Timedora

2 Duration and Latency Measurement

Learn how to define, record, summarize, and interpret duration and latency measurements of behavior.

What and measure

and are time-based measures, but they answer different questions. measures how long a behavior lasts; measures how long it takes for a behavior to begin after a specified event. Both require clear, consistent rules for when timing starts and stops.

Measuring

is the elapsed time from behavior to behavior . Use it when the length of a behavior matters, such as how long a student works, cries, or remains out of their seat.

Before measuring, establish an with observable start and end points. For example, work might begin when a student makes a task-related mark or selects a task item, and end when the student stops task-related activity for 10 consecutive seconds. This 10-second rule is part of that example definition, not a universal standard. Apply the chosen rule consistently.

To record , start the timer when the defined behavior begins and stop it when the behavior ends. Record each episode in a consistent unit, such as seconds, repeat for each episode during the observation, and add episode times to calculate total .

Measuring

is the elapsed time between a defined and the beginning of the specified behavior. For example, it can measure the time from a teacher's direction to begin work until the student's first task-related action.

Define the timing points in advance. A practical rule is to start timing at the end of the direction and stop at the first observable action that meets the behavior definition. If a direction takes time, consistently use the same timing point, such as its end. Record each opportunity separately.

A shorter means the behavior began sooner after the ; a longer means more time elapsed before it began. Interpret the measurement in light of the specific behavior, direction, and setting.

Summarizing and interpreting results

Use summaries that match the question being asked. Total combines all episodes in an observation, while mean per episode describes the average length of an episode. Mean describes the average elapsed time across measured opportunities.

Total duration=∑episode durations\text{Total duration} = \sum \text{episode durations}
Mean duration per episode=total durationnumber of episodes\text{Mean duration per episode} = \frac{\text{total duration}}{\text{number of episodes}}
Mean duration per observation=total duration across observationsnumber of observations\text{Mean duration per observation} = \frac{\text{total duration across observations}}{\text{number of observations}}

Mean per observation is useful when comparing observations of the same length. Label what was averaged and include the unit, such as “72.5 seconds per episode” or “10 seconds per direction.”

Mean latency=sum of recorded latenciesnumber of measured opportunities\text{Mean latency} = \frac{\text{sum of recorded latencies}}{\text{number of measured opportunities}}
Range=largest value−smallest value\text{Range} = \text{largest value} - \text{smallest value}

Range is a simple indication of variation. It does not show how values are distributed between the minimum and maximum. When comparing observations, check that observation lengths, behavior definitions, and timing rules are comparable.

Worked examples and recording limits

A behavior occurs from 9:02:10 to 9:03:25, then from 9:05:00 to 9:06:10. The episodes last 75 seconds and 70 seconds.

Total duration=75+70=145 seconds\text{Total duration} = 75 + 70 = 145\text{ seconds}
Mean duration per episode=1452=72.5 seconds\text{Mean duration per episode} = \frac{145}{2} = 72.5\text{ seconds}

The total is 2 minutes and 25 seconds. If the observation lasted 10 minutes, or 600 seconds, the behavior occupied:

145600×100≈24.2% of the observation\frac{145}{600} \times 100 \approx 24.2\%\text{ of the observation}

Use this percentage only when the observation period and the behavior's timing are defined consistently. Total and mean per episode answer different questions, so label which measure is reported.

After four directions, recorded latencies are 8, 12, 5, and 15 seconds:

Total latency=8+12+5+15=40 seconds\text{Total latency} = 8 + 12 + 5 + 15 = 40\text{ seconds}
Mean latency=404=10 seconds per direction\text{Mean latency} = \frac{40}{4} = 10\text{ seconds per direction}
Range=15−5=10 seconds\text{Range} = 15 - 5 = 10\text{ seconds}

If a behavior does not occur during an observation, record zero . If it does not begin before the observation ends, report that no response occurred within the available time—for example, “no response within 30 seconds”—rather than treating the cutoff as an exact .

Practice results and measurement consistency

For two episodes, a behavior starts at 10:14:20 and ends at 10:15:05, then starts at 10:18:00 and ends at 10:18:40. The episodes last 45 and 40 seconds. Total is 45+40=8545 + 40 = 85 seconds, mean is 852=42.5\frac{85}{2} = 42.5 seconds per episode, and range is 45−40=545 - 40 = 5 seconds.

After three directions, latencies are 6, 11, and 7 seconds. Mean is 6+11+73=8\frac{6 + 11 + 7}{3} = 8 seconds per direction, and range is 11−6=511 - 6 = 5 seconds. The mean represents the average elapsed time from the defined direction point to the beginning of the defined behavior across the three opportunities.

Reliable interpretation depends on clear operational definitions, consistent timing points and units, and comparable observation conditions. Report the measured quantity and its unit so that the meaning of each total or average is clear.