On Monday, a learner calls out 12 times during a 20-minute observation.
On Tuesday, the learner calls out 18 times during a 60-minute observation.
Which day shows more calling out?
If you mean total recorded responses, Tuesday. If you mean responses per minute, Monday.
Both statements can be true, because they answer different questions. That difference is the heart of count versus rate—and it is why the word frequency needs more care than many study guides give it.
A count tells you how many responses were observed. A rate tells you how quickly they occurred relative to observed time.
The BCBA 6th Edition Test Content Outline places the central skill under C.3: Measure occurrence. The clinical selection problem also relates to C.9: Select a measurement procedure to obtain representative data that accounts for the critical dimension of the behavior and environmental constraints.
The exam can test whether you can calculate a rate, but harder questions ask which comparison is valid, whether observation opportunities are comparable, or whether an occurrence measure answers the clinical question at all.
Count is a tally of observed occurrences of a defined response. If a student raises a hand seven times, the count is seven. That number does not, by itself, specify the pace of responding.
Counts are valuable when the observation window is fixed or when the total number of events is itself important: for example, the number of medication administration errors during a defined shift, or the number of completed independent requests in a standardized practice period.
The catch is that twelve responses could represent twelve responses in ten minutes, twelve responses over a school day, or twelve responses across a week. Without the relevant time frame and opportunity information, those observations cannot be compared responsibly.
“Count has no time component” is too crude. A count is collected during some observation period; the point is that count itself does not normalize the number of responses by the length of that period. Record the observation window even when you report a raw tally.
Rate relates the number of recorded responses to the amount of time observed. The calculation is straightforward:
Response rate = number of responses ÷ observation time
For Monday's 12 call-outs in 20 minutes, 12 ÷ 20 = 0.60 responses per minute. For Tuesday's 18 call-outs in 60 minutes, 18 ÷ 60 = 0.30 responses per minute.
Expressed per hour, those same rates are 36 responses per hour and 18 responses per hour, respectively. These are equivalent rate expressions, not totals observed across an hour. Changing the time unit changes the numerical expression, not the observed pace. Always include the unit; “the rate was 0.6” is incomplete reporting.
Consider two sessions in which the client is expected to request help only when presented with difficult tasks. If one session includes 30 minutes of those tasks and another includes five, dividing by the entire session duration may hide important differences in opportunity.
For some targets, a response per opportunity measure or a proportion of opportunities with a response may answer the question better. That is not automatically the same measure as responses per minute. The appropriate denominator depends on the target and the question.
Similarly, if observers record only during selected periods, use the actual observation time to calculate the observed rate. Do not silently divide by the whole school day and present that value as though behavior had been monitored continuously.
Here is the terminology problem: frequency has been used to mean different things in behavior-analytic writing and instruction. Sometimes it means a simple count; elsewhere it means count per unit time and is treated as equivalent to rate.
Merbitz, Merbitz, and Pennypacker (2016) examined this terminology problem in On Terms: Frequency and Rate in Applied Behavior Analysis. They noted that BACB terminology and major textbooks have not used the word consistently, and argued that within applied behavior analysis frequency should refer to responses per unit of time rather than to a simple count. They also emphasized reporting the observation time base so that a measure can be interpreted.
Carr, Nosik, and Luke (2018) later examined how frequency is used in contemporary behavior-analytic journals and textbooks. They found that it was predominantly used to mean count, and recommended retaining that convention, with rate reserved for responses per unit of time.
This is a genuine disagreement about terminology, not about the mathematics, and it is not a trick invented for test prep. A count identifies how many responses occurred; a rate relates those responses to observation time. Both papers reinforce the importance of clear measurement reporting, even though their preferred terminology differs.
For the current BCBA 6th Edition TCO, the operative task wording is “Measure occurrence”, not a stand-alone definition of frequency. Therefore, do not invent an official rule that “frequency always equals count” or “frequency always equals rate” for every exam question.
The safest practice: identify what the question actually reports. “Nine responses” is a count. “Nine responses in three minutes” supplies both count and time, from which a rate of three per minute can be calculated. If a stem says “frequency,” look for operational definitions, units, observation duration, and intended interpretation.
Do not choose an answer based solely on the word frequency when another choice precisely identifies the measurement as a tally or as responses per minute. Context and units should resolve the ambiguity. If an assessment treats frequency as a synonym for count, that is a convention in that context—not a universal property of the term.
| Measure | Example | What it tells you | Key limitation |
|---|---|---|---|
| Count | 15 responses | Number observed | Does not normalize differing observation lengths |
| Rate | 0.5 responses/minute | Responses relative to observed time | Depends on a meaningful observation denominator |
| “Frequency” | “Frequency = 15” or “15/hour” | Depends on the author's usage | Ambiguous unless units and conventions are clear |
Suppose aggression is recorded 10 times in a 15-minute baseline observation and 12 times in a 40-minute intervention observation. The count increased from 10 to 12. But the observed rate changed from roughly 0.67 per minute to 0.30 per minute.
That difference is important. It also does not establish, by itself, that intervention caused the change. A change in observation context, opportunity, response definitions, or other conditions could affect the data. A defensible interpretation first describes the measured relation accurately; a causal conclusion requires appropriate evidence.
Normalize by time when time differs and pace is the question. Do not normalize automatically when another dimension matters more.
Rate is useful for discrete, countable responses where the observation period and opportunities make sense. But it is not the correct default for every behavior. Duration and latency, for example, fall under C.4: Measure temporal dimensions of behavior, and sometimes they answer the clinical question better than any occurrence measure.
These four questions require more than recognizing a definition. Decide what was observed, what denominator is justified, and what conclusion the resulting number actually supports.
During a 15-minute observation, a client makes 9 requests. During a later 30-minute observation, the client makes 12 requests. Both observations use the same response definition and comparable instructional conditions.
Which description of the observed request rate is accurate?
Why: Nine divided by fifteen is 0.60 requests per minute; twelve divided by thirty is 0.40 per minute. The later observation has a higher count but a lower time-adjusted rate.
Why the others miss: more total responses do not guarantee a faster rate; a threshold such as eight responses does not establish equal rates; reversing the values reverses the direction of change. The observation conditions were specified as comparable to keep the arithmetic interpretation meaningful.
A technician writes, “The learner emitted 24 hand raises during the observation.” No observation duration or standardized time unit appears in the report.
Which measurement label is most precise for the information actually reported?
Why: the report provides a tally. Without observation time, 24 cannot be converted into a rate per minute or hour.
Why the others miss: no time unit, total engagement duration, or event-to-response interval was reported. The ambiguous word frequency would add less precision than stating count.
A teacher asks students 10 individually scored questions in one lesson and 25 in another. The team wants to compare how consistently a student gives an independent correct answer when asked, not how rapidly the teacher presents questions.
Which measure best matches that question?
Why: the team wants success conditional on an opportunity. Dividing independent correct responses by the number of presented questions yields the proportion or percentage of opportunities with a correct response.
Why the others miss: raw totals are affected by how many questions were presented. A per-minute rate reflects both student responding and presentation pace. Interresponse time measures spacing between responses, not success across opportunities.
A clinician observes a behavior 16 times across 40 minutes during baseline and 10 times across 20 minutes during intervention. The behavior definition is unchanged, but observation contexts have not been demonstrated equivalent.
Which conclusion is best supported by these observations alone?
Why: baseline rate is sixteen divided by forty, or 0.40 per minute. Intervention rate is ten divided by twenty, or 0.50 per minute. It is reasonable to describe that observed increase; attributing it to intervention would require stronger evidence and attention to context.
Why the others miss: a drop in raw count obscures the unequal observation times; option C reverses the two rates and blames the shorter session; and option D overstates the problem, because unequal periods are exactly where rate allows a comparison, provided the result is interpreted cautiously.
A reliable reasoning sequence is to ask:
In part 1, we separated measurement quality from agreement statistics. In part 2, we separated three percentages that looked identical. Here, we separate two ways of describing the same recorded responses.
Those distinctions are connected: a number is only useful when you know what produced it, what it represents, and what it cannot establish.
Before interpreting “more” or “less” behavior, ask: more or less of what—total responses, responses per unit time, or responses per opportunity?
BA Prep practice questions are designed to make you defend the measurement decision, not simply remember a formula. If you can identify the right denominator before you calculate, you are doing the harder part of the problem.
Start Free TrialBehavior Analyst Certification Board. (2022). BCBA test content outline (6th ed.; updated September 2024). Tasks C.3 and C.9. Official TCO.
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. https://doi.org/10.1002/jaba.449.
Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.). Pearson.
Merbitz, C. T., Merbitz, N. H., & Pennypacker, H. S. (2016). On terms: Frequency and rate in applied behavior analysis. The Behavior Analyst, 39, 333–338. https://doi.org/10.1007/s40614-015-0048-z.