Choosing a behavior data method

By the IEP Atlas editorial teamLast updated November 1, 2026

Short answer

Pick the behavior data method from the behavior's shape. Count behaviors with a clear start and end that happen in short bursts (frequency). Time behaviors that vary in length (duration). Time how long a student takes to start (latency). Use intervals for high-rate or continuous behavior. Use ABC data only while you are looking for the function.

Most behavior data problems are method problems. A teacher counts tantrums that last anywhere from 30 seconds to 25 minutes, and the count drops while the class loses more time than ever. Or a para tries to tally every hum for a student who hums for half the day. The data are honest, but they answer the wrong question.

This page matches 12 common classroom behaviors to the method that fits. The students and numbers are fictional. For the general rules on choosing any IEP data method, see how to choose an IEP data collection method. For where behavior data feed the FBA and the plan, see the FBA and BIP guide.

The decision table

Start with the question the team needs answered.

Question	Method	Fits behaviors that	Data sheet
How often?	Frequency recording	Have a clear start and end, last a few seconds	Frequency sheet
How long?	Duration recording	Vary in length from episode to episode	Duration sheet
How fast does it start?	Latency recording	Follow a cue, such as a direction	Latency sheet
How much of the time?	Interval recording	Are high rate, continuous, or hard to count	Interval sheet
Why?	ABC data	Any behavior during an FBA	ABC narrative or checklist

If two methods seem to fit, pick the one the goal will use. The baseline and the goal must share a unit.

12 behaviors and the method that fits

Each example names the behavior, the method, and what makes the choice strong.

Behaviors with a clear start and end

1. Calling out during instruction. Malik calls out about 15 times in a 45-minute lesson. Method: frequency, converted to rate per lesson. Why it is strong: each call-out is brief and countable, and the rate shows change when lesson length varies.

2. Hitting peers. Sofia hits about 3 times a day. Method: frequency per day. Why it is strong: low-rate, serious behavior should be counted every time. Sampling would miss it.

3. Leaving the classroom. Jonah leaves about twice per math class. Method: frequency plus duration of each exit. Why it is strong: the count shows how often, and the minutes out show the instruction lost.

4. Swearing. Grace swears about 6 times a day in two classes. Method: frequency during those two classes only. Why it is strong: the count stays manageable because it is limited to the hot-spot periods.

Behaviors that vary in length

5. Tantrums. Leo's tantrums last 1 to 22 minutes. Method: duration per episode, plus total minutes per day. Why it is strong: a drop in count could hide a rise in total minutes; duration catches it.

6. Out of seat during work time. Ben stands and wanders for stretches of 10 seconds to 6 minutes. Method: total duration per 30-minute work block. Why it is strong: minutes in seat are what the teacher cares about.

7. Crying. Ava cries for 2 to 15 minutes at drop-off. Method: duration from arrival. Why it is strong: the goal is shorter episodes, and duration measures exactly that.

Behaviors that follow a cue

8. Starting work. Eli takes a median of 6 minutes to begin after directions. Method: latency in seconds or minutes. Why it is strong: the problem is the delay, not whether he ever starts.

9. Following a transition direction. Kai takes 2 to 4 minutes to leave the computer. Method: latency from the direction to standing up. Why it is strong: it matches a goal like "within 60 seconds."

High-rate or continuous behaviors

10. Humming. Owen hums for long stretches with short breaks. Method: momentary time sampling every 2 minutes. Why it is strong: counting each hum is not possible, and a snapshot every 2 minutes is.

11. On-task behavior. Rosa is on task for some of each work period. Method: momentary time sampling every 30 seconds for 10 minutes. Why it is strong: it estimates the share of time on task without watching every second. The guide on partial vs whole interval recording explains which interval type fits.

While looking for the function

12. Pushing materials off the desk. Carlos does this during some tasks. Method: ABC data for 2 weeks, plus a frequency count. Why it is strong: the ABC entries show the trigger and the payoff, and the count sets the baseline. The steps are in how to take and analyze ABC data.

Weak and strong choices side by side

Behavior	Weak choice	Why it fails	Strong choice
Tantrums of 1 to 22 minutes	Count per day	Count fell from 4 to 2 while minutes rose from 30 to 45	Total minutes per day
Humming most of the period	Tally every hum	Para lost count after 5 minutes	Momentary time sampling every 2 minutes
Hitting 3 times a day	15-minute interval sampling	Missed 4 of 11 hits in one week	Frequency every time
Slow to start work	"On task yes/no" at the end of class	Does not show the 6-minute delay	Latency from direction to first mark on paper
Calling out in lessons of 20 to 50 minutes	Raw count per lesson	Longer lessons look worse	Rate per 10 minutes
Refusal	"Good day / bad day" rating	No definition, no number	Frequency of refusals after a direction, with an operational definition

The operational definitions of behavior guide shows how to define each behavior first. No method rescues a vague definition.

Measurement section: frequency recording, filled in

Frequency is the method most behavior goals use, so here it is in detail.

Set it up. Write the operational definition at the top. Set the observation window, such as "math block, 9:45 to 10:45." Record the start and end time each day, so you can convert to a rate if the window changes.

Probe frequency. Count every day during the observation window while the plan is new. Once the trend is clear, many teams drop to 3 days a week. Keep the same days and times so the data stay comparable. Graph weekly.

A filled-in frequency sheet. Malik, grade 4, call-outs during math (fictional).

Date	Window	Minutes	Tally	Count	Rate per 10 min
Mon 10/6	9:45 to 10:45	60	16 marks	16	2.7
Tue 10/7	9:45 to 10:30	45	13 marks	13	2.9
Wed 10/8	9:45 to 10:45	60	15 marks	15	2.5
Thu 10/9	9:45 to 10:45	60	14 marks	14	2.3
Fri 10/10	9:45 to 10:15	30	8 marks	8	2.7

Baseline: median 2.7 call-outs per 10 minutes (range 2.3 to 2.9). The raw counts range from 8 to 16 only because the windows differ, which is why the rate is the number to report. A goal built from it might read: "During math instruction, Malik will raise his hand and wait to be called on, reducing call-outs to 0.5 or fewer per 10 minutes across 4 consecutive weeks, measured by daily frequency recording." The frequency data sheet has the columns above, ready to print.

For how often to graph and review behavior data once the goal is in place, see progress monitoring for behavior goals.

Who collects the data, and when

A method only works if the person collecting it can do it while working. A co-teacher leading a lesson can tally call-outs with a counter clipped to a lanyard. The same teacher cannot time 6 overlapping tantrums. Assign the method to the adult whose hands are free during that block, often the para; see paraprofessional data collection for training tips.

Check agreement now and then. Have a second adult collect data on the same session once every 2 or 3 weeks. If the counts differ by more than 1 or 2, tighten the definition. The guide on interobserver agreement shows the math.

The IEP must state how progress toward each goal will be measured.^[1] The What Works Clearinghouse practice guide on classroom behavior recommends collecting data on the specific behavior to guide decisions.^[2] To check that your goal names the same measure as your data sheet, paste it into the goal checker.

Frequently asked questions

What is the most common behavior data method?

Frequency recording, because most target behaviors have a clear start and end and happen in short bursts. It stops working when episodes vary a lot in length or happen too fast to count. Then switch to duration or interval recording.

Can I use two data methods for one behavior?

Yes. Frequency plus duration is common for tantrums or classroom exits. Keep it to two at most, and decide which one the IEP goal uses. The second measure supports decisions but should not crowd out the first.

Is ABC data enough for a behavior goal baseline?

Usually not. ABC data show why a behavior happens, but the entries are rarely a complete count. Pair them with frequency or duration data taken the same way the goal will be measured.

Use it in IEP Atlas

Sources

34 CFR §300.320: Definition of individualized education program, Legal Information Institute, Cornell Law School
Reducing Behavior Problems in the Elementary School Classroom, What Works Clearinghouse, Institute of Education Sciences

Written and checked by the IEP Atlas editorial team under our editorial policy. This guide describes general practice and federal rules. It is not legal advice; your state may add requirements, and the IEP team makes the decisions for each student.