplars-only VMVO rapidly raised MSWO procedural integrity to about 90%, performance was maintained at five-week follow-up, and adding nonexemplars did not further improve outcomes — supporting the most-stripped-down version of VMVO as a defensible default when training resources are limited (Bartle et al., 2025).

But not every brief package works. Sherman and colleagues evaluated a more elaborate Articulate-Rise interactive computer training (ICT) module — text, graphics, video models, knowledge checks — across nine human-service staff and found only small-to-moderate gains; every participant except one still required supplemental in-vivo feedback or modeling to reach mastery (Sherman et al., 2025). The implication is operational rather than damning: ICT is useful as a refresher or partial-training adjunct, but supervisors should plan brief corrective-feedback rounds to bring staff to 90% procedural fidelity rather than treating a completed module as evidence of mastery (Sherman et al., 2025).

MSWO is under-used in routine practice

Morris, Conine, and colleagues' survey of 260 BCBAs and RBTs delivering ABA services found that only 26% reported using formal preference assessments such as MSWO to identify reinforcers; the majority relied on asking parents or clients or on informal observation (Morris et al., 2024). Self-report likely overstates actual MSWO use, so the real rate is probably lower. The practice-pattern data complicate the assumption that MSWO drives day-to-day reinforcer selection in EIBI and adult services; in practice, the brief MSWO is what experienced clinicians actually run mid-session, and the institutional adoption gap is the chief barrier — not procedural difficulty (Morris et al., 2024) (Wishnowski et al., 2018).

MSWO rank does not always predict reinforcer efficacy under instruction

Kronfli and colleagues ran MSWOs (three iterations per participant, ranks averaged) with two preschoolers to rank fruit/vegetable versus salty/sweet food reinforcers and then compared sight-word acquisition under each category; learning speed did not consistently track MSWO rank — fruits and vegetables sometimes produced faster acquisition even when they ranked below salty/sweet items on the MSWO (Kronfli et al., 2024). The take-home is the same as the broader preference-versus-reinforcement caution: MSWO produces a choice rank, which is correlated with but not identical to reinforcer efficacy under instructional contingencies. Treat the top one or two MSWO items as candidates for instruction, then validate with brief in-task reinforcer probes during initial sessions (Kronfli et al., 2024) Melanson et al. (2023).

The MSWO literature is dominated by single-subject experimental designs and methodology papers, with one survey, one decision-making model, and one quasi-experimental adult-decision-making validation Melanson et al. (2023) (Morris et al., 2024).

Single-subject experimental designs. Most of the MSWO corpus sits here. Melanson and colleagues (n=17 children with autism, 40 MSWOs) anchor the intra-session stability claim — round-to-round correspondence ~60%, edibles less stable than leisure Melanson et al. (2023). Conine and colleagues (n=49, 157 MSWOs) compare 1-, 2-, and 3-session MSWO hierarchies and document strong but imperfect correspondence Conine et al. (2021). Castillo and colleagues (n=4 adults with IDD) document the "save the best for last" pattern via paired MSWO + progressive-ratio reinforcer assessments Castillo et al. (2022). Brodhead and colleagues (n=4 children with autism) validate the brief no-access video MSWO against standard MSWO with access Brodhead et al. (2017). Hoffmann and colleagues (n=3 boys with ASD) document a 0.94 rank-order correlation between tablet app-icon MSWO and tangible MSWO Hoffmann et al. (2019). Kronfli and colleagues (n=2 preschoolers) show that MSWO rank does not always predict instructional reinforcer efficacy (Kronfli et al., 2024). Ford and colleagues (n=6 older adults with NCD) document the limits of MSWO for cognitively impaired populations and the relative advantages of single stimulus Ford et al. (2022). Staff-training SCED includes Arnal Wishnowski and colleagues (n=10) on computer-aided self-instruction (Wishnowski et al., 2018), O'Handley and colleagues (n=3 preservice school psychologists) on brief BST O’Handley et al. (2021), Bartle and colleagues (n=3 novice staff) on exemplar-only VMVO (Bartle et al., 2025), and Sherman and colleagues (n=9 staff) on interactive computer training as a complicating data point (Sherman et al., 2025).

Methodology and decision-making papers. Curiel and colleagues describe a free, browser-based MSWO tool that digitizes the entire workflow Curiel et al. (2020). Lill, Shriver, and Allen's SPADS articulates the prerequisite-skill check that gates pictorial MSWO Lill et al. (2021). Poling and colleagues' conceptual paper on motivating operations and discriminative stimuli provides the framework for pre-session MO state assessment Poling et al. (2020).

Quasi-experimental. Jarmolowicz and colleagues (n=41 college students) validated MSWO criterion validity by showing that purchase-task valuations mirrored MSWO-identified rank order in an adult decision-making context Jarmolowicz et al. (2016).

Survey. Morris, Conine, and colleagues' survey of 260 BCBAs and RBTs is the largest practice-pattern data point and complicates the assumption that MSWO drives day-to-day reinforcer selection — only 26% reported using formal preference assessment (Morris et al., 2024).

Bottom line. Convergent SCED evidence supports MSWO as a valid, efficient rank-order procedure that reproduces across tangible, picture, video, tablet, and web platforms when prerequisites are intact Hoffmann et al. (2019) Brodhead et al. (2017) Curiel et al. (2020); complicating evidence — intra-session instability, save-the-best-for-last patterns, MSWO rank ≠ instructional reinforcer efficacy, and population limits with cognitive impairment — converges on the recommendation to run multiple rounds and validate top items with a downstream reinforcer probe rather than treating MSWO output as definitive Melanson et al. (2023) Castillo et al. (2022) (Kronfli et al., 2024) Ford et al. (2022).

The format-selection question is rarely "MSWO or paired stimulus?" in isolation; it is "what is the client's repertoire, what is the safety profile, and what is this assessment for?" The decision tree below reflects the modern literature's recommendations for when to choose MSWO, when to default to a different format, and when to switch variants mid-protocol.

1. Default rank-order assessment for clients with intact scanning and selection skills. Run a standard 5–7 item, 3-round MSWO. This is the most-validated rank-order format with the densest variation literature Melanson et al. (2023) Brodhead et al. (2017) Hoffmann et al. (2019).
2. Older adult with neurocognitive disorder, severe disability, or limited choice repertoire. Do not default to MSWO. Switch to single stimulus, which produces fewer implementation errors, is faster to run, and predicts engagement better than rank-order formats for these populations Ford et al. (2022).
3. Tangible-maintained problem behavior or high-risk safety profile. MSWO withholds preferred items between trials and reliably evokes higher problem-behavior rates than free operant; default to free operant or response-restriction free operant for this population. Reserve MSWO for after problem behavior has been brought to acceptable levels through treatment.
4. Time-pressed in-session reinforcer adjustment. Run a brief (single-round, 1–3 minute) MSWO, accepting that the resulting rank will diverge from a multi-session MSWO in roughly 40% of cases Conine et al. (2021). This is the format experienced clinicians actually use mid-block (Morris et al., 2024).
5. High-stakes reinforcer selection (BIP, skill-acquisition program, transition planning). Run the full three-session MSWO and validate the top one or two items with a downstream reinforcer test. A single-session brief MSWO is not adequate when a wrong rank propagates through weeks of treatment Conine et al. (2021) (Kronfli et al., 2024).
6. Items will be delivered digitally (videos, apps, streaming). Use a tablet- or web-based MSWO with verified picture-matching prerequisites; both formats reproduce tangible-MSWO accuracy when prerequisites are intact Hoffmann et al. (2019) Curiel et al. (2020).
7. Items cannot be delivered contingently after selection (sanitation, single-use, time-prohibitive video). Use a brief no-access video MSWO; rank correspondence with standard MSWO is moderate-to-strong Brodhead et al. (2017).
8. Whole-class reinforcer identification. Use a group brief MSWO via Plickers cards or a similar simultaneous-response tool; group brief MSWO has been shown to produce identical top edible reinforcers to individual four-choice MSWO across elementary classrooms (covered in detail on the broader preference-assessment overview linked under Related Topics).
9. Picture or icon arrays without verified matching prerequisites. Do not run pictorial MSWO. Probe object-to-picture and picture-to-object matching first; if the learner fails those probes, choose tangible-item MSWO or train prerequisite matching skills Lill et al. (2021).
10. Adult populations where direct consumption is impractical or inappropriate (vocational planning, adult choice contexts, social-partner preference). MSWO with photograph or video stimuli has been validated as criterion-valid for adult rank-order decisions when paired with a downstream task probe Jarmolowicz et al. (2016).

Clinic and outpatient

Clinic is the densest setting for MSWO research and the one with the cleanest procedural fit. Standard 5-item, 3-round MSWO with tangibles and brief contingent access is the default Melanson et al. (2023). Concurrent-operants reinforcer validation is procedurally cheap in clinic — two-chair setups, alternating-treatments designs, and tablet-based video delivery are all standard Hoffmann et al. (2019). The web-based MSWO tool offers a one-click workflow that eliminates paper data sheets and exports selections to spreadsheet, which simplifies documentation when multiple BCBAs and RBTs share a clinical caseload Curiel et al. (2020). Build pre-session MSWO into the start of every long block and brief MSWO checks into every 30–40 minute mid-block engagement drop.

School and special education

Classroom MSWO diverges from clinic in two ways: per-student time is short, and group-level reinforcer identification often matters more than individual hierarchies for class-wide token systems and group rewards. For individual student programming, a brief tablet- or web-based MSWO is more feasible than a 15-minute paper assessment, and BST or VMVO training packages bring teaching staff and school psychologists to MSWO mastery in one session O’Handley et al. (2021) (Bartle et al., 2025). School-based MSWO also feeds IDEA-mandated transition planning by surfacing vocational and leisure preferences that drive IEP goal selection — covered in detail on the broader preference-assessment overview linked under Related Topics.

Home and parent-administered

Home-based MSWO leans on simpler formats and caregiver implementation. The 5- to 10-minute brief MSWO, in-person or via video on a tablet, is the practical default; full multi-session MSWO is generally too time-intensive for typical home routines. Computer-aided self-instruction packages reach 87% accuracy for direct-care staff with 7- to 17-day maintenance, offering a workable training path for parents and home-based RBTs without on-site supervision (Wishnowski et al., 2018). The web-based MSWO tool's video-upload feature allows parents to identify preferred YouTube clips, app screens, or family-specific stimuli that a clinic-based item array would never include Curiel et al. (2020).

Telehealth-implemented MSWO

Telehealth MSWO is now a defensible standard. The web-based MSWO tool runs in any browser; tablet picture MSWO produces rank-order correlations of 0.94 with tangible MSWO when picture-matching prerequisites are intact; brief BST and VMVO packages can themselves be delivered via telehealth real-time feedback to bring direct-care staff and parents to MSWO mastery within a few sessions Curiel et al. (2020) Hoffmann et al. (2019) O’Handley et al. (2021). The remaining constraints are practical (internet connectivity, prerequisite verification, contingent delivery of digital reinforcers) rather than procedural — telehealth MSWO is no longer a methodological compromise.

Adult disability services and residential

Adult and residential settings concentrate two challenges: cognitive impairment that defeats rank-order formats for some clients, and dispersed staff who cannot all be trained the same way. For older adults with NCD, MSWO produces poorer caregiver–assessor agreement than SS, requires more training, and runs slower; default to single stimulus and use MSWO only with significant procedural modifications Ford et al. (2022). For higher-functioning adults in vocational or community settings, MSWO of photograph or video stimuli (job tasks, leisure activities, partner choices) is criterion-valid against downstream choice probes Jarmolowicz et al. (2016). VMVO and computer-aided training packages give residential teams a way to install fidelity across dispersed staff without on-site travel (Bartle et al., 2025) (Wishnowski et al., 2018).

The MSWO is procedurally simple to describe and easy to err on. The step-by-step below reflects the protocol used across the modern MSWO literature; specific parameter choices (array size, access duration, ITI) have small but real effects on the resulting hierarchy and should be held constant across re-administrations Melanson et al. (2023).

1. Generate the candidate item list. Use a caregiver/teacher interview, a RAISD-style open-ended questionnaire, or direct observation to identify 5–8 candidate items. Indirect data are inputs only — they generate the array, not the hierarchy Lill et al. (2021).
2. Verify prerequisites for picture or video formats. If using pictures, app icons, or video, confirm that the learner can match each picture to its corresponding tangible item or comprehend the video before running the assessment; missing this prerequisite produces silently invalid hierarchies Lill et al. (2021) Hoffmann et al. (2019).
3. Confirm motivating-operation state. Run the MSWO at a time when the learner is not currently satiated on edibles or recently saturated with leisure items; pre-session MO checks (deprivation/satiation note in the data sheet) help interpret later rank shifts Poling et al. (2020).
4. Set the array. Place all items simultaneously in front of the learner in an arc, equidistant, with positions randomized between rounds. Use a 5–7 item array; smaller arrays under-discriminate, larger arrays risk scanning overload and cognitive fatigue.
5. Deliver the prompt. A neutral discriminative cue ("Pick one") is standard. Avoid model prompts ("Pick this one") or any verbal commentary on items, which can bias selection.
6. Score the selection. Record which item the learner picks first. Provide brief contingent access (typically 30 seconds for tangibles, single consumption for edibles, full play for short videos) — or, for the no-access variant, score the selection and immediately remove the item without delivery Brodhead et al. (2017).
7. Remove the selected item. Permanently remove the chosen item from the array for the remainder of this round.
8. Re-present the remainder. After a brief inter-trial interval (5–10 seconds), re-arrange the remaining items into a new randomized arc and re-prompt. Continue until all items have been selected, or until the learner persistently refuses to choose.
9. Handle "no choice" responses. If the learner does not select within a defined latency window (typically 30 seconds), prompt once more; if there is still no choice, score that round and either terminate or move to the next round. Do not repeatedly re-prompt — a non-response is data, not a procedural failure Melanson et al. (2023).
10. Repeat for multiple rounds. Run 3–5 rounds. The hierarchy is the average rank order across rounds, the mode across rounds, or the rank order from the final round (each defensible; pick one, document it, hold it constant across re-assessments) Melanson et al. (2023) Conine et al. (2021).
11. Compute and document. A modern web-based MSWO tool can auto-record selections and export to spreadsheet Curiel et al. (2020). If using paper data sheets, compute mean rank or modal rank per item and store with the array configuration, presentation parameters, and assessor identity Curiel et al. (2020).
12. Validate the top items with a downstream reinforcer test. Before committing the top one or two items to a BIP or skill-acquisition plan, run a brief concurrent-operants reinforcer probe under conditions resembling the actual treatment schedule (Kronfli et al., 2024) Castillo et al. (2022).

Brief MSWO (single-round, 1–3 minutes)

The brief MSWO is a single round of the standard procedure used as a rapid screen for in-session reinforcer adjustment, group-level identification, or rapid digital-stimulus screening. The hierarchy is the rank order from that single round; expect that this ranking will diverge from a three-session MSWO in roughly 40% of cases Conine et al. (2021) Melanson et al. (2023). Use brief MSWO when:

• A teaching block has stalled and you suspect the current reinforcer has lost potency — a 1–3 minute brief MSWO drawn from items the learner has not had access to in the last 30 minutes will surface the moment's actual top item.
• You need a whole-class reinforcer for a group reward without pulling each student aside.
• You are screening digital stimuli (videos, apps, streaming content) where running the full MSWO would consume more session time than the resulting hierarchy is worth.

Do not use brief MSWO as the basis for selecting reinforcers in a BIP, a skill-acquisition program targeting a long-term goal, or any context where a wrong rank cascades into weeks of ineffective treatment Conine et al. (2021).

Pre-session vs in-session MSWO

A pre-session MSWO (run at the start of a teaching block, before any learning trials) is the format most graduate programs teach and the format most format-comparison research uses; it produces the cleanest hierarchy because MO state is more controlled and the learner is not yet in an instructional contingency Melanson et al. (2023) Poling et al. (2020). An in-session MSWO (typically a brief MSWO, run mid-block when responding drops) corrects for satiation effects and within-session preference shifts and is what the field's largest practice-pattern survey suggests experienced clinicians actually do (Morris et al., 2024). The optimal protocol uses both: a pre-session full MSWO to set the day's working hierarchy plus brief in-session checks every 30–40 minutes when responding meaningfully drops (Morris et al., 2024) Melanson et al. (2023).

Tangible vs edible array considerations

Edible arrays produce less stable within-session hierarchies than leisure items in pediatric samples, with rank shifts likely driven by accumulating satiation across rounds Melanson et al. (2023). Several practical implications follow. Keep edible array sizes small (4–5 items) and portion sizes minimal — large portions multiply the satiation problem Poling et al. (2020). Avoid mixing edibles and tangibles in the same array unless you specifically want to identify which class is more potent in the moment; the cross-class hierarchy is harder to interpret and may misrepresent within-class preferences. For learners on restricted diets or with food sensitivities, default to leisure-item MSWO and validate with a separate edible probe only when the BIP specifically calls for food reinforcers (Kronfli et al., 2024). Standardize stimulus size and portion across administrations: even subtle changes in presentation parameters shift mid- and lower-rank items, which matters whenever the full hierarchy — not just the top item — is operationally relevant.

• Treating round one as the hierarchy. Within-session preference correspondence is roughly 60%; rank shifts in 40% of round comparisons mean a single round is not the assessment Melanson et al. (2023). Run at least two — preferably three — rounds before adopting a final hierarchy.
• Assuming first-selected equals most-reinforcing. Some learners systematically save the highest-value item for last, and a follow-up reinforcer assessment may identify the last-selected item as the strongest reinforcer Castillo et al. (2022).
• Skipping the prerequisite check on picture and icon formats. Pictorial MSWO produces silently invalid hierarchies when learners cannot match the picture to its tangible referent. Probe object-to-picture and picture-to-object matching before running pictorial MSWO; if the probe fails, switch to tangible MSWO or train the prerequisite first Lill et al. (2021).
• Arrays that are too large for the learner's scanning capacity. Arrays larger than 7 items risk cognitive overload and selection by spatial bias rather than preference; arrays smaller than 4 under-discriminate. Stay in the 5–7 range unless population-specific evidence supports otherwise Ford et al. (2022).
• Insufficient sampling time per item. Brief access periods that do not allow consummatory contact with the item undermine the contingency that the procedure depends on. Use 30 seconds for tangibles, full consumption for edibles, and at least one full play-through for short videos.
• Ignoring "no choice" responses. A non-selection is data: it can signal satiation, poor MO state, or that the array contains nothing the learner currently values. Score it, do not over-prompt, and adjust the array or MO before re-running Melanson et al. (2023).
• Repeating MSWO without sanitization or stimulus refresh. Saliva contamination on edibles, contaminated toy surfaces, and degraded video files all introduce assessor-side confounds. Use a no-access video MSWO when sanitation is the constraint Brodhead et al. (2017).
• Testing under satiation. Running an edible MSWO right after lunch produces an unrepresentative hierarchy. Document MO state on the data sheet and run when the learner is at typical motivational baseline for the stimulus class Poling et al. (2020).
• Mixing stimulus classes without justification. Cross-class arrays (edibles + leisure) are harder to interpret and may misrepresent within-class preferences. Use single-class arrays unless the explicit clinical question is which class is currently more potent.
• Treating a completed staff-training module as evidence of mastery. Interactive computer training alone produced only small-to-moderate gains in MSWO accuracy across nine staff; most still required in-vivo feedback to reach mastery. Pair any computer-based training with at least one corrective-feedback round under direct observation (Sherman et al., 2025).
• Assuming MSWO rank predicts instructional reinforcer efficacy. MSWO rank does not always track learning rate during instruction. Validate the top one or two items with a brief in-task reinforcer probe before committing them to a long-term skill-acquisition program (Kronfli et al., 2024).
• Failing to standardize stimulus presentation. Stimulus size, portion, and modality shift the rank of mid- and low-preferred items. Hold presentation parameters constant across re-administrations whenever the full hierarchy — not just the top item — matters.

• Persistent inability to identify any preferred item across MSWO and alternative formats. If MSWO, brief MSWO, single stimulus, and free operant all fail to surface any reliably approached stimulus across multiple sessions and several MO states, escalate to a specialist team rather than committing to a treatment plan without an empirical reinforcer.
• Sudden hierarchy collapse in a previously stable learner. Abrupt rejection of items previously ranked high — particularly food refusal — can mask GI, dental, or sensory issues; document a medical referral before re-running or re-validating the MSWO.
• Cognitive impairment beyond the procedure's reach. When a learner cannot meet the scanning, selection, and inter-trial-interval demands of MSWO even after array reduction and procedural simplification, refer to a team experienced with single-stimulus and adapted preference protocols Ford et al. (2022).
• Staff training requirements beyond local capacity. When local staff cannot reach 90% MSWO procedural fidelity through brief BST, VMVO, or telehealth packages within 3–5 sessions, refer to a regional consultation team rather than running an underpowered assessment in-house O’Handley et al. (2021) (Sherman et al., 2025).

The MSWO literature has strong operational claims about format selection, brief variants, and staff training, but several gaps remain. Intra-session stability has been quantified at roughly 60% rank correspondence within-administration, but the field lacks consensus on which specific re-assessment cadence (daily, weekly, MO-triggered) optimally balances precision against staff time Melanson et al. (2023). The "save the best for last" pattern observed in some adults with IDD has not been characterized at scale — its prevalence, predictors, and clinical implications all remain open Castillo et al. (2022). Conine and colleagues' analysis of one-, two-, and three-session MSWO was retrospective; a prospective comparison of treatment outcomes when clinicians use one-session versus three-session MSWO data to select BIP reinforcers would directly answer the operational question of how much MSWO is enough Conine et al. (2021). The institutional-adoption gap — only 26% of surveyed clinicians report using formal preference assessment — is an implementation-science problem more than a procedural one and warrants systems-level intervention research rather than further procedural refinement (Morris et al., 2024). Finally, the recent finding that interactive computer training alone does not produce mastery-level MSWO accuracy raises the question of which specific training-package components actually drive procedural fidelity — exemplar-only VMVO works, ICT alone does not, and the difference is not yet well-characterized (Sherman et al., 2025) (Bartle et al., 2025).

1. Run at least three rounds before adopting a final hierarchy. Within-session rank shifts occur in roughly 40% of round comparisons; round one is not the assessment Melanson et al. (2023).
2. Validate the top one or two MSWO items with a downstream reinforcer test. Choice rank is correlated with but not identical to reinforcer efficacy; some learners save the best for last, and rank does not always predict instructional learning rate Castillo et al. (2022) (Kronfli et al., 2024).
3. Use brief MSWO for in-session adjustment, not for high-stakes reinforcer selection. Single-session and three-session MSWO correlate strongly but do not produce identical rankings; reserve the three-session version for BIP and skill-acquisition program design Conine et al. (2021).
4. Default to tablet or web-based MSWO for digital stimuli. Tablet app-icon MSWO produces rank-order correlations of 0.94 with tangible MSWO when picture-matching prerequisites are intact Hoffmann et al. (2019) Curiel et al. (2020).
5. Use brief no-access video MSWO when sanitation or contingent delivery is impractical. Rankings show moderate-to-strong correspondence with the standard format Brodhead et al. (2017).
6. Verify picture-matching prerequisites before any pictorial MSWO. Missing prerequisites produce silently invalid hierarchies Lill et al. (2021).
7. Switch to single stimulus for older adults with NCD and learners with limited choice repertoire. SS produces fewer implementation errors and runs faster than MSWO in these populations Ford et al. (2022).
8. Train staff with brief BST or exemplar-only VMVO; do not rely on ICT alone. Brief BST raises preservice school psychologists from 9–19% to 73–93% MSWO accuracy; exemplar-only VMVO maintains 90% procedural integrity at 5-week follow-up; ICT alone yields only small-to-moderate gains and requires supplemental in-vivo feedback for most staff O’Handley et al. (2021) (Bartle et al., 2025) (Sherman et al., 2025).
9. Hold stimulus presentation parameters constant across re-administrations. Size, portion, and modality shift mid- and lower-rank items; standardization protects the comparability of repeated MSWOs.
10. Build pre-session MSWO into the start of every long block and brief MSWO into mid-block engagement drops. This is what experienced clinicians actually do and the practice-pattern survey supports as the routine reinforcer-management workflow (Morris et al., 2024).
11. Document MO state, array configuration, and assessor identity on every MSWO data sheet. Re-interpretation of rank shifts is impossible without these contextual variables Poling et al. (2020).
12. Use the web-based MSWO tool to standardize implementation and eliminate paper data sheets. Auto-recorded selections and spreadsheet exports remove a meaningful source of clinical-team variability Curiel et al. (2020).

Primary research synthesized in this guide. DOIs link to the original source.

• Sherman, J., Vedora, J., Hotchkiss, R., & Colón-Kwedor, C. (2025). An analysis of interactive computer training on staff acquisition of MSWO preference assessment implementation. Journal of Organizational Behavior Management, 45(4), 307–323. https://doi.org/10.1080/01608061.2024.2438013 https://doi.org/10.1080/01608061.2024.2438013
• Kronfli, F. R., Vollmer, T. R., Hack, G. O., & DuBois, G. D. (2024). Optimizing learning outcomes when teaching sight words using fruits and vegetables as reinforcers. Behavior Analysis in Practice, 17, 908–914. https://doi.org/10.1007/s40617-024-00912-6 https://doi.org/10.1007/s40617-024-00912-6
• Bartle, G. E., Ruby, S. A., & DiGennaro Reed, F. D. (2025). The effects of video modeling containing different exemplar types on procedural integrity. Journal of Organizational Behavior Management. https://doi.org/10.1080/01608061.2025.2476425 https://doi.org/10.1080/01608061.2025.2476425
• O’Handley, R. D., Pearson, S., Taylor, C., & Congdon, M. (2021). Training Preservice School Psychologists to Conduct a Stimulus Preference Assessment. Behavior Analysis in Practice, 14(2), 445-450. https://doi.org/10.1007/s40617-020-00537-5 https://doi.org/10.1007/s40617-020-00537-5
• Castillo, M. I., Frank‐Crawford, M. A., Liesfeld, J. E., Doan, T. M., Newcomb, E. T., Rooker, G. W., & Borrero, J. C. (2022). Do persons with intellectual and developmental disabilities prefer to save the best for last in an MSWO? A preliminary investigation. Behavioral Interventions, 37(4), 1133-1148. https://doi.org/10.1002/bin.1883 https://doi.org/10.1002/bin.1883
• McCammon, M. N., Wolfe, K., & Check, A. R. (2024). A review of the environmental variables included in mand training interventions. The Analysis of Verbal Behavior, 40, 345–378. https://doi.org/10.1007/s40616-024-00211-9 https://doi.org/10.1007/s40616-024-00211-9
• Curiel, H., Curiel, E. S. L., Adame, A., & Li, A. (2020). Multiple‐stimulus‐without‐replacement preference assessment tool. Behavioral Interventions, 35(4), 680-690. https://doi.org/10.1002/bin.1732 https://doi.org/10.1002/bin.1732
• Brodhead, M. T., Abston, G. W., Mates, M., & Abel, E. A. (2017). Further refinement of video‐based brief multiple‐stimulus without replacement preference assessments. Journal of Applied Behavior Analysis, 50(1), 170-175. https://doi.org/10.1002/jaba.358 https://doi.org/10.1002/jaba.358
• Conine, D. E., Morris, S. L., Kronfli, F. R., Slanzi, C. M., Petronelli, A. K., Kalick, L., & Vollmer, T. R. (2021). Comparing the results of one‐session, two‐session, and three‐session MSWO preference assessments. Journal of Applied Behavior Analysis, 54(2), 700-712. https://doi.org/10.1002/jaba.808 https://doi.org/10.1002/jaba.808
• Arnal Wishnowski, L., Yu, C., Pear, J., Chand, C., & Saltel, L. (2018). Effects of computer‐aided instruction on the implementation of the MSWO stimulus preference assessment. Behavioral Interventions, 33(1), 56-68. https://doi.org/10.1002/bin.1508 https://doi.org/10.1002/bin.1508
• Melanson, I. J., Thomas, A. L., Brodhead, M. T., Sipila‐Thomas, E. S., Miranda, D. R. G., Plavnick, J. B., Joy, T. A., Fisher, M. H., & White‐Cascarilla, A. N. (2023). An evaluation of preference stability within MSWO preference assessments for children with autism. Journal of Applied Behavior Analysis, 56(3), 638-655. https://doi.org/10.1002/jaba.988 https://doi.org/10.1002/jaba.988
• Jarmolowicz, D. P., Lemley, S. M., Mateos, A., & Sofis, M. J. (2016). A multiple‐stimulus‐without‐replacement assessment for sexual partners: Purchase task validation. Journal of Applied Behavior Analysis, 49(3), 723-729. https://doi.org/10.1002/jaba.313 https://doi.org/10.1002/jaba.313
• Hoffmann, A. N., Brady, A. M., Paskins, R. T., & Sellers, T. P. (2019). Using Pictures Depicting App Icons to Conduct an MSWO Preference Assessment on a Tablet Device. Behavior Analysis in Practice, 12(2), 335-342. https://doi.org/10.1007/s40617-018-00309-2 https://doi.org/10.1007/s40617-018-00309-2
• Poling, A., Lotfizadeh, A. D., & Edwards, T. L. (2020). Motivating Operations and Discriminative Stimuli: Distinguishable but Interactive Variables. Behavior Analysis in Practice, 13(2), 502-508. https://doi.org/10.1007/s40617-019-00400-2 https://doi.org/10.1007/s40617-019-00400-2
• Ford, M. N., Bayles, M. W., & Bruzek, J. L. (2022). Assessing Preference and Stability of Preference for Individuals with Neurocognitive Disorder. Behavior Analysis in Practice, 15(3), 782-795. https://doi.org/10.1007/s40617-021-00648-7 https://doi.org/10.1007/s40617-021-00648-7
• Lill, J. D., Shriver, M. D., & Allen, K. D. (2021). Stimulus Preference Assessment Decision-Making System (SPADS): A Decision-Making Model for Practitioners. Behavior Analysis in Practice, 14(4), 1144-1156. https://doi.org/10.1007/s40617-020-00539-3 https://doi.org/10.1007/s40617-020-00539-3
• Morris, S. L., Conine, D. E., Slanzi, C. M., Kronfli, F. R., & Etchison, H. M. (2024). A survey of why and how clinicians change reinforcers during teaching sessions. Behavior Analysis in Practice, 17, 815–830. https://doi.org/10.1007/s40617-023-00847-4 https://doi.org/10.1007/s40617-023-00847-4