Interval Sampling and Permanent-Product Recording

Sometimes a behavior plan does not ask you to record every instance of a behavior. It may instead ask for one of two other kinds of record:

• A yes-or-no mark for each short block of time.
• A measurement of something the behavior left behind.

This lesson shows how to follow each kind of plan, and what the resulting numbers can and cannot tell you.

What you will be able to do

• Score one timeline using partial-interval, whole-interval and momentary time sampling.
• Calculate an interval percentage when some intervals were not observed, following the data sheet's rule.
• Explain why an interval percentage is not a count or an exact duration.
• Decide whether a permanent product can be credited to the target behavior, and state what it cannot show.

You only need one piece of arithmetic: a percentage is part ÷ whole × 100. Your supervisor chooses the procedure, the interval length and the observation instant. Your job is to carry out the plan accurately and report problems.

The basic idea: one mark per interval

An interval is a short, equal block of time on the data sheet. For each interval, you record one mark: did the behavior meet the rule, yes or no? You do not record how many times it happened.

This is called discontinuous measurement. In plain terms, the record has gaps: it does not capture every instance or every second.

There are three common rules for deciding yes or no:

• Partial interval: Mark yes if the behavior happened at any time during the interval, even briefly.
• Whole interval: Mark yes only if the behavior happened throughout the entire interval.
• Momentary time sampling (MTS): Mark yes only if the behavior is happening at one instant set in advance, such as the last second of the interval. What happens before or after that instant does not count.

One timeline, three scores

This example is invented. The interval length is chosen only for practice; a real plan sets its own.

Setup

• Target behavior: humming, defined as audible humming with lips closed.
• Observation: 30 seconds, divided into three 10-second intervals.
  • Interval 1: 0:00 to 0:10, with the MTS check at 0:10.
  • Interval 2: 0:10 to 0:20, with the MTS check at 0:20.
  • Interval 3: 0:20 to 0:30, with the MTS check at 0:30.

What you saw

• Hum A: from 0:03 to 0:05.
• Hum B: from 0:09 to 0:21, without stopping.

No hum starts or stops exactly on a boundary, so it is always clear which interval each moment belongs to.

Scoring each interval

• Interval 1
  • Partial: yes (Hum A).
  • Whole: no (silent from 0:00 to 0:03).
  • MTS: yes, because Hum B is under way at 0:10.
• Interval 2
  • Hum B covers all of it, so partial, whole and MTS are all yes.
• Interval 3
  • Partial: yes (Hum B continues until 0:21).
  • Whole: no (silent after 0:21).
  • MTS: no (silent at 0:30).

Calculating the percentages

You observed all three intervals, so:

Interval percentage = intervals scored yes ÷ intervals observed × 100.

• Partial: 3 ÷ 3 × 100 = 100%
• Whole: 1 ÷ 3 × 100 ≈ 33%
• MTS: 2 ÷ 3 × 100 ≈ 67%

What actually happened

There were two hums, lasting 2 + 12 = 14 seconds in total. That is 14 ÷ 30 × 100 ≈ 47% of the observation time. The same behavior produced three different scores, and none of them is 47%.

What an interval percentage means

An interval percentage is the share of intervals scored yes. It is not a count, and it is not an exact duration. A score of "100% partial interval" does not mean someone hummed the whole time or hummed three times.

Each rule has its own tendency:

• Partial interval tends to overestimate how much of the time the behavior occurred. A 2-second hum earns credit for the whole interval (100% versus the actual 47%).
• Whole interval tends to underestimate it. Any pause turns the interval into a no (33%).
• MTS samples moments. Its percentage estimates the share of the observation period during which the behavior was occurring. It can miss brief events completely: Hum A never appears in the MTS record.

A tempting mistake: Some study materials say MTS estimates the number of occurrences. That is not the standard interpretation. In the example, Hum B was one hum but earned two yes marks, and Hum A earned none. If the plan needs a count, that requires a different measure.

Compare like with like. A score only means something alongside its procedure, interval length and observation instant. For example:

• 50% partial interval and 50% whole interval do not describe the same behavior.
• Two MTS records can disagree if one checks at the start of each interval and the other checks at the end.

Missing intervals are not zeros

Sometimes you cannot observe an interval, for example because you are responding to a safety concern. Do not mark it as no. A "no" claims you watched and saw no behavior, and you did not.

Instead, record the gap the way your approved data sheet or your supervisor's protocol says to, such as a "not observed" mark or the sheet's symbol for it. The protocol also decides how missed intervals affect the percentage. If the rule is unclear, ask your supervisor rather than guessing.

Example: Suppose your protocol leaves not-observed intervals out of the denominator. Twelve intervals were planned. You missed 2, and you scored yes in 5 of the 10 you observed.

• Following that protocol: 5 ÷ 10 × 100 = 50%
• Wrong: scoring the missed intervals as no gives 5 ÷ 12 × 100 ≈ 42%. That treats 2 intervals you never watched as if you saw no behavior.

Report the result with its denominator, such as "50% of 10 observed intervals; 2 not observed," and tell your supervisor why the gap happened.

Safety still comes first. MTS changes when behavior is recorded; it does not suspend any safety or supervision duties assigned by the plan, workplace, or applicable requirements.

Record observations, not interpretations. Write what you saw, such as "hummed." Do not write what you think it meant, such as "was anxious." The same applies to products.

Permanent products: measuring what's left behind

A permanent product is a lasting, measurable result of a behavior that you can check after the behavior is over. Answers on a worksheet are an example.

Before treating a product as evidence of the target behavior, confirm two things:

1. Each occurrence of the target behavior naturally leaves the same kind of product.
2. The product can be produced only by the target behavior.

Basic example

• Target: Riley folds towels in thirds during chore time.
• Riley does the chore alone. No one helps, and no one else folds towels during that time.
• Afterward, you count 8 towels folded in thirds.
• Each fold leaves one folded towel (rule 1), and only Riley's folding could have produced these towels (rule 2). So you record 8.

Change one fact

Riley's parent mentions that they folded a few towels too. Rule 2 now fails, because some of the towels came from someone else. You cannot credit 8 to Riley.

Instead, record what you know ("8 towels folded in thirds; caregiver reports folding some") and tell your supervisor. Always check three things:

• Who produced the item?
• Did anyone help?
• Could the item have come about another way?

What a product cannot tell you

A finished product does not show how the behavior happened. Eight folded towels do not tell you whether the folding took 5 minutes or 40, or whether Riley needed prompts. "Riley folded them independently" is an interpretation unless someone observed it.

Checkpoint: interval recording

Setup

• Target: out of seat, defined as no part of the seat of the pants touching the chair.
• Four 15-second intervals: 0:00–0:15, 0:15–0:30, 0:30–0:45 and 0:45–1:00.
• The MTS check is at the end of each interval.
• Your data sheet's rule: mark missed intervals "not observed" and leave them out of the denominator.

What happened

• The student was out of seat from 0:05 to 0:09.
• The student was out of seat from 0:20 to 0:28.
• The student left the seat at 0:40 and was still out at 0:45.
• During interval 4, you were steadying a tipping step stool, so you did not observe that interval.

Questions

1. Score each observed interval with all three rules, then calculate each percentage.
2. Suppose the plan actually places the MTS check at the start of each interval (0:00, 0:15, 0:30). What is the MTS percentage now?
3. Can you report that the student left the seat three times?

Answer: interval recording

Question 1

• Interval 1: partial yes, whole no, MTS no (seated at 0:15).
• Interval 2: partial yes, whole no, MTS no (seated at 0:30).
• Interval 3: partial yes, whole no, MTS yes (out of seat at 0:45).
• Interval 4 is not observed. Under this data sheet's rule, it is left out, so the denominator is 3.

Results:

• Partial: 3 ÷ 3 = 100%
• Whole: 0 ÷ 3 = 0%
• MTS: 1 ÷ 3 ≈ 33%

If you got 75%, 0% and 25%, you scored interval 4 as a no. That records an observation you never made.

Question 2

At 0:00, 0:15 and 0:30 the student was seated, so MTS = 0 ÷ 3 = 0%. Moving the observation instant changed the score. Results with different instants are not interchangeable.

Question 3

No. Three partial-interval yes marks match three episodes here only by coincidence:

• In the humming example, one long hum earned several yes marks.
• Two brief episodes inside a single interval would earn only one.

You also have no information about interval 4.

Checkpoint: permanent product

Setup

• Target: writes each spelling word correctly when it is read aloud.
• A worksheet comes back with 9 of 10 words correct.
• It was completed at home.
• The word list is printed at the top of the page.

What can you record, and what can you not conclude?

Answer: permanent product

What you can record: "Returned worksheet: 9 of 10 words written correctly." That is an observable fact.

What you cannot do: You cannot credit the result to the target behavior.

• The printed list means the words could have been copied, which is another way to produce the same product.
• Completing it at home leaves authorship and help unknown.
• The sheet shows nothing about how long the task took or whether prompts were used.

Report these limits to your supervisor. Do not record "90% spelling accuracy" as if the target behavior were confirmed.

Recap

• Partial interval = any occurrence. Whole interval = throughout. MTS = at a predetermined instant.
• Percentage = yes intervals ÷ intervals in the denominator × 100. Never mark a missed interval "no"; record it and count it the way your approved data sheet or supervisor's protocol says, and report the denominator.
• Interval percentages are not counts or exact durations:
  • Partial tends to overestimate.
  • Whole tends to underestimate.
  • MTS estimates the share of time and can miss brief events.
• Compare scores only when they share the same procedure, interval length and observation instant.
• MTS changes when you record, not your safety or supervision duties.
• Credit a permanent product only when the behavior reliably produces it and nothing else could have. Check who made it and whether they had help.
• A finished product cannot show process, duration or independence.

Sources

• Behavior Analyst Certification Board. RBT Test Content Outline (3rd ed.), 2023, updated 09/2024.
• Behavior Analyst Certification Board. RBT Ethics Code (2.0), 2021, updated 08/2024.
• Positive Environments, Network of Trainers (PENT), California. Data Collection.
• National Center on Intensive Intervention. Momentary Time-Sampling, Academic Engagement (progress-monitoring tool chart).

Independent study resource

This study resource is not affiliated with, sponsored by, or endorsed by the Behavior Analyst Certification Board (BACB). BACB, Registered Behavior Technician and RBT are registered trademarks of the Behavior Analyst Certification Board. Source citations do not imply endorsement by their authors or publishers.