DRAGENwiki · Article
Frequency and Rate: Counting Behavior Against the Clock
Customize your experience
Sources checked. Clinical review pending.
Frequency starts with a count: how many times a behavior happened. A count becomes useful when you also know how long you watched. Count divided by time is rate, and in much of behavior analysis "frequency" means exactly that.
Frequency and rate in 60 seconds
- Count each one. Event recording "encompasses any procedure in which the frequency of each behavior is recorded during an observation (aka frequency recording)." [4]
- Then add the clock. In applied behavior analysis, frequency generally means "a count (usually of behavior) divided by the time during which it occurred." [3]
- The word has two meanings. "In statistics, however, the term refers to a count of items in a data set." [3] When Merbitz and colleagues wrote in 2016, one major textbook and the BACB used it that way too. [3]
- The fix is simple. State both numbers. Merbitz and colleagues suggest you "simply state both count and time (e.g., ten per class period)." [3]
- It fits some behaviors best. Counting works for behaviors with "clear beginnings and endings" that do not happen too fast to record. [4]
Count, rate and frequency
Count is the tally. Five requests for help is a count.
Rate is the count divided by the time you watched. Five requests in 25 minutes is a rate of one every 5 minutes, or 0.2 per minute.
Frequency is where the confusion lives. Merbitz, Merbitz and Pennypacker (2016) traced how the word has been used. [3]
- In applied behavior analysis, frequency generally means rate: a count per unit of time. [3]
- In statistics, a frequency is simply a count. [3]
- The task lists changed over time. The 1997 second edition, from Florida's certification program, directed readers to "Use rate (sometimes referred to as frequency)." [3] The BACB's third edition (2004) equated frequency with count, and the fourth edition kept that usage. [3]
Today's outlines name these measures without defining them. The technician outline lists frequency under "Implement continuous measurement procedures (e.g., frequency, duration, latency, interresponse time)." [2] It lists rate under "Calculate and summarize data in different ways (e.g., rate, mean duration, percentage)." [2] The BCBA outline's item reads simply "Measure occurrence." [1]
Merbitz and colleagues concluded that "a consistent distinction is elusive." [3] Their advice is to treat rate and frequency as synonyms, "with neither taken to mean simple count, outside of statistical use." [3]
How DRAGEN writes it. On DRAGEN pages, "count" means the tally and "rate" means count divided by time. When we write "frequency," we give the time with it, as in "6 in 30 minutes." That way a reader never has to guess.
Why a count needs a clock
A count on its own can mislead. Merbitz and colleagues put it plainly: "Consider the implications of three in a day or week versus three in a decade." [3]
The same is true in a classroom. Six requests in a 60-minute block and six requests in a 20-minute block are the same count. They are not the same behavior. The first is 1 every 10 minutes. The second is 3 every 10 minutes.
LeBlanc and colleagues warn that it is "not particularly useful to have a frequency count without making note of the observation time." [4] Even a schedule that looks fixed can change: "Even when it seems like an observation window will always be the same (e.g., the school day), certain factors may change the duration of the observation at some point (e.g., the child leaves school ill, an unexpected field trip occurs)." [4]
So write two numbers every time: the count and the minutes you watched.
The same week, told four ways
Merbitz and colleagues show why this matters with a made-up week in a group home. [3] Six incidents were reported, all during bath time. They describe that one week four ways:
- Six incidents last week.
- Four on Tuesday, two on Thursday, none on the other days.
- Four during a two-hour bathing attempt on Tuesday, two during a 30-minute attempt on Thursday.
- All six happened while one aide was helping. Another aide helped on the other days and reported none.
The second version looks like improvement: four a day, then two. The third shows the opposite. The rate doubled, from two per hour to four per hour. [3] The fourth points somewhere new. In the authors' words, it "invites investigation into exactly what each of the aides is doing and reporting." [3]
The authors note that all four are accurate statements about rate. They differ in how precise they are and how much they help a team act. [3] Notice where the fourth version points: at what the adults are doing, not only at the person.
How to turn a count into a rate
- Write the start and stop times. Do it every session, even when the schedule never seems to change. [4]
- Take out time when the behavior could not happen. The simple formula "may conceal complexities such as periods during which the behavior cannot occur." [3] LeBlanc and colleagues advise a denominator that matches the time the behavior could occur. [4] For example, if a learner is out of the room for 10 minutes, leave those minutes out.
- Divide the count by the minutes. LeBlanc and colleagues describe this as the usual path: "a frequency (count) is often converted to a rate measure by dividing it by the duration of the observation window." [4]
- Pick one unit and keep it. Per minute, per 10 minutes, per hour or per school day. Rates in different units "can be converted mathematically from one to another." [3] Use the same unit for every point on a graph.
A quick check. 12 in 40 minutes is 0.3 per minute, 3 per 10 minutes, or 18 per hour. All three say the same thing.
Event recording, step by step
Event recording marks each occurrence as it happens. [4]
- Define one occurrence. Write what the behavior looks like and what ends it, so every observer counts the same thing. See operational definition.
- Decide when it counts again. If a request repeats within a few seconds, does it count once or twice? Write the rule down.
- Tally as it happens. Paper, a hand counter or an app all work. The continuous measurement page compares tools.
- Record the window. Start time, stop time and any minutes when the behavior could not happen.
- Summarize. Event data are "typically summarized as the frequency (count) of responses or response rate (frequency divided by time)." [4]
When counting fits, and when it does not
LeBlanc and colleagues name the conditions. [4]
- A clear start and end. Event recording is "only appropriate for behaviors that have clear beginnings and endings and do not occur so frequently that it is impossible to accurately record them." [4]
- Similar lengths. "Event recording is also best suited for behaviors that occur for comparable durations." [4] One 10-second episode and one 20-minute episode both count as 1.
- Someone watching the whole time. Event recording "requires the observer to constantly monitor the behavior, which may not be feasible in many service environments." [4]
- Other dimensions matter less. When duration and intensity "are not particularly important or are relatively equal across instances of the behavior, event recording is the recommended procedure." [4]
When these do not hold, another method fits better. Long or uneven episodes call for duration. When nobody can watch every second, see discontinuous measurement.
Free and restricted: could it happen right now?
"Many behaviors can occur at any time and in any setting (i.e., a free operant) while others can only occur under certain circumstances (i.e., a restricted operant)." [4]
- Free operant. A comment during play can happen at any moment of the play session. Count it and divide by the time.
- Restricted operant. Answering a question can only happen after someone asks one. A raw rate here mostly tracks how many questions were asked.
For restricted behaviors, LeBlanc and colleagues describe two options. Event data can be summarized as "the percentage of opportunities in which behavior occurred." [4] Or you can keep a continuous measure "but use a denominator that reflects the restricted observation window in which the behavior can occur." [4]
Ask before you divide: could this behavior have happened during every minute I am dividing by?
Other measures give you a count too
You do not always need a separate tally. LeBlanc and colleagues note that "duration recording in which the duration of each behavior is documented also generates a frequency measure." [4] The same holds for latency: "latency recording also generates a frequency measure." [4]
So when another dimension, such as length or intensity, 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." [4] One record answers two questions: how long, and how often.
Rate shows pace
Rate tells you how often something happens in a stretch of time, which a percentage alone does not show. Merbitz and colleagues open by noting that "Skinner once cited rate of response and the cumulative record as his two greatest contributions." [3]
Their examples range from "correct math problems solved in 60 seconds" to affectionate comments to a spouse, counted per week. [3] A short window may hold little besides the behavior you are counting. A long window holds many other behaviors, and stretches when the behavior cannot happen at all, such as times the spouse is away. [3] Both are rates, and both need the time written down.
Measuring yes and no
A count fits a learner's "no" well. Each one matters, and many are brief and easy to mark.
Breaux and Smith note that "A learner will likely use a myriad of topographies to give or withdraw assent" [5], so define each learner's own ways of saying no before you count them. They describe practice in which "Data on the provision of assent, the withdrawal of assent, and the related skills of choice making and functional communication are collected, analyzed, and used as decision-making tools within and across intervention sessions." [5] They add that these skills are not just things to measure: "these are also important outcomes of intervention." [5]
A count of "no" needs a clock too. Four in a 10-minute activity and four across a whole morning tell very different stories.
DRAGEN's rule. We record a learner's "no" as information about the task, the setting or the relationship. It is never "noncompliance" and never a target to reduce. A rising count of "no" is a reason to revisit the plan with the learner and the family. When a behavior can cause harm, it gets its own definition, its own measure and its own safety plan, kept separate from the learner's ways of saying no.
A made-up example: Student D
Student D is 6 years old. D, the family and the teacher agree on a goal: more requests for help or a break during the morning work block, by word, sign or card.
- The definition. A request is D asking an adult for help or a break by word, sign or card. Asking again within 5 seconds counts once.
- The method. The aide uses a hand counter and writes down the start and stop time of each work block. When D leaves for speech therapy, those minutes come out of the time.
- Two days in week one. Monday: 3 requests in 40 minutes. Tuesday: 3 requests in 20 minutes, because an assembly shortened the block. The count is the same. The rates are 0.75 and 1.5 per 10 minutes.
- The plan. Adults prompt and honor every request. A break means a real break.
- Some weeks later. 12 requests in a 40-minute block, a rate of 3 per 10 minutes.
- The "no" line. A separate line records each time D says no or pushes the work away. Each one is honored, and the team looks at what came just before it.
- The check. Once a week a second adult counts too. They record 11 and 12 requests, about 92% agreement on the total. See the lesson on checking your data with IOA.
- Reading it honestly. More requests is the goal here. If requests climb on days with new, harder work, that is information about the work, and the team adjusts the work.
Common myths
- "Frequency always means count." In applied behavior analysis it generally means rate. In statistics it means count. [3] Write the time and nobody has to guess.
- "Same count, same behavior." Not when the observation times differ. [3] [4]
- "Any behavior can be counted." Counting needs clear beginnings and endings and a pace slow enough to record. [4]
- "Counting is the easy option." It asks someone to watch the whole time. [4]
- "One is one." A 10-second episode and a 20-minute episode are each 1 in a count. When lengths vary, measure duration too. [4]
- "The school day is always the same length." Illness and field trips change it. [4]
Words to know
- Count: the number of times a behavior happened.
- Rate: count divided by the time observed, such as 3 per 10 minutes.
- Frequency: in applied behavior analysis, generally the same as rate. In statistics, a count. On DRAGEN pages, always given with its time.
- Event recording: marking each occurrence as it happens. Also called frequency recording. [4]
- Observation window: the minutes you watched, minus any time the behavior could not happen.
- Free operant: a behavior that can happen at any time in the setting. [4]
- Restricted operant: a behavior that can happen only under certain conditions, such as after a question. [4]
- Opportunity: one chance for a restricted behavior to happen.
- Unit of time: the time base you report, such as per minute or per hour.
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:
- Frequency count sheet (PDF)
- Blank graph for counts (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 · discontinuous measurement · operational definition · duration · latency · celeration · interobserver agreement
What remains unknown
- LeBlanc and colleagues write that their selection model "has not been empirically tested." [4] They call it "a potentially useful starting point until validation data can be generated." [4]
- Merbitz and colleagues leave the details of how to count and time each behavior to the situation and the analyst's goals, calling them "beyond the scope of this paper." [3]
- Both meanings of "frequency" were in use when Merbitz and colleagues wrote in 2016. [3] The technician outline names frequency and rate without defining them [2], and the BCBA outline says only "Measure occurrence." [1]
- 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
- 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. https://doi.org/10.1007/s40614-015-0048-z (free full text: https://pmc.ncbi.nlm.nih.gov/articles/PMC6701258/)
- 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/)
- 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
- Duration: Measuring How Long Behavior Lasts
- 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