How Do I Run a Preference Assessment? (With Examples)

A preference assessment is a short, structured way to find out what a learner actually wants to work for, right now. It sounds basic, and it is — but it is also one of the most skipped steps in applied practice. Sessions stall, demands get refused, and the "reinforcer" sitting on the table turns out to be something the learner stopped caring about weeks ago.

This guide walks you through the five standard methods — caregiver interview, free-operant observation, single-stimulus, paired-choice, and multiple-stimulus arrays — when to use each one, and how to record and use the results. There is a full worked example in section 5 so you can see exactly what the data look like.

Everything here is doable with items you already have in your session bag or classroom. No special materials, no software — just a tally sheet and a few minutes at the start of your session.

1. Why Preference Assessments Matter (and Why Once Is Never Enough)

Reinforcers change. A preference assessment replaces guessing with data, so you stop wasting session time on items that no longer motivate.

Reinforcement only works if the "reward" is actually rewarding to the learner in that moment. What a learner loved last month — or this morning — may do nothing for them right now. When you guess wrong, you get slower acquisition, more refusals, and more problem behavior, and it is easy to misread all of that as the learner "having a bad day" when the real issue is that nothing on the table was worth working for.

A preference assessment gives you a ranked list of what the learner wants most right now. That ranking — often called a preference hierarchy — lets you match your strongest items to your hardest demands and save weaker items for easy tasks. It takes a few minutes and pays for itself many times over across a session.

2. Ask First — but Verify

Caregiver and teacher interviews are a great starting point for what to test, but direct observation beats report every time.

Start by asking the people who know the learner: "What does she play with at home? What foods does he ask for? What does she do when she has free time?" This gives you a candidate list of items and activities to bring into the assessment.

Then verify. Reports are honest but often out of date or filtered ("he likes puzzles" may mean "we bought him puzzles"). The learner's own choices, observed directly, are the real data. Use the interview to decide what to test, and use one of the direct methods below to decide what to use.

3. Free-Operant Observation: Watch What They Choose

Give free access to a set of items and record where the learner spends their time. Least intrusive, no demands, no removal.

Set out several items in a room or on a table, let the learner move freely, and time how long they engage with each item over a set period (for example, 5 minutes). The items that hold attention longest are your high-preference candidates.

Free-operant observation is a good first choice for learners who get upset when items are taken away, because nothing is ever removed — you only watch and record. The trade-off is that it may not give you a clean ranking: a learner may camp on one item the whole time and tell you nothing about the rest.

4. Single-Stimulus: One Item at a Time

Present one item, record whether the learner approaches it. Use this for learners who cannot yet choose between two items.

Present each item by itself and record the learner's response: did they reach for it, engage with it, eat it, push it away, ignore it? Present every item several times in mixed order, then compare approach rates across items.

This is the right method for learners who do not yet scan an array or make a selection between options. It is slower and gives a rougher ranking than the choice-based methods — many learners approach almost everything when it is the only thing offered — but it gets you real data when nothing else will.

5. Paired-Choice (Forced-Choice): The Workhorse — With a Worked Example

Present items two at a time, record the pick each trial, and you get a full ranked hierarchy. This is the most commonly used method.

Every item is paired with every other item, one pair per trial. The learner picks one; the other is removed. Because the learner must choose between items, you get a much sharper ranking than approach data alone.

Worked example — 5 toys. Say your candidate items are bubbles, a spinner, blocks, a toy car, and a picture book. With 5 items, every possible pair gives you 10 trials. Each trial: put two items on the table about a foot apart, say "pick one," let the learner engage with the pick for about 30 seconds, remove both, and tally the pick. Swap which side each item appears on across trials so a side preference does not contaminate the data. Your tally sheet at the end might look like this (each item appears in 4 pairs):

To compute the hierarchy, divide times chosen by times presented for each item and rank from highest to lowest. In this example, bubbles are today's strongest reinforcer, the spinner is a solid backup, and the picture book should not be your go-to for hard work today. (This is a generic illustration of the procedure, not data from a real learner.)

6. MSW and MSWO: Fast Hierarchies From an Array

Present the whole array at once. Multiple-stimulus-without-replacement (MSWO) gives you a full ranking in just a few trials.

Instead of pairs, lay out all the items in a row (usually 5 to 8) and ask the learner to pick one. In multiple stimulus with replacement (MSW), the chosen item goes back into the array each trial. In multiple stimulus without replacement (MSWO), the chosen item is set aside, the remaining items are re-ordered, and the learner picks again from what is left — the order in which items leave the array is your hierarchy.

MSWO is faster than paired-choice: 5 items rank in as few as 5 picks instead of 10 trials. It works well for learners who can scan a row of items and make a clear selection. If your learner grabs the nearest item regardless of what it is, drop back to paired-choice or single-stimulus.

7. Run a Brief Mini-Assessment Before Every Session

Top items shift daily. A 2-minute MSWO or a few quick paired trials at the start of each session keeps your reinforcers current.

You do not need the full procedure every day. A brief version — one quick MSWO round with your learner's top 4 or 5 known items, or 3 to 4 paired trials — is enough to confirm what is strong today.

Build it into your session-opening routine, right after rapport-building: pull out the tally sheet, run the mini-assessment, and set up your reinforcers before the first demand. Two minutes here saves you twenty minutes of fighting a flat session later.

8. Using the Results — and the Mistakes That Undo Them

Reserve top-ranked items for the hardest demands, rotate reinforcers to prevent satiation, and never treat one assessment as permanent truth.

Match reinforcer strength to task difficulty: the #1 item is for the hardest, newest, or most-refused demands; mid-ranked items cover maintenance tasks and easy trials. If the top item is available for everything, it stops being special — and if the learner just had 20 minutes of free access to it, it will not function as a reinforcer at all.

Rotate. Even a strong reinforcer weakens with repeated use inside a session (satiation). Cycle among your top 2 or 3 items, and retire an item for a few days when its pull visibly drops.

The common mistakes to avoid:

• Assessing once a year (or once at intake) and running on stale data forever. Preferences drift in days, not semesters.
• Items too similar — five nearly identical fidget toys tell you nothing. Sample across categories: edible, sensory, toy, activity, social.
• Letting the assessment become a demand. If you drill "pick one!" with a flat tone and no access time, the assessment itself turns aversive and the data go bad. Keep it brisk, low-pressure, and always give real access to the chosen item.
• Ignoring side bias. If the learner always picks the left item, counterbalance positions and re-run before trusting the ranking.

When to Get More Help

Bring your supervising BCBA in when the data will not settle: every item ranks about the same, the learner refuses to choose at all, the assessment reliably triggers problem behavior, or the "top" reinforcer still will not hold behavior during demands. Those patterns usually mean the method needs to change (for example, moving from an array to single-stimulus, or testing activities and social interaction instead of tangibles) — and method selection is a supervision conversation, not a solo call. If you are a parent working through this at home and feel stuck, our free Start Here page can point you to the right next step.