Partial interval recording is a behavior observation method where you divide an observation period into equal time segments and simply note whether a behavior happened at any point during each segment. If the behavior occurs even once during an interval, you mark it as an occurrence. If it never appears during that interval, you mark it as a non-occurrence. It’s one of the most common tools used in applied behavior analysis and special education classrooms to track how often problem behaviors or target skills show up over time.
How It Works in Practice
The setup is straightforward. You choose a total observation period, say 10 or 20 minutes, and break it into equal intervals. These intervals are typically short, often 10, 15, or 30 seconds, though the exact length depends on the behavior you’re tracking. You then watch the student or client during each interval and record a simple yes or no: did the behavior happen at any point during those seconds?
The key rule is that it doesn’t matter whether the behavior happened once or five times within a single interval. Both get the same mark. A student could tap their desk once in a 15-second window or tap it repeatedly for the full 15 seconds, and the recording would look identical. This makes the method quick and easy to use but also limits the detail it captures.
After the observation, you calculate the percentage of intervals in which the behavior occurred. If a behavior was marked in 12 out of 20 intervals, you’d report it as occurring in 60% of intervals. This gives a general picture of how spread out the behavior is across a session.
What It’s Best Suited For
Partial interval recording works best for behaviors that are brief and don’t last the entire interval. Think of quick, discrete actions: a student calling out in class, a child flapping their hands for a few seconds, or brief instances of aggression. The method is designed to catch behaviors that pop in and out rather than ones that persist continuously. It’s particularly useful when you need to estimate how frequently a behavior occurs rather than how long it lasts.
Research supports this distinction. A study in the Journal of Applied Behavior Analysis concluded that partial interval recording is the recommended discontinuous measurement method for estimating the frequency of responses, while momentary time sampling is better suited for behaviors measured by duration. If you’re trying to answer “how often does this happen?” rather than “how long does this last?”, partial interval recording is the stronger fit.
The Overestimation Problem
The most important thing to understand about partial interval recording is that it consistently overestimates how much a behavior actually occurs. This isn’t a minor technical footnote. It’s a well-documented, systematic bias that can significantly distort your data.
Here’s why it happens: if a behavior occurs for just 2 seconds within a 30-second interval, the entire interval gets marked as an occurrence. Across a full session, this can make a behavior look far more prevalent than it really is. A computer simulation published in the Journal of Applied Behavior Analysis found that in some cases, the relative error was enormous, reaching up to 10,000% of the actual observation period when behavior durations were short and interval durations were long. Even under more typical conditions, the bias consistently pushed estimates upward.
The overestimation tends to be worst when the actual behavior is brief and infrequent, which is ironic since those are exactly the behaviors the method is most often used to track. Shorter intervals reduce the error somewhat, but reviews of the research literature have found that partial interval recording overestimates duration even when relatively small intervals are used.
This bias has real consequences for treatment decisions. Overestimated baseline levels can mask the effects of an intervention, making it look like a treatment isn’t working when it actually is (a false negative). Conversely, the method can also produce false positives, detecting intervention effects that haven’t actually occurred. Both outcomes can lead practitioners down the wrong path.
How It Differs From Whole Interval Recording
Whole interval recording flips the rule. Instead of marking an interval as an occurrence if the behavior happens at any point, whole interval recording requires the behavior to be present throughout the entire interval. If a student is on task for 28 out of 30 seconds, whole interval recording would mark that as a non-occurrence.
This means the two methods produce opposite biases. Partial interval recording overestimates behavior because any brief appearance counts. Whole interval recording underestimates behavior because anything short of continuous engagement gets missed. Whole interval recording is better suited for sustained behaviors like staying in a seat, maintaining eye contact, or engaging with a task, where you want to know if the behavior persisted rather than just whether it appeared.
How It Compares to Momentary Time Sampling
Momentary time sampling takes a different approach entirely. Instead of watching the entire interval, you only look at the exact moment the interval ends. If the behavior is happening right at that instant, you record it. If not, you don’t, regardless of what happened during the rest of the interval.
Multiple studies have found momentary time sampling to be more accurate than partial interval recording for estimating how long a behavior lasts. One foundational study compared both methods to continuous duration recording of a secretary’s in-seat behavior during 20-minute sessions and found that partial interval recording consistently overestimated duration, while momentary time sampling sometimes overestimated and sometimes underestimated, with a much smaller margin of error overall. Later research with children exhibiting repetitive behaviors confirmed the same pattern: momentary time sampling produced more accurate estimates across low, moderate, and high levels of the behavior.
The tradeoff is that momentary time sampling can miss brief behaviors entirely since you’re only checking at a single moment. For behaviors that are very short, like a quick vocalization or a single hit, partial interval recording is more likely to catch them. The choice between the two depends on whether you care more about capturing every instance (favoring partial interval recording) or getting a more accurate picture of overall levels (favoring momentary time sampling).
Choosing the Right Interval Length
The length of your intervals directly affects data quality. Shorter intervals produce more accurate results because there’s less room for the overestimation bias to inflate your numbers. If a behavior lasts 3 seconds and your interval is 10 seconds, the distortion is smaller than if your interval is 60 seconds.
Research confirms that increasing interval size decreases accuracy. The relationship between interval length and the actual duration of each behavior occurrence is what drives error levels. When intervals are much longer than the behavior itself, error grows substantially. A general guideline is to keep intervals short enough that the behavior you’re tracking could plausibly fill a meaningful portion of the interval. For most classroom or clinical applications, intervals of 10 to 15 seconds offer a reasonable balance between accuracy and practicality.
Longer intervals are easier to manage, especially when one person is simultaneously teaching and collecting data. But the convenience comes at the cost of precision. If your data will be used to make treatment decisions or evaluate whether an intervention is working, shorter intervals are worth the extra effort.
Practical Strengths and Limitations
The biggest advantage of partial interval recording is simplicity. It doesn’t require the observer to count every instance of a behavior or track exactly how long each episode lasts. You just need to notice whether it happened at all during each interval. This makes it manageable in real-world settings where a teacher or therapist is juggling multiple responsibilities.
It also works well as a screening tool. If you’re trying to get a rough sense of whether a behavior is happening frequently enough to warrant a more detailed assessment, partial interval recording gives you a fast, low-burden starting point. And because the recording format is standardized, it’s relatively easy to train new observers and achieve consistency across data collectors.
The limitations are equally clear. It cannot tell you how many times a behavior occurred within an interval, how long each instance lasted, or whether the behavior is getting shorter or longer over time. It flattens all of that detail into a binary yes or no. For tracking subtle changes during an intervention, especially when you need to detect small improvements, this lack of sensitivity can be a serious problem. The overestimation bias can push scores toward a ceiling of 100%, leaving little room to detect meaningful progress even when the behavior is genuinely decreasing.