When to use it
Use duration recording when how long a behavior lasts matters more than how often it happens. Examples:
- Reduction targets such as tantrums, crying, or time out of seat.
- Increase targets such as time engaged in a task, time playing with peers, or time tolerating a haircut.
Duration needs a clear onset and offset. If observers can’t agree when an episode starts or stops, fix the definition before collecting data.
Two duration measures are common. Record episode start and stop times, and you can calculate both:
- Duration per occurrence: how long each episode lasts. Use it when the length of each episode matters, such as “tantrums are getting shorter.”
- Total duration: the sum of all episodes in a session. Use it when overall time matters, such as “total minutes engaged.”
How to fill it in
- Write the operational definition with an explicit onset and offset. Example: “Tantrum: crying with screaming or dropping to the floor. Onset: first scream or drop. Offset: 10 consecutive seconds with no crying, screaming, or dropping.”
- Record the session start and stop times at the top of the sheet. You need these to calculate percent of session.
- Start the timer at onset. Write the clock time in the Start column.
- Stop the timer at offset. Write the clock time in the Stop column. The offset includes the full “calm” window in your definition, but the timed duration ends when the behavior last occurred. Decide on one convention and write it down, because practice varies.
- Write the episode duration in minutes and seconds (for example, 2:15).
- Add notes on context, such as the activity, what came before, and any safety procedures used.
Common mistakes:
- Guessing duration afterward instead of timing it live.
- Not agreeing on an offset rule, so one observer records 2 minutes and another records 6.
- Forgetting to record session length, which makes percent of session impossible to calculate.
- Mixing “total duration” and “average duration” across sessions on the same graph.
The sheet
Client:Date:Observer:Setting:
Target behavior and definition (onset / offset):
Session start:Session stop:Session length (min):
| Episode # | Start time | Stop time | Duration (min:sec) | Activity / context | Notes | Initials |
|---|---|---|---|---|---|---|
| 1 | ||||||
| 2 | ||||||
| 3 | ||||||
| 4 | ||||||
| 5 | ||||||
| 6 | ||||||
| 7 | ||||||
| 8 | ||||||
| 9 | ||||||
| 10 | ||||||
| 11 | ||||||
| 12 |
Number of episodes:Total duration:Mean duration per episode:Percent of session:
Summarizing the data
- Total duration = sum of all episode durations.
- Mean duration per occurrence = total duration divided by number of episodes.
- Percent of session = total duration divided by session length, times 100.
Worked example: a 30-minute session with four tantrums lasting 0:45, 2:10, 0:30, and 1:35.
- Convert to seconds: 45 + 130 + 30 + 95 = 300 seconds = 5 minutes total duration
- Mean duration per episode = 300 / 4 = 75 seconds (1:15)
- Percent of session = 5 / 30 x 100 = 16.7%
If you also want frequency, the number of episodes (4) is already on the sheet. Rate = 4 per 30 minutes = 8 per hour.
Tips
- A stopwatch that keeps running totals (start and stop without resetting) makes total duration easy. Still write each episode separately so you can calculate per-occurrence data.
- For increase targets, such as engagement, graph total duration or percent of session. For “tolerating” targets, graph duration per trial.
- For IOA, total duration IOA (shorter total divided by longer total, times 100) is common. Mean duration-per-occurrence IOA is more sensitive to disagreement on single episodes.
- If episodes are so frequent that timing each one is impractical, consider partial or whole interval recording as an estimate. Remember that interval data only estimates duration.