Interval recording is a method of observing and documenting behavior by dividing an observation session into equal time segments, then noting whether a target behavior occurred during each segment. Instead of counting every single instance of a behavior, you simply mark “yes” or “no” for each interval. This makes it a practical tool in classrooms, clinics, and research settings where tracking every occurrence of a behavior would be difficult or impossible.
Why Interval Recording Exists
Some behaviors are easy to count. A child raises their hand, you make a tally mark. But many behaviors don’t have a clear start and stop point, or they happen so frequently that counting each one becomes impractical. Think of a student who rocks back and forth in their chair, a child who hums throughout a lesson, or someone who engages in repetitive self-stimulatory movements. These behaviors blur together in a way that makes simple frequency counts unreliable.
Continuous recording, where a dedicated observer tracks every instance of a behavior with precise timing, gives the most accurate data. But in real clinical and educational settings, therapists and teachers are responsible for monitoring multiple individuals or multiple behaviors at once. Interval recording solves this by giving observers a structured, manageable way to estimate how much a behavior is happening without needing to catch every moment.
How It Works in Practice
The setup is straightforward. You decide on a total observation period (say, 5 minutes), then break it into equal intervals, typically between 5 and 15 seconds long. A 5-minute session with 10-second intervals gives you 30 intervals to score. During the observation, a timer or audible beep signals the end of each interval, and the observer marks whether the behavior occurred.
The result is a data sheet with a series of “occurrence” or “non-occurrence” marks. To calculate a percentage, you divide the number of intervals where the behavior was recorded by the total number of intervals and multiply by 100. If a student was off-task in 12 out of 30 intervals, you’d report that the behavior occurred in 40% of intervals. This percentage becomes your estimate of how prevalent the behavior is, and you can track it over days or weeks to see whether an intervention is working.
Whole Interval Recording
In whole interval recording, you only mark an interval as “occurrence” if the behavior lasted for the entire duration of that interval. If a child was out of their seat for 8 seconds of a 10-second interval, you would mark it as non-occurrence because it didn’t persist through the full 10 seconds.
This method has a built-in bias: it consistently underestimates how much a behavior is actually happening. A behavior could be present for most of an interval and still not get counted. Computer simulations published in the Journal of Applied Behavior Analysis found that whole interval recording underestimates cumulative behavior duration, and the underestimation gets worse as interval lengths increase. In some scenarios, the error reached as high as 100%, meaning the behavior was completely missed.
Because of this tendency to undercount, whole interval recording is generally considered a conservative measure. It works best when you want to track behaviors you’re trying to increase, like on-task engagement, since it sets a high bar for what counts as “present.”
Partial Interval Recording
Partial interval recording flips the rule. You mark an interval as “occurrence” if the behavior happens at any point during that interval, even briefly. If a child calls out for just 1 second within a 10-second window, the entire interval gets scored as an occurrence.
This method overestimates behavior. A behavior that occurs for a fraction of an interval gets the same score as one that fills the entire interval. Simulation research found that the overestimation can be dramatic, with relative error reaching up to 10,000% of actual behavior duration in extreme cases where behaviors were brief and intervals were long. Even under more typical conditions, partial interval recording inflates estimates in a predictable direction.
This overestimation makes partial interval recording a better fit for behaviors you’re trying to decrease, like aggression or self-injury. Since it catches even brief instances, it provides a sensitive measure that won’t let short bursts of a problem behavior slip through the data unnoticed.
Momentary Time Sampling
Momentary time sampling is a related method that’s sometimes grouped with interval recording, though it works differently. Instead of watching throughout each interval, the observer only looks at one specific moment, usually the last 1 to 2 seconds of each interval. If the behavior is happening at that exact moment, it’s scored as an occurrence. If not, it’s scored as non-occurrence, regardless of what happened during the rest of the interval.
This is the least demanding method for observers because it doesn’t require sustained attention throughout each interval. You can glance up at the right moment, record what you see, and return to other tasks. In a study published in the Journal of Applied Behavior Analysis, researchers used 2-second observation windows at the end of each 10-second interval. The tradeoff is that momentary time sampling can miss behaviors that are brief or infrequent since there’s a good chance they won’t be happening at the exact moment you check. Research has shown it can actually overestimate or underestimate behavior depending on the situation, making it less predictable in its bias than the other two methods.
Choosing the Right Method
The three types of interval recording each carry known biases, and picking the right one depends on what you’re measuring and why.
- Whole interval recording works well for behaviors you want to see more of, like academic engagement or appropriate play. Its tendency to underestimate means your data shows conservative progress, so improvements you document are likely real.
- Partial interval recording is better suited for behaviors you want to reduce, like disruptive vocalizations or stereotyped movements. Its tendency to overestimate ensures you don’t prematurely conclude that a problem behavior has disappeared.
- Momentary time sampling is most practical when you’re monitoring multiple people or behaviors simultaneously and can’t dedicate sustained attention to one target.
Interval Length Matters
Shorter intervals produce more accurate data but require more effort from the observer. Intervals between 5 and 15 seconds are standard for most behavioral observations. Longer intervals, like 30 seconds or more, reduce observer burden but amplify measurement error in both directions. Simulation research confirmed that as interval duration increases, both overestimation (in partial interval recording) and underestimation (in whole interval recording) get worse.
The interval length should also roughly match the nature of the behavior. Very brief behaviors, like a single tic or a quick hand flap, are poorly captured by long intervals in whole interval recording because they’ll almost never fill an entire 15- or 20-second window. For those behaviors, shorter intervals or partial interval recording gives a more useful picture. Behaviors that tend to persist, like sustained off-task posture, are better suited to longer intervals and whole interval recording.
Whatever interval length you choose, consistency matters more than perfection. Using the same interval duration across observation sessions ensures your data is comparable over time, which is the entire point of collecting it.