How interval recording works

Interval recording divides an observation session into equal time blocks and scores each block based on whether a target behaviour occurred. Instead of counting every instance, you end up with a percentage of intervals — a proportion score. For high-rate, continuous, or overlapping behaviours where precise frequency or rate counts are impractical, interval recording gives you a repeatable, comparable measure that staff can actually take in real settings.

Before starting, write a clear operational definition for each behaviour you plan to score. An ambiguous definition corrupts your data — two observers watching the same student will score differently if they disagree on what counts.

Momentary time sampling

At each interval boundary — the beep — you look up and record whether the behaviour is occurring at that exact instant, not what happened during the preceding interval. Because the scoring decision is anchored to a single moment rather than continuous watching, momentary time sampling produces the least biased estimate of how much of a session the behaviour actually occupies. It is the most practical method in classroom settings: teach right up to the beep, glance at the student, mark your score, and return immediately.

Partial-interval recording

Score the interval as + if the behaviour occurs at any point during it — even once, even briefly. The practical upside is sensitivity: short or low-rate behaviours that might fall between momentary snapshots will still be captured. The cost is accuracy. Because any occurrence at all scores the whole interval, partial-interval recording systematically over-estimates how much of the session the behaviour fills. A behaviour that flickered for two seconds in a 30-second interval scores the same as one that lasted the full 30 seconds. Use it when missing an occurrence would be clinically meaningful — for example, tracking brief instances of self-injurious behaviour you cannot afford to miss.

Whole-interval recording

Score the interval as + only if the behaviour continues for the entire interval without stopping. This is the mirror image of partial: it under-estimates true occurrence, because a behaviour that lasted 29 of 30 seconds scores zero. That under-estimation is sometimes useful — for behaviours where sustained engagement is the goal (on-task behaviour, remaining in seat, sustained communication) the whole-interval criterion holds a higher standard and makes small improvements visible as gains in percentage.

Which method should you use?

Start with the accuracy trade-offs: partial over-estimates, whole under-estimates, momentary provides the least biased estimate for ongoing behaviour. If you want the truest picture of how long a behaviour occupies a session, momentary time sampling is usually the right call. If you need high sensitivity to catch brief events, partial interval is more appropriate. If you are measuring a sustained behaviour and want the data to reflect genuine maintenance, whole interval raises the bar appropriately.

Whatever method you choose, pair the data with solid measurement hygiene. After the session, use the IOA Calculator to check interobserver agreement — interval-by-interval agreement tells you whether two independent observers are scoring the same events the same way. If agreement is low, revisit your operational definition before drawing conclusions from the data. If you are also recording what triggered the behaviour, the ABC Data Sheet Generator gives you a structured antecedent-behaviour-consequence format to run alongside interval sessions.