TL;DR: Choosing the wrong data collection method can produce numbers that look clean but mislead clinical decisions. Here’s how to pick the right system for the behavior you’re actually measuring.
Data is the heartbeat of ABA practice. Without it, behavior analysts are guessing. With the wrong system, they’re measuring a different question than they think they’re asking.
Choosing a data collection method is not a paperwork decision — it’s a clinical decision. The method you choose shapes what you learn about a behavior, and that shapes whether your interventions actually work.
The Core Data Collection Methods
Frequency (Event) Recording
Track how many times a behavior occurs within a defined observation period. Best for discrete, clearly observable behaviors with a definite beginning and end — hitting, vocal requests (mands), or instances of elopement.
Use when: The behavior has a clear start and stop, and you care about how often it happens. Pair frequency with a consistent observation window if you need rate (responses per minute or hour).
Avoid when: The behavior is long-duration or continuous — like stereotypy or on-task engagement — where counting individual instances isn’t clinically meaningful.
Duration Recording
Measure how long a behavior lasts each time it occurs. Best for behaviors where length matters more than count — tantrums, on-task engagement, time in a designated area.
Use when: You need to know how long something is happening, not just how often. A child may have three tantrums a day (frequency) but one that lasts 40 minutes. Duration tells the clinical story frequency misses.
Latency Recording
Measure the time between a cue (like an instruction) and the behavior’s onset. Useful for assessing compliance, transition speed, or skill fluency.
Use when: You care how quickly someone responds to a stimulus — relevant for instruction-following, task initiation, and manding.
Interval Recording
Divide your observation window into equal time blocks and mark whether the behavior occurred during each. Three common variants:
- Whole Interval: Mark “yes” only if the behavior occurred for the entire interval. Tends to underestimate — best for measuring sustained behaviors.
- Partial Interval: Mark “yes” if the behavior occurred at any point during the interval. Tends to overestimate — useful when behaviors occur too rapidly to count discretely.
- Momentary Time Sampling (MTS): Check whether the behavior is occurring only at the end of the interval. Less intrusive; good for naturalistic or classroom settings where you can’t watch continuously.
Trial-by-Trial Recording
Standard in Discrete Trial Training. Mark correct, incorrect, or prompted for each trial. Gives you precise response accuracy and prompt-dependency data to guide fading and mastery decisions.
Choosing the Right System
The most common mistake is defaulting to partial interval for everything — it’s easy to collect but systematically overestimates behavior. Before setting up a data system, ask: what clinical question am I actually trying to answer?
| Behavior type | Best system |
|---|---|
| Discrete, short-duration behaviors | Frequency |
| Long-duration or sustained behaviors | Duration or Whole Interval |
| Response speed / compliance | Latency |
| High-rate or continuous behaviors | Partial Interval or MTS |
| Structured teaching programs | Trial-by-trial |
Train the System Before You Go Live
Whichever method you choose, your whole team needs to collect the same way. Inter-observer agreement (IOA) isn’t just a BACB requirement — it’s how you catch drift before it corrupts your dataset. Build IOA checks in from day one, and coach your RBTs on the system before they run sessions.
Here’s where practice matters as much as knowledge. Knowing what interval recording is and being able to run it accurately in a live session are two different skills. Practitioners who haven’t rehearsed a new data system before using it tend to miss trials, default to familiar shortcuts, or collect inconsistently under session pressure.
How Kipr Helps
Kipr’s simulation environment lets BCBAs and RBTs practice data collection in realistic session scenarios before applying those skills with actual clients. You can run a simulated DTT session, practice interval recording on a simulated learner, and build consistency — without any real learner bearing the cost of your learning curve.
If you’re onboarding a new RBT cohort or trying to build data fidelity across your clinical team, simulation-based practice can close the gap between understanding a method and executing it reliably under the pressure of a live session.
Join the Kipr waitlist for early access.