TL;DR: A preference assessment is the foundation of any reinforcement-based program. Pick the wrong format and your “reinforcers” stop reinforcing. This post walks through the five formats most BCBAs and RBTs encounter — single-stimulus, paired-stimulus, MSWO, MSW, and free-operant — and when each one earns its place in your session.
If you’ve ever watched a learner refuse the goldfish crackers that “worked yesterday,” you already know the practical truth behind preference assessments: preferences shift, satiation is real, and reinforcer assumptions decay fast. The job isn’t to run a preference assessment once and file it away. The job is to know which assessment to use, how often to repeat it, and how to read the data without overinterpreting it.
This post is written for BCBAs designing reinforcement systems and RBTs running daily sessions. We’ll keep the theory tight and the application concrete.
Why preference assessments matter (briefly)
A reinforcer is defined by its function, not by what an adult thinks is “fun.” A preference assessment is the cleanest, fastest way to predict what items or activities are likely to function as reinforcers — before you commit to a token economy, a teaching trial structure, or a behavior reduction plan that depends on access contingencies.
Skipping the assessment, or relying on caregiver report alone, is one of the most common reasons a program stalls. The contingency is fine. The “reinforcer” just isn’t reinforcing.
The five formats and when to use each
1. Single-Stimulus (SS)
How it works: Present one item at a time. Record whether the learner approaches, engages with it, and for how long.
When to use it: Early learners, learners who don’t yet discriminate between two items, or learners with limited motor or scanning skills. SS is the lowest-demand format.
Trade-off: Tells you whether something is liked, not whether it is preferred over anything else. You’ll get a “high-engagement” item that may still lose to a stronger reinforcer in a head-to-head.
2. Paired-Stimulus (PS) — “Forced Choice”
How it works: Present two items at a time, in every possible pairing. Record which item the learner selects.
When to use it: Learners who can scan and choose between two stimuli. Great for ranking a small set (5–8 items) when you need a clean preference hierarchy.
Trade-off: Time-intensive. With 8 items you’ll run 28 trials. Worth it for a strong baseline; less practical mid-session.
3. Multiple-Stimulus Without Replacement (MSWO)
How it works: Lay out an array of items (typically 5–7). The learner picks one. Remove the chosen item. Re-array the rest. Repeat until items are gone or the learner stops engaging.
When to use it: This is the workhorse. Faster than paired-stimulus, gives you a ranked hierarchy in one pass, and is easy to teach to RBTs. Use it at the start of a programming cycle and as a periodic re-check.
Trade-off: Requires a learner who can scan an array and tolerate items being removed. Watch for position bias — some learners always grab the leftmost item regardless of preference.
4. Multiple-Stimulus With Replacement (MSW)
How it works: Same array setup as MSWO, but the chosen item gets replaced into the array each time, and a new item replaces the unselected ones across trials.
When to use it: When you suspect one strong item is dominating and you want to test whether other items have any value at all. MSW exposes a single highly-preferred item but is poor at producing a hierarchy.
Trade-off: You learn that one thing is preferred. You don’t learn the order of the rest.
5. Free-Operant (FO)
How it works: Let the learner roam a contrived environment with multiple items freely available. Record what they engage with and for how long during a fixed window (often 5–10 minutes).
When to use it: Naturalistic settings, NET-heavy programming, learners who refuse structured tasks, or any time you want to know what a learner actually gravitates toward when no demands are placed.
Trade-off: Lower control. A novel item may dominate simply because it’s novel. Run it more than once to filter signal from noise.
Picking the right format quickly
| Situation | Best fit |
|---|---|
| Brand-new learner, unclear if anything functions as a reinforcer | Single-Stimulus |
| Need a ranked top-5 reinforcer list | MSWO |
| Suspect one item is washing out the rest | MSW |
| Learner won’t tolerate structured trials | Free-Operant |
| Setting up a token economy or full reinforcement schedule | Paired-Stimulus or MSWO |
Common mistakes practitioners make
- Running it once. Preferences shift with satiation, mood, time of day, and program demands. Quick MSWOs at the start of each session beat one comprehensive assessment from three months ago.
- Confusing engagement with preference. A learner may pick up a fidget because it’s there, not because it’s preferred over the iPad sitting in the cabinet. Make sure your array has real competition.
- Letting RBTs improvise the procedure. MSWO has a defined re-array step. Position bias is a defined risk. If the procedure drifts, the data drifts.
- Treating top-rank as durable. Today’s #1 reinforcer is tonight’s saturated stimulus. Build rotation into the plan.
How often to re-assess
A reasonable default for active programs:
- Brief MSWO (3–5 items): at the start of each session for high-density teaching
- Full MSWO or PS: every 2–4 weeks, or whenever a reinforcer “stops working”
- Free-operant probe: monthly, especially for learners with limited verbal repertoire
How Kipr Helps
Preference assessments are exactly the kind of procedure that looks simple in a textbook and falls apart on the floor — position bias, learners who refuse arrays, supervisors who can’t watch every session live. Kipr gives BCBAs and RBTs a way to practice running each assessment format with simulated learners before applying them to real clients. You can rehearse an MSWO, get the procedural drift flagged, and try again — without burning a real session or a real reinforcer hierarchy on a learning rep.
The goal is competence and confidence before the data point counts.
Join the Kipr waitlist for early access.