Frequency vs Duration Recording in ABA: Differences, Examples and When to Use Each
Frequency vs duration in ABA: how frequency recording differs from duration recording, when to convert to rate, a worked example and 6 practice questions.
By Praxis Notes Team Editorial team
Updated 14 min read
On this page
Frequency recording counts how many times a behavior occurs; duration recording times how long it lasts, from onset to offset. Count brief behaviors with a clear start and end, and convert the count to rate when sessions differ in length. Time behaviors whose length matters or varies, such as tantrums or on-task work (LeBlanc et al., 2016).
This page is for RBT candidates and BCBA trainees first, then RBTs who want to know why their plan asks for a tally or a stopwatch. Its centre is one fictional learner whose count, rate and duration point three different ways, followed by six practice questions. More explainers are in Behavior Analysis Concepts.
What is the difference between frequency and duration data in ABA?
Frequency recording differs from duration recording in what it captures: frequency data show how many times a behavior happened, and duration data show how long it lasted. Both are continuous measures: the observer records every instance across the whole observation instead of sampling it.
| Question | Frequency recording | Duration recording |
|---|---|---|
| What it measures | How many times the behavior occurs (a count) | How long the behavior lasts, from onset to offset |
| How you record it | A tally mark, clicker or app tap for each occurrence | A timer started at onset and stopped at offset for each occurrence |
| How it is summarized | Count with the observation time, or rate (count ÷ time) | Total duration, mean duration per occurrence, or percentage of time |
| Fits | Brief behavior of about the same length each time | Behavior whose length varies or matters: tantrums, on-task work |
| Does not fit | Behavior that lasts long or varies widely in length | Very brief behavior, where every occurrence lasts a second or two |
| Example result | “Hitting: 8 times in a 40-minute session” | “Crying: 3 episodes, 12 minutes in total, in a 60-minute session” |
In LeBlanc and colleagues’ (2016) measurement-selection model, event recording (frequency) is “only appropriate for behaviors that have clear beginnings and endings” and “best suited for behaviors that occur for comparable durations,” while duration recording is the preferred procedure when how long the behavior lasts is the dimension of interest (LeBlanc, Raetz, Sellers & Carr, 2016).
What is frequency recording?
Frequency recording, also called event recording, is counting each occurrence of a defined behavior during an observation. LeBlanc and colleagues (2016) note that event-recording data are “typically summarized as the frequency (count) of responses or response rate (frequency divided by time).”
How to record it:
- Read the operational definition, including what counts as one occurrence and when a second one starts.
- Note the start time of the observation.
- Mark every occurrence as it happens.
- Note the end time, and write the count with the observation time: “Mands: 14 in 60 minutes.”
A count is only meaningful next to its observation time. LeBlanc and colleagues (2016) say it is “not particularly useful to have a frequency count without making note of the observation time.” Fourteen requests in an hour and fourteen in a school day are different levels of behavior.
Frequency fits behaviors such as hitting, mands, hand-raising and throwing: each occurrence is short, has a clear start and end, and lasts about the same time as the others.
What is duration recording?
Duration recording is timing each occurrence of a behavior from the moment it starts (onset) to the moment it stops (offset). LeBlanc and colleagues (2016) say duration data are “typically summarized as the mean duration of each behavior during an observation or the total duration of all behaviors during an observation.”
How to record it:
- Read the operational definition, including the onset and offset rules (for example, “crying ends after 10 seconds of quiet”).
- Start the timer at onset and stop it at offset, every time.
- Write each episode’s time, not only the running total.
- Summarize the session three ways if the plan asks for them:
- Total duration: all episode times added together.
- Mean duration per occurrence: total duration ÷ number of episodes.
- Percentage of the session: total duration ÷ observation time × 100.
Writing each episode down matters because “duration recording in which the duration of each behavior is documented also generates a frequency measure” (LeBlanc et al., 2016): the number of episodes is the count.
Duration fits behaviors such as tantrums, crying, out-of-seat time, on-task work and tolerating a wait, where one occurrence can last seconds or many minutes.
What is rate, and when do you convert frequency to rate?
Rate is the count divided by the observation time, such as 0.25 responses per minute (15 per hour). Convert a count to rate whenever observation periods differ in length, because a raw count then mixes up how much behavior happened with how long you watched.
Rate = count ÷ observation time
Fictional example. Noah’s RBT counts his mands. On Monday the session lasts 60 minutes and she counts 14 mands; on Wednesday it lasts 30 minutes and she counts 9.
- Monday: 14 ÷ 60 = 0.23 mands per minute (about 14 per hour)
- Wednesday: 9 ÷ 30 = 0.30 mands per minute (18 per hour)
The count says Noah asked for things less often on Wednesday. The rate says he asked more often. Only the rate can be compared across the two sessions. RBTs calculate rate under task A.6 of the RBT Test Content Outline (3rd ed.), “Calculate and summarize data in different ways (e.g., rate, mean duration, percentage).” The Praxis Learn lesson on calculating and summarizing data has more examples. Rate is also compared with percentage measures in our guide to rate vs percentage of occurrence.
Worked example: when count, rate and duration disagree
Fictional example. Leo, 6, has tantrums during work periods. His plan defines a tantrum as crying, screaming or dropping to the floor; it ends after 30 seconds without any of these. His RBT records the start and end of every tantrum.
| Session | Observation | Tantrum lengths (minutes) | Count | Rate | Total duration | Mean duration per tantrum | Percentage of session |
|---|---|---|---|---|---|---|---|
| Monday | 120 minutes | 1, 2, 1, 3, 2, 1 | 6 | 0.05/min (3/hour) | 10 minutes | 1.7 minutes (1 min 40 s) | 8.3% |
| Thursday | 60 minutes | 4, 7, 9 | 3 | 0.05/min (3/hour) | 20 minutes | 6.7 minutes (6 min 40 s) | 33.3% |
The working:
- Rate: Monday 6 ÷ 120 = 0.05 per minute; Thursday 3 ÷ 60 = 0.05 per minute.
- Mean duration: Monday 10 ÷ 6 = 1.7 minutes; Thursday 20 ÷ 3 = 6.7 minutes.
- Percentage of session: Monday 10 ÷ 120 × 100 = 8.3%; Thursday 20 ÷ 60 × 100 = 33.3%.
Figure 1 draws Thursday’s session on one timeline and adds the two other continuous measures, latency and interresponse time (IRT), from a fictional instruction at minute 0 and fictional tantrum start times.

What each measure says about Thursday:
- Count: half as many tantrums (3 instead of 6). Looks better, but Thursday’s session was half as long.
- Rate: no change (3 per hour both days).
- Duration: much worse. Each tantrum lasted about four times as long, and tantrums filled a third of the session instead of a twelfth.
If the team cares about how much of Leo’s day tantrums take up, which is usually the point for a long, variable behavior, only duration shows the problem. Recording each tantrum’s start and end gave all three measures from one data sheet. One fictional pair of sessions shows how the measures work, not what any real learner’s data look like.
Frequency or duration: which should you use?
Use frequency (and rate) when the behavior is brief, has a clear start and end and lasts about the same each time; use duration when how long it lasts is what matters or when its length varies. The supervising BCBA chooses the measure; the RBT records it as written and reports when it does not fit the behavior.
| The question the team needs answered | Measure to record |
|---|---|
| How often does a brief behavior happen? | Frequency, reported with the observation time |
| How often, when sessions vary in length? | Frequency converted to rate |
| How long does each episode last? Is it getting shorter? | Duration per episode, summarized as mean duration |
| How much of the session does the behavior fill? | Total duration, or percentage of the session |
| How long does the learner keep working or waiting? | Duration (a behavior to increase) |
| How quickly does the learner start after an instruction? | Neither: latency (see latency vs IRT) |
| The behavior has no clear start or end, or the observer cannot watch continuously | A discontinuous method such as momentary time sampling |
Figure 2 turns the table into a decision chart.

The choice follows the dimension of the behavior that matters, not its function. BCBA task C.9 of the BCBA Test Content Outline (6th ed.) asks candidates to “select a measurement procedure to obtain representative data that accounts for the critical dimension of the behavior and environmental constraints.” The Praxis Learn BCBA lesson on selecting a measurement procedure covers that task, and the lesson on measuring occurrence covers count and rate (C.3).
A clear operational definition comes first for both: frequency needs a rule for when one occurrence ends and the next begins, and duration needs onset and offset rules. See operational definitions in ABA.
What is a disadvantage of using rate as a measurement?
The main disadvantage of rate is that it says nothing about how long each occurrence lasts: in the worked example, Leo’s rate was the same on both days while his tantrum time doubled. Rate has two other limits:
- It assumes the behavior could happen throughout the observation. Merbitz, Merbitz and Pennypacker (2016) point out that “count divided by time may conceal complexities such as periods during which the behavior cannot occur,” such as time asleep or away from the setting. LeBlanc and colleagues (2016) suggest a denominator “that reflects the restricted observation window in which the behavior can occur.”
- It fits poorly when the teacher controls the opportunities. In discrete trial teaching the number of trials is set by the instructor, so correct responses are usually reported as a percentage of trials, not per minute; LeBlanc and colleagues (2016) note that behavior with limited opportunities is often summarized as a percentage of opportunities.
Rate also needs both numbers to be right: a wrong session length gives a wrong rate even when the count is perfect.
Does “frequency” mean count or rate?
On the RBT exam, read “frequency” as a count unless the question gives a time unit (“per minute,” “per hour”). In our reading, the current outline points the same way: the RBT Test Content Outline (3rd ed.) lists frequency under recording (A.1) and rate separately under calculating (A.6). The term is used both ways in behavior analysis:
- Carr, Nosik and Luke (2018) surveyed journals and textbooks and found “the predominant usage of frequency was as count, not rate,” and they “encourage behavior analysts to use frequency as a synonym for count” (JABA, 51(2)).
- Merbitz, Merbitz and Pennypacker (2016) argue the opposite: within ABA, frequency “should not refer to ‘count’ but instead to ‘rate’,” and they note that Cooper, Heron and Heward’s 2007 edition equated frequency and rate. For students they suggest “use frequency as a synonym for count for the BACB examinations” (The Behavior Analyst, 39(2)).
Merbitz and colleagues (2016) add that counts reported without the observation time should be avoided, and LeBlanc and colleagues (2016) make the same point. In our reading, the safest habit on the job is to write both numbers (“14 mands in 60 minutes”) so the data work under either definition.
Can you record frequency and duration together?
Yes: if you time every episode, you also have the count, and from the two you can calculate rate, total duration and mean duration. That is why many plans for tantrums ask for the start and end time of each episode rather than a single total.
Recording both is most useful when a behavior varies in length, as in the worked example. For brief, uniform behaviors such as hitting, timing each one adds work without adding information; the count is enough.
Practice questions on frequency vs duration
Every scenario is fictional. Decide before you open the answer.
1. A.1: Episodes of different lengths
Jayden screams in episodes that last anywhere from a few seconds to 15 minutes. His BCBA wants to know whether a new routine is making the episodes shorter. What should the RBT record?
- A. A tally of each scream episode
- B. The rate of scream episodes per hour
- C. The duration of each episode, from onset to offset
- D. The percentage of trials with screaming
Answer and rationale
C. Whether episodes are getting shorter is a question about duration, and timing each episode also gives the count. A and B count episodes but cannot show that a 15-minute episode became a 2-minute one. D is for behavior measured against a set number of trials. Task A.1.
2. A.1: Brief, uniform behavior
Ella raises her hand to answer questions. Each hand-raise lasts a second or two and has a clear start and end. Her teacher wants to know whether she is participating more often. Which measure fits best?
- A. Frequency (event recording), with the observation time noted
- B. Duration of each hand-raise
- C. Latency from the question to the hand-raise
- D. Whole interval recording
Answer and rationale
A. Brief behavior with a clear start and end and similar length each time is what event recording suits (LeBlanc et al., 2016). B times something that barely varies. C answers how quickly she responds, not how often. D would almost never score an interval, because a hand-raise does not last a whole interval. Task A.1.
3. A.6: Sessions of different lengths
An RBT counts 12 hits in a 40-minute session on Tuesday and 15 hits in a 75-minute session on Friday. Which statement is accurate?
- A. Hitting increased on Friday, because 15 is more than 12
- B. Hitting was the same on both days
- C. The two sessions cannot be compared
- D. The rate of hitting was higher on Tuesday
Answer and rationale
D. Tuesday: 12 ÷ 40 = 0.30 per minute. Friday: 15 ÷ 75 = 0.20 per minute. A compares raw counts from sessions of different lengths. B is wrong because the rates differ: 0.30 against 0.20 per minute. C is wrong because converting to rate makes the sessions comparable. Task A.6.
4. A.6: The limit of rate
A learner’s aggression occurs at 2 episodes per hour in both baseline and treatment, but the episodes in treatment last much longer. What does this show about rate?
- A. Rate is inaccurate when sessions are the same length
- B. Rate does not show how long each occurrence lasts
- C. Rate should be replaced with a raw count
- D. Rate overestimates behavior that occurs in long episodes
Answer and rationale
B. Rate is count per unit of time, so two sessions with the same number of episodes per hour have the same rate however long the episodes last. A is backwards: rate works fine for equal sessions. C loses the time base. D invents a bias: rate counts each episode once, whatever its length. Task A.6.
5. A.1: Reading the word “frequency”
A practice question says: “The RBT recorded the frequency of elopement.” No time unit is given. What did the RBT most likely record?
- A. How long each elopement lasted
- B. The time from an instruction to the elopement
- C. The number of times elopement occurred during the observation
- D. The percentage of intervals with elopement
Answer and rationale
C. Without a time unit, read “frequency” as a count (Carr, Nosik & Luke, 2018; Merbitz et al., 2016 give the same advice for BACB exams). A is duration, B is latency and D is an interval method. Task A.1.
6. A.1: Spot the wrong definition
A study card says: “Duration recording documents the amount of time between behaviors.” What is wrong with it?
- A. Nothing; that is the definition of duration
- B. It describes latency
- C. It describes rate
- D. It describes interresponse time (IRT); duration is how long each behavior lasts
Answer and rationale
D. Time between one response and the next is IRT; duration runs from the onset to the offset of the behavior itself. B is close but latency starts at a stimulus, such as an instruction, not at the previous response. C is a count per unit of time. Task A.1.
For more items on Domain A, try the RBT Data Collection and Graphing practice exam.
What to study next
- Task A.1 as a whole (frequency, duration, latency and IRT, with a session timeline to measure): continuous measurement for RBT task A.1, and the Praxis Learn RBT A.1 continuous measurement lesson.
- Rate, mean duration and percentage (task A.6): the Learn lesson on calculating and summarizing data.
- When you cannot watch every second: partial interval vs whole interval vs momentary time sampling.
Sources
- Behavior Analyst Certification Board. RBT Test Content Outline (3rd ed.). Updated 09/2024. Tasks A.1, A.6.
- Behavior Analyst Certification Board. BCBA Test Content Outline (6th ed.). Updated 09/2024. Tasks C.3, C.9.
- Carr, J. E., Nosik, M. R., & Luke, M. M. (2018). On the use of the term ‘frequency’ in applied behavior analysis. Journal of Applied Behavior Analysis, 51(2), 436–439.
- LeBlanc, L. A., Raetz, P. B., Sellers, T. P., & Carr, J. E. (2016). A proposed model for selecting measurement procedures for the assessment and treatment of problem behavior. Behavior Analysis in Practice, 9(1), 77–83.
- Merbitz, C. T., Merbitz, N. H., & Pennypacker, H. S. (2016). On terms: Frequency and rate in applied behavior analysis. The Behavior Analyst, 39(2), 333–338.
All examples on this page are fictional. Sources last checked on October 29, 2026.
Updated October 29, 2026: rewritten around the choice between count, rate and duration, with a worked example and practice questions mapped to the RBT Test Content Outline (3rd ed.). Removed unsupported claims, including a rule tying the measure to the function of the behavior and references to the retired RBT Task List as current.
Reviewed by the Praxis Notes team on October 29, 2026.