Frequency vs rate in ABA
Frequency is how many times a behavior happened, and rate is that count divided by the time you observed. Use rate whenever observation periods differ in length, because raw counts from a 20-minute session and a 50-minute session cannot be compared fairly.
What is frequency in ABA?
Frequency is a plain count of how many times a behavior occurs during an observation. You make a tally mark or click a counter each time the behavior starts and stops as a complete instance. It works best for behaviors with a clear beginning and end, such as requests, hits, or correct answers in a drill.
A count alone is only useful when every observation is the same length. Five requests in one 15-minute session and five in another 15-minute session mean the same thing.
What is rate in ABA?
Rate is a count divided by the time spent observing, written as responses per minute or per hour. It keeps the count tied to its observation window, so sessions of different lengths can be compared on one graph. To calculate it, divide the number of responses by the number of minutes observed.
Example: 24 vocal requests in 40 minutes is 24 divided by 40, or 0.6 requests per minute. In 8 minutes the same learner made 6 requests, which is 0.75 per minute. The second count is far smaller, yet the rate is higher.
What is the one-question test for frequency vs rate?
Ask whether the observation time is part of the measure. If the result is a number of times, with no time unit attached, it is frequency. If the result includes per minute, per hour or per session of a stated length, it is rate. When sessions vary in length, rate is the measure that fits.
- Check whether every session is the same length.
- If yes, a count can work, but rate still does no harm.
- If no, convert each count to a rate before comparing.
- Use the same time unit on every data point.
Frequency and rate side by side
| Frequency | Rate | |
|---|---|---|
| What it records | Number of occurrences | Occurrences per unit of time |
| Time included | No | Yes |
| Comparing uneven sessions | Misleading | Fair |
| Example | 9 elopements | 9 elopements in 3 hours, or 3 per hour |
| Best fit | Fixed-length sessions | Sessions of different lengths |
How do you choose between frequency, rate and other measures?
Match the measure to the question the supervisor is asking. How many times points to count or rate. How long points to duration. How long before starting points to latency. How much time between responses points to interresponse time. Each answers a different question, so one behavior may need more than one measure.
Scenario 1. A supervisor wants to know whether a learner is asking for help more often, and sessions run 30, 45 or 60 minutes. Lengths differ, so collect a count and convert it to rate.
Scenario 2. A learner cries for long stretches, and the concern is how long each episode lasts. Counting episodes would hide the real problem. This calls for duration.
Scenario 3. After an instruction to line up, a learner takes a long time to begin moving. The time from instruction to the first step is latency, not rate.
Scenario 4. A learner bites his nails, and the team wants to know how much time passes between one episode and the next. That gap is interresponse time, or IRT.
Scenario 5. Every session is exactly 20 minutes, and the technician records 7 tantrums. A count of 7 is fine here, though converting it to 0.35 per minute keeps the graph ready if session lengths change.
Which measures are related to frequency and rate?
Frequency and rate sit in the continuous measurement family with duration, latency and interresponse time. All of them record every instance during the observation. Interval recording methods such as partial interval, whole interval and momentary time sampling only estimate behavior, so they are discontinuous and answer a different question.
| Measure | Question it answers |
|---|---|
| Duration | How long did each instance last? |
| Latency | How long after the cue did it start? |
| Interresponse time | How long between one instance and the next? |
What are common mistakes with frequency and rate?
The biggest mistake is comparing raw counts across sessions of different lengths and reporting a false improvement or decline. Others include dividing by the wrong time, mixing minutes and hours on one graph, counting long behaviors that really need duration, and assuming every source uses the word frequency in exactly the same way.
- Forgetting to write down the observation length.
- Leaving out time spent on breaks when the plan says to subtract it, or the reverse.
- Graphing counts and rates on the same axis.
- Counting a behavior with no clear start or end.
Practice questions
During a 30-minute session a technician tallies 12 instances of hand-raising. In a 60-minute session the next day she tallies 18. Which measure lets her compare the two sessions fairly?
A supervisor wants to know how long a learner's tantrums last from start to finish. Which measure should the technician collect?
On a line graph of correct responses per session, the data points rise steadily across the last six sessions after a new teaching procedure started. How is this best described?
Questions people ask
Do some sources use frequency to mean rate?
Yes. Some texts treat the terms as close to the same. On a question, check whether time is part of the measure.
Is rate a continuous measure?
Yes. Rate records every instance across the whole observation, then adds the time.
What unit should I use for rate?
Use the unit your supervisor sets, usually per minute or per hour, and keep it the same on every data point.
How ready are you?
Ten questions, about 3 minutes, no account. Get your readiness score and your weakest domains.
Check my readiness