Interval Recording Explained: Partial, Whole, Momentary

Interval recording splits an observation into equal chunks of time and marks whether a behavior happened in each chunk. The result is a percentage of intervals, such as "on task in 60% of intervals." There are three versions. Partial interval marks a chunk if the behavior happened at any point. Whole interval marks it only if the behavior lasted the entire chunk. Momentary time sampling marks it only if the behavior is happening at the exact moment the chunk ends. The three give different numbers from the same session, so the version you choose, and the interval length, need to be written into the goal and onto the sheet.

When interval recording is the right tool

Interval recording fits behaviors that are too continuous to count and too frequent or fuzzy-edged to time. Off-task behavior, engagement during group work, humming, and fidgeting are common examples. If a behavior has a clear start and stop and happens a few times a period, a frequency count is simpler and more precise. If what matters is how long each episode lasts, use duration recording. Choosing a measurement type walks through that decision.

Interval data is an estimate. It trades exact duration or counts for a number a person can collect without a stopwatch running all period. Each method has a known direction of error, and knowing it is what lets you read the number honestly.

The three methods, and how each one errs

Partial interval gives full credit for a single second. A student who is off task for five seconds of a one-minute interval gets the same mark as one who is off task for all sixty. That inflates the percentage. Many teams use it for behaviors they want to reduce, since the inflation works against showing improvement too early, and for brief behaviors that a momentary check would miss.

Whole interval requires the behavior for the whole interval. A student who is on task for 55 of 60 seconds gets no mark. That deflates the percentage. It is commonly used for behaviors the team wants to increase, and it is demanding: the observer has to watch continuously.

Momentary time sampling only asks about one instant per interval. It is the least demanding for the observer, because you can teach and glance up when a vibrating timer goes off. With short intervals over a long enough session it tends to track the real proportion of time well. With long intervals it can miss brief behaviors entirely.

One session, scored three ways

The fastest way to see the difference is to score the same observation with all three methods.

None of these numbers is wrong. Each is an accurate answer to a slightly different question. The problem only comes when a team compares a partial-interval baseline with a momentary-sampling progress check, or reads a 70% partial-interval figure as "off task 70% of the time."

Choosing an interval length

Shorter intervals make every method more accurate, because each mark covers less time. They also make recording harder. The right length is the shortest one you can keep up with for the whole observation, every time.

• 10 to 30 seconds is common for a dedicated observer with nothing else to do, such as a behavior specialist during an FBA observation.
• 1 minute works for partial or whole interval when a paraeducator can watch one student for a short block.
• 1 to 2 minutes is common for momentary time sampling by a teacher who is also teaching, using a vibrating timer.
• Longer than 5 minutes turns momentary sampling into a handful of snapshots. It can miss the behavior entirely and is hard to interpret.

Keep the observation length the same across sessions too. Twenty one-minute intervals on Monday and eight on Thursday are hard to compare, because the short session has more room for chance.

Calculating the percentage

The formula is the same for all three methods:

Percent of intervals = intervals marked ÷ intervals observed × 100

The word that matters is observed. If a fire drill, a phone call, or a student leaving for speech means you could not watch an interval, cross it out and leave it out of the denominator. Scoring it as "did not occur" makes the behavior look rarer than it is.

Writing it into the goal

A goal that says "on task in 80% of intervals" is incomplete, because the same student can hit 80% with partial interval and 40% with whole interval. A complete version names the method and interval length:

"During 20-minute independent work periods, Aaliyah will be on task in at least 80% of 1-minute intervals, measured by momentary time sampling, across 3 consecutive observations."

The free interval recording sheet has room for 30 intervals per page, and its instructions ask you to write the method and interval length at the top before you begin. When the data is in, graphing IEP progress data covers plotting percent-of-interval points against an aim line.

Reporting interval data to a family

"On task in 62% of intervals" means little to most families, and it can sound more exact than it is. Translate the method along with the number. For momentary time sampling, a plain version is: "Every minute during reading, we checked whether Aaliyah was working. She was working at about 6 of every 10 checks, up from 4 of every 10 in September." For partial interval on a behavior to reduce: "We split math into one-minute chunks. In October, calling out happened in about half of them. Now it is about one in four."

Keep the comparison inside one method. If the team switched from partial to momentary sampling partway through the year, say so, and do not compare the two numbers directly. Mark the switch on the graph the same way you would mark a change in instruction.

Practical tips

• Define the behavior before choosing the method. "Off task" needs a definition two observers would agree on, such as the three-second rule above.
• Use a vibrating timer or an interval app. Watching a wall clock during a lesson produces drift within minutes.
• Observe at the same time of day. An observation during a preferred activity and one during a hard task are not comparable.
• Check agreement once. Have a second adult score the same session. If you are more than about 10 points apart, tighten the definition before collecting more data.