DRAGENwiki · Article
Duration: Measuring How Long Behavior Lasts
Customize your experience
Sources checked. Clinical review pending.
Duration is how long a behavior lasts, from the moment it starts to the moment it stops. You time each episode, then report the total, the average or the share of the session it filled. Choose duration when how long matters more than how many.
Duration in 60 seconds
- Time each episode. Duration recording "involves the measurement of the amount of time each behavior occurs during an observation." [3]
- Start and stop. Each record runs from onset to offset. LeBlanc and colleagues put it simply: "Record the amount of time from onset to offset for each behavior as it occurs". [3]
- Three summaries. The same record gives you "Total duration, mean duration, percentage duration". [3]
- A count comes with it. A duration record "also generates a frequency measure". [3]
- It costs attention. Duration recording "requires constant vigilance" and a timer. [3]
Where duration fits
Duration is one of the temporal dimensions of behavior, along with latency and interresponse time. The BCBA outline asks analysts to "Measure temporal dimensions of behavior (e.g., duration, latency, interresponse time)." [1] The technician outline lists duration in "Implement continuous measurement procedures (e.g., frequency, duration, latency, interresponse time)." [2] It also names a way to summarize it: "Calculate and summarize data in different ways (e.g., rate, mean duration, percentage)." [2]
- Duration asks how long each episode lasted.
- Latency asks how long it took to start after a cue.
- Frequency and rate ask how many, and how many per minute.
Duration is a form of continuous measurement. In their review of journal articles, Mudford, Taylor and Martin counted records as continuous when they "contained second-by-second records of occurrences of discrete behaviors or the onsets and offsets of behaviors with duration." [5] Duration needs both ends of every episode.
Duration is common in research, though less common than rate. Summarizing that review, Mudford, Martin, Hui and Taylor report that 95% of the articles reported rate, and "duration measures were reported in 36% of articles reviewed." [4]
When duration is the right measure
LeBlanc and colleagues call duration recording the preferred procedure when how long the behavior lasts "is the most relevant dimension of interest." [3] Some questions where length is the point:
- How long does a learner stay with a book, a game or a friend? More minutes of chosen reading or play can be a goal.
- How long does a hard moment last? When a plan aims to help a person feel better sooner, with support, the minutes show whether that is happening.
- How long does a routine take? Getting dressed, or finishing a page of math problems.
A count hides length. "Event recording is also best suited for behaviors that occur for comparable durations." [3] A 10-second episode and a 20-minute episode each count as 1 on a tally. When episodes vary that much, the count alone misses most of the story.
The rule also works the other way. When the duration and intensity of a behavior "are not particularly important or are relatively equal across instances of the behavior, event recording is the recommended procedure." [3] In that case, see frequency and rate.
Onset and offset: where each episode starts and stops
A timer is only as good as the rule that starts and stops it. In one accuracy study, observers measuring duration "were to press the code key (or touch the name) for its onset and again for its offset." [4] Every duration record depends on those two moments.
Write both into the operational definition:
- Onset. The first moment that counts. "Sits at the game table and takes a turn" is clearer than "starts playing."
- Offset. The moment the episode ends. "Leaves the table" is a clear offset.
- The pause rule. How long a break must last before the episode is over. Without one, a single observer may stop the clock at a 5-second pause while another keeps it running. Write a number, such as 30 seconds, and say whether the clock stops at the start of that pause or at its end.
Two people reading the same definition should start and stop their timers at the same moments. If they would not, the definition needs more detail.
Total, mean and percentage duration
LeBlanc and colleagues write that "Duration-recording data are typically summarized as the mean duration of each behavior during an observation or the total duration of all behaviors during an observation." [3] Their table adds a third summary: percentage duration. [3]
A worked example. In a 30-minute session, a learner reads three times: for 4 minutes, 6 minutes and 5 minutes.
- Total duration: 4 + 6 + 5 = 15 minutes.
- Mean duration: 15 ÷ 3 = 5 minutes per episode.
- Percentage duration: 15 ÷ 30 = 50% of the session.
- Count: 3 episodes, or 1 per 10 minutes.
Each summary answers a different question:
- Total tells you how much time the behavior filled.
- Mean tells you how long a typical episode lasts. A total alone cannot tell one 15-minute episode from fifteen 1-minute episodes.
- Percentage lets you compare sessions of different lengths. In Mudford and colleagues' study, relative durations were "expressed as the percentage of each session in which the behavior was observed". [4]
Percent of what? Divide by the time when the behavior could happen. LeBlanc and colleagues advise keeping a continuous measure "but use a denominator that reflects the restricted observation window in which the behavior can occur in the conversion to a rate or percentage duration measure". [3] If the book corner is closed for 10 of the 30 minutes, reading can fill at most 20 minutes. Then 15 minutes is 75% of the time reading was possible.
Duration gives you a count too
You do not need a second tally sheet. LeBlanc and colleagues note that "duration recording in which the duration of each behavior is documented also generates a frequency measure". [3] Count the episodes in your record and you have a count. Divide by the minutes observed and you have a rate.
They go further. When another dimension can vary greatly, they suggest you "capture a measure of that dimension (e.g., duration, intensity) and derive the frequency count and rate from the number of observed instances." [3] One duration record answers two questions: how long, and how often.
Recording duration, step by step
- Write the onset, offset and pause rule. Keep them where every observer can see them.
- Have a timer ready. "Duration recording also requires a timing device (e.g., stopwatch, clock app) that must be easily accessible yet unobtrusive." [3]
- Write the session's start and end times. Note any minutes when the behavior could not happen.
- Start at onset, stop at offset. Write each episode on its own line, either as minutes and seconds or as start and stop clock times.
- Sum it up. Use the same summary every session (total, mean or percentage) so each point on the graph means the same thing.
Paper works. The duration recording sheet gives each episode its own row. The continuous measurement page compares paper, counters and apps.
What makes duration hard
It takes constant attention. "Like event recording, however, duration recording requires constant vigilance, which might limit its practicality." [3] An adult who is also teaching can miss an onset or an offset. When no one can watch every second, discontinuous measurement can estimate duration instead. LeBlanc and colleagues add a caution: partial-interval recording "does not produce complete data about any behavioral dimension but instead generates an estimate of the frequency and duration of behavior." [3]
Many short episodes are hard to catch. Mudford, Martin, Hui and Taylor (2009) had observers record from video samples. Then "Data files were compared with criterion records to determine observer accuracy." [4] For duration, "Observers' accuracy at recording the medium-duration behavior was the lowest by all measures of accuracy". [4]
The total time was not what made that sample hard. "Among the 26 occurrences, 21 were less than 10 s in duration, of which 18 were less than 5 s in duration." [4] The authors suggest that "the complexity of the observers' task in recording the medium-duration behavior may be the underlying reason for lower accuracy." [4] Each brief episode asks for two quick presses, one at the start and one at the end.
If a behavior comes in many brief bursts, ask whether length matters for the goal. If it does not, a count may serve the team better.
Checking duration data
A second adult can time the same session, and you compare the two records. Three findings from Mudford and colleagues' study are worth knowing:
- Agreement is not accuracy. "Agreement between observers generally exceeded accuracy when compared with criterion records of the behavior." [4] Two people can agree and both be off.
- The method of comparing matters. "Block-by-block and exact agreement algorithms were susceptible to inflated agreement and accuracy estimates at lower rates and durations." [4]
- Leeway helps duration less. The authors also tried a tolerance: when two records are compared, a press within a few seconds of the other record still counts as agreement. More tolerance helped counts more than duration: "Accuracy in recording behaviors with duration increased as tolerance was increased; however, gains were less evident than those observed for event recording." [4]
The interobserver agreement page and the lesson How to check your data with interobserver agreement (IOA) show how to compare two records.
Measuring yes and no
Time can show choice. A learner who stays with a game for 10 minutes, free to leave at any moment, is telling you something about the game. That condition matters: minutes of engagement mean something only when leaving is truly allowed.
Breaux and Smith's table of measures for assent and assent withdrawal lists counts and latencies, such as "Frequency of learner initiating engagement in treatment activities or with treatment materials". [6] Duration can sit beside measures like these. "A learner will likely use a myriad of topographies to give or withdraw assent" [6], so define each learner's own ways of saying no before you measure anything around them.
DRAGEN's rule. We never time a learner's "no" in order to make it shorter. A "no" is information about the task, the setting or the relationship. It is never "noncompliance" and never a target. When minutes of engagement drop, the team asks what changed in the activity. When a behavior could cause harm, it gets its own definition, its own measure and its own safety plan, separate from the learner's ways of saying no.
A made-up example: Student E
Student E is 8. E, the family and the teacher agree on a goal: more time playing board games with classmates during the 20-minute free-choice block.
- The definition. Onset: E sits at a game with at least one classmate and takes a turn or watches a classmate's turn. Offset: E leaves the game, or 30 seconds pass with no play. The clock stops at the start of that pause.
- The method. The aide uses a phone stopwatch, writes each episode's minutes and seconds, and notes the block's start and end times. When E leaves early for speech therapy, those minutes come out of the window.
- First week, Monday. Two episodes: 2 minutes and 3 minutes. Total 5 minutes, mean 2.5 minutes, 25% of the 20-minute block, count 2.
- The plan. E picks the game and the partner. An adult joins at first to show turn-taking, then steps back. E may leave any game at any time.
- Some weeks later. Three episodes: 4, 6 and 5 minutes. Total 15 minutes, mean 5 minutes, 75% of the block.
- The "no" line. A separate line notes each time E says no to a game or picks something else. Each one is honored, and the team looks at what was offered.
- The check. Once a week, a second adult times the same block. One week their totals were 15 minutes and 14.5 minutes. See the lesson How to check your data with interobserver agreement (IOA).
- Reading it honestly. More minutes is the goal only while E is choosing them. If the minutes rose because E was not allowed to leave, the number would no longer measure what the team cares about.
Common myths
- "Longer is always better." Only when the learner is choosing the minutes, and only for behavior the person wants more of.
- "Total duration says it all." One 15-minute episode and fifteen 1-minute episodes both total 15 minutes. Report the mean or the count as well.
- "You need special equipment." A stopwatch or clock app works. [3]
- "Timing and counting are separate jobs." A duration record also gives you a count. [3]
- "A lower total means less of the behavior." Not if the session was shorter. Percentage duration accounts for that. [3]
- "Brief episodes are easy to time." In one study, many brief episodes were the hardest for observers to record accurately. [4]
Words to know
- Duration: how long an episode lasts, from onset to offset.
- Onset: the moment an episode starts.
- Offset: the moment an episode ends.
- Episode: one occurrence of the behavior, from onset to offset.
- Total duration: all episodes added together. [3]
- Mean duration: the total duration divided by the number of episodes. [3]
- Percentage duration: the total duration divided by the time observed, times 100. [3] Also called relative duration. [4]
- Pause rule: how long a break must last before an episode ends.
- Timing device: a stopwatch, clock app or recording program. [3]
Need a PDF?
Create it here, for free: the DRAGEN app. Use it at your next session or team meeting.
Or print one we already made, also free:
- Duration recording sheet (PDF)
- Blank graph for percentages (PDF)
- Which measure? card (PDF)
- IOA check sheet (PDF)
- Every free PDF
More from DRAGEN
- Lesson: How to count a behavior: frequency, duration, latency and interval recording
- Lesson: How to check your data with interobserver agreement (IOA)
- Lesson: How to read a behavior graph in sixty seconds
- Free app for parents: DRAGENparent
- BCBA care: DRAGENtherapy
- Books: DRAGENbooks
- Talk with Chase: Contact
- Home: dragen.cloud
Related DRAGENwiki pages: measuring behavior · continuous measurement · frequency and rate · latency · discontinuous measurement · operational definition · percentage · interobserver agreement
What remains unknown
- LeBlanc and colleagues write that their selection model "has not been empirically tested." [3] They offer it as "a potentially useful starting point until validation data can be generated." [3]
- When Mudford and colleagues wrote in 2009, "there are no empirical studies to guide researchers on interpreting agreement or accuracy indexes produced by time-window analysis." [4]
- The certification outlines name duration and mean duration, but they do not say which summary to report. [1] [2]
- This page is general information, not medical or legal advice.
- This page was last checked on November 12, 2026.
Sources
- Behavior Analyst Certification Board. BCBA Test Content Outline (6th ed.). https://www.bacb.com/wp-content/bcba-outline-6thEd/
- Behavior Analyst Certification Board. RBT Test Content Outline (3rd ed.), effective January 1, 2026. https://www.bacb.com/rbt-test-content-outline-3rd-ed
- 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. https://doi.org/10.1007/s40617-015-0063-2 (free full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC4788644/)
- Mudford, O. C., Martin, N. T., Hui, J. K. Y., & Taylor, S. A. (2009). Assessing observer accuracy in continuous recording of rate and duration: Three algorithms compared. Journal of Applied Behavior Analysis, 42(3), 527–539. https://doi.org/10.1901/jaba.2009.42-527 (free full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC2741073/)
- Mudford, O. C., Taylor, S. A., & Martin, N. T. (2009). Continuous recording and interobserver agreement algorithms reported in the Journal of Applied Behavior Analysis (1995–2005). Journal of Applied Behavior Analysis, 42(1), 165–169. https://doi.org/10.1901/jaba.2009.42-165 (free full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC2649837/)
- Breaux, C. A., & Smith, K. (2023). Assent in applied behaviour analysis and positive behaviour support: Ethical considerations and practical recommendations. International Journal of Developmental Disabilities, 69(1), 111–121. https://doi.org/10.1080/20473869.2022.2144969 (free full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC9897747/)
Pages that link here
- Celeration: How Fast Is Learning Growing?
- Continuous Measurement: Recording Every Time a Behavior Happens
- Discontinuous Measurement: Recording a Sample of Behavior
- Frequency and Rate: Counting Behavior Against the Clock
- Latency: Measuring How Long It Takes to Start
- Measuring Behavior: Turning What You See Into Numbers You Can Trust
- Operational Definition: Describing Behavior So Anyone Can Count It