Three students each get a score of 60%.
One was measured with momentary time sampling. One was measured with partial-interval recording. One was measured with whole-interval recording.
Do those three percentages mean the same thing?
No.
And that is the entire reason these questions get harder than they look. The number on the page may be identical, but the scoring rule that produced the number changes what you are allowed to conclude from it.
A percentage is not a behavioral dimension. The measurement procedure determines what the percentage means.
The current BCBA 6th Edition Test Content Outline separates several measurement skills that work together here:
That last one is especially important. The exam does not have to ask, “What is momentary time sampling?” It can give you a clinical outcome, a staffing limitation, and four plausible measurement procedures, then ask which combination of procedure and interpretation is most defensible.
With continuous measurement, the observer attempts to capture every relevant occurrence or the full relevant temporal dimension during the observation period. Event recording, duration, latency, and interresponse time are familiar examples.
With discontinuous measurement, the observer samples behavior. The observation period is divided into intervals or observation moments, and the scoring rule determines which intervals or moments count as positive.
That makes discontinuous measurement useful when continuous observation is impractical. It also guarantees a tradeoff: you are no longer recording the complete behavioral record.
Do not choose a discontinuous procedure merely because staff are busy. First identify the behavioral dimension that matters.
If the clinical question requires exact count or duration and continuous measurement is feasible, a discontinuous estimate may be the weaker choice.
The strongest answer matches dimension + observation opportunity + scoring rule + interpretation.
With partial-interval recording (PIR), an interval is scored positive if the target behavior occurs at any time during that interval.
Suppose you divide a 10-minute observation into twenty 30-second intervals. During one interval, the behavior occurs for only 2 seconds.
That entire interval still receives a positive score.
If 12 of 20 intervals are positive, you report:
60% of intervals with behavior.
You do not know from that number alone that the person engaged in the behavior for 60% of the observation.
Because even a very brief response can make an entire interval positive, PIR tends to overestimate percentage time when compared with continuous duration. The degree of error depends on properties such as interval length, response duration, behavior level, and session characteristics.
When PIR is 60%, the clean statement is:
The behavior occurred at least once in 60% of the scored intervals.
That statement stays faithful to the scoring rule without converting the percentage into a duration measure it never directly produced.
With whole-interval recording (WIR), an interval is scored positive only if the target behavior occurs throughout the entire interval.
Now use the same 30-second interval.
If engagement lasts for 28 of those 30 seconds, the interval is still scored negative.
That makes whole-interval recording conservative when the percentage is interpreted as an estimate of time engaged. Compared with continuous duration, WIR tends to underestimate percentage time.
Again, do not turn that into a universal slogan like “whole interval measures low behavior.” What matters is the scoring rule.
If WIR is 60%, the clean statement is:
The behavior occurred throughout 60% of the scored intervals.
That does not mean it occurred for only 60% of the session. Behavior may also have occurred during portions of many intervals that were scored negative.
With momentary time sampling (MTS), the observer checks behavior at predetermined observation moments—for example, every 30 seconds, every 2 minutes, or another planned interval.
The observer does not need to know what happened throughout the entire interval. The question is simply:
Is the behavior occurring at this sample moment?
If behavior is occurring at 12 of 20 sample moments, the result is 60% positive samples.
That is also the practical advantage. With whole-interval recording, the observer must monitor behavior throughout the interval. With partial-interval recording, the observer must monitor until the behavior occurs—or through the entire interval if it never occurs. With MTS, the observer needs to observe only at the predetermined sample moment, leaving the rest of the interval available for other responsibilities.
Unlike partial- and whole-interval recording, MTS is not tied to a characteristic direction of error. Under appropriate sampling conditions, the proportion of positive sample moments can be used as an estimate of the proportion of time behavior is occurring.
But “estimate” matters.
MTS can miss behavior that begins and ends between sample moments. Its accuracy depends on factors including interval length, behavior duration and frequency, and whether the observations adequately sample the period of interest.
If 60% of MTS samples are positive, the observer did not directly measure 60% duration.
The direct result is that behavior was present at 60% of sampled moments. Under suitable conditions, that sample can support an estimate of time allocation.
| Procedure | Positive score when... | Clean interpretation | Typical percentage-time bias |
|---|---|---|---|
| Partial interval | Behavior occurs at any point in the interval | % of intervals touched by behavior | Tends to overestimate |
| Whole interval | Behavior occurs throughout the entire interval | % of intervals fully occupied by behavior | Tends to underestimate |
| Momentary time sampling | Behavior is occurring at the sample moment | % of sampled moments with behavior; can estimate time allocation | No fixed direction of error |
A common mistake is learning the procedure but ignoring the sampling schedule.
Shorter intervals generally reduce the amount of sampling error because the measurement system checks behavior more often. But shorter intervals also increase observer response effort and can become impractical when staff are teaching, supervising multiple clients, or tracking several behaviors at once.
There is no universal “correct” interval length that works for every behavior and every setting.
A 2-minute MTS interval might be usable for relatively long episodes of pacing and a poor choice for 2-second instances of aggression. The same clock interval can be reasonable for one response class and badly mismatched to another.
Interval length is part of the measurement system, not a formatting choice.
Many candidates learn a shortcut that sounds something like this:
Use partial interval for behaviors you want to reduce; use whole interval for behaviors you want to increase.
That mnemonic points toward the directional bias of the scoring systems, but it is too rigid to use as a selection rule.
The rationale behind it is real, and worth knowing: because PIR tends to overestimate and WIR tends to underestimate, PIR errs on the conservative side when the goal is to reduce behavior (improvement is not overstated), and WIR errs on the conservative side when the goal is to increase it (progress is not overstated). Just do not treat that as the whole selection rule.
The better question is:
What dimension matters clinically, what can the observer actually record, and what will the resulting percentage mean?
If exact duration matters and staff can record continuously, continuous duration may be better than either interval procedure. If only brief point observations are feasible and the clinical question concerns time allocation, MTS may be a better fit. If you use PIR or WIR, interpret the output according to its actual scoring rule rather than pretending the percentage is exact duration.
These questions are designed to make you identify the dimension, observation constraint, scoring rule, and defensible interpretation before choosing an answer.
A client's pacing occurs in long episodes, typically lasting 5–15 minutes. The clinical outcome is the proportion of session time spent pacing. Continuous duration recording is not feasible because staff supervise several clients, but staff can reliably make a brief observation every two minutes.
Which measurement-and-interpretation pair is BEST?
Why: the target dimension is time allocation, but the observation opportunity is limited to brief point observations. MTS matches that observation pattern and can support an estimate of proportion of time when the sampling arrangement is appropriate. It remains an estimate—not direct duration. Episodes this long are unlikely to begin and end between two-minute sample moments.
Why the others miss: PIR and WIR require information about what happened across the interval, and neither percentage is exact duration. Event recording answers how often episodes occurred, not how much session time they occupied.
A BCBA uses 30-second partial-interval recording for yelling. Yelling is scored in 18 of 30 intervals.
Which conclusion is MOST defensible?
Why: partial interval tells you whether the behavior occurred at any point during each interval. Eighteen positive intervals out of thirty means the behavior touched 60% of intervals.
Boundary: because a one-second response and a twenty-nine-second response can both produce the same positive interval score, the percentage cannot be converted into exact duration.
A teacher uses 20-second whole-interval recording for academic engagement. Engagement is scored positive in 21 of 30 intervals.
What does the 70% result mean?
Why: whole interval requires the behavior to persist throughout the full interval. A negative interval does not mean zero engagement occurred; even 19 seconds of engagement in a 20-second interval would still be scored negative.
That is why WIR tends to underestimate percentage time when compared with continuous duration.
A client's hand-to-head hits are brief, discrete, and easy to count. The clinical team needs the exact response rate, and a dedicated observer can continuously monitor the entire 20-minute session.
Which measurement procedure is BEST matched to the clinical question?
Why: the team wants exact rate, the responses are discrete and countable, and continuous observation is available. There is no reason to replace a direct count with a sampling estimate simply because interval methods are familiar.
This is C.9 reasoning: match the procedure to the critical dimension and the actual environmental constraints.
The calculations themselves are easy:
positive intervals or samples ÷ total intervals or samples × 100.
The difficult part is deciding what the numerator means.
For PIR, a positive interval means some behavior happened.
For WIR, a positive interval means behavior persisted throughout.
For MTS, a positive sample means behavior was occurring at the observation moment.
Those are not interchangeable events. Therefore, the resulting percentages are not interchangeable measurements.
Last week's article made one measurement point repeatedly: a statistic only answers the question it was designed to answer.
IOA answers agreement, not accuracy. A validity judgment answers whether the measure addresses the relevant target or dimension, not whether two observers happened to agree.
The same principle applies here.
A discontinuous measurement percentage is not self-interpreting. You have to know the procedure that generated it.
Do not ask, “What percentage did I get?” Ask, “What had to happen for an interval or sample to count as positive?”
Once you answer that, most of the interpretation problem disappears.
BA Prep questions are built to make you interpret the measurement system, not just recognize vocabulary. If you can explain what a positive interval means before you calculate the percentage, you are solving the harder part of the problem.
Start Free TrialBehavior Analyst Certification Board. BCBA Test Content Outline (6th ed.). See Tasks C.5, C.6, and C.9. BACB 6th Edition TCO
Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.). Pearson.
Fiske, K., & Delmolino, L. (2012). Use of discontinuous methods of data collection in behavioral intervention: Guidelines for practitioners. Behavior Analysis in Practice, 5(2), 77–81. https://doi.org/10.1007/BF03391826
Powell, J., Martindale, A., & Kulp, S. (1975). An evaluation of time-sample measures of behavior. Journal of Applied Behavior Analysis, 8(4), 463–469. https://doi.org/10.1901/jaba.1975.8-463
Powell, J., Martindale, B., Kulp, S., Martindale, A., & Bauman, R. (1977). Taking a closer look: Time sampling and measurement error. Journal of Applied Behavior Analysis, 10(2), 325–332. https://doi.org/10.1901/jaba.1977.10-325
Saudargas, R. A., & Zanolli, K. (1990). Momentary time sampling as an estimate of percentage time: A field validation. Journal of Applied Behavior Analysis, 23(4), 533–537. https://doi.org/10.1901/jaba.1990.23-533
LeBlanc, L. A., Lund, C., Kooken, C., Lund, J. B., & Fisher, W. W. (2020). Procedures and accuracy of discontinuous measurement of problem behavior in common practice of applied behavior analysis. Behavior Analysis in Practice, 13(2), 411–420. https://doi.org/10.1007/s40617-019-00361-6