3 Interval Measurement and Time Sampling

Purpose and method choices

Interval measurement divides an observation into equal time periods and records whether a defined behavior meets a specified rule in each period. It can be useful when behavior is frequent, lasts for varying amounts of time, or is difficult to count continuously.

The three common methods are whole-interval recording, partial-interval recording, and momentary time sampling (MTS). Because they use different scoring rules, their percentages are not interchangeable.

How the scoring rules differ

Each method applies a different rule to the same interval:

• Whole-interval recording: Score an interval positive only if the behavior occurs throughout the entire interval. This method is often used for behaviors desired to continue, such as sustained engagement. It tends to underestimate behavior because any interruption makes the interval negative.

• Partial-interval recording: Score an interval positive if the behavior occurs at any time during it, even briefly. This can be useful for detecting behaviors one wants to reduce. It tends to overestimate how much time the behavior occurred. Multiple instances within one interval still count as only one positive interval.

• Momentary time sampling (MTS): Score whether the behavior is occurring at one specified instant, commonly at the end of each interval. Observation is needed only at those moments, but brief behavior between them may be missed. The estimate can be high or low depending on the behavior and sampling schedule.

Scoring an observation

Consider an observation lasting 80 seconds, divided into eight 10-second intervals. The behavior occurs during the first 2 seconds of interval 1; throughout interval 2; from second 4 through second 7 of interval 3; not at all in interval 4; during the final 2 seconds of interval 5; throughout interval 6; during only the first second of interval 7; and not at all in interval 8. For MTS, assume the observer checks at the end of each interval.

A plus means the method's rule is met; a minus means it is not.

• Interval 1: whole interval −; partial interval +; MTS −.

• Interval 2: whole interval +; partial interval +; MTS +.

• Interval 3: whole interval −; partial interval +; MTS −.

• Interval 4: whole interval −; partial interval −; MTS −.

• Interval 5: whole interval −; partial interval +; MTS +.

• Interval 6: whole interval +; partial interval +; MTS +.

• Interval 7: whole interval −; partial interval +; MTS −.

• Interval 8: whole interval −; partial interval −; MTS −.

The positive-interval totals are 2 for whole-interval recording, 6 for partial-interval recording, and 3 for MTS. The Percentage of intervals positive is calculated as follows:

For this example, whole-interval recording gives $\frac{2}{8} \times 100 = 25\%$, partial-interval recording gives $\frac{6}{8} \times 100 = 75\%$, and MTS gives $\frac{3}{8} \times 100 = 37.5\%$.

Interpreting the percentages

These percentages summarize the proportion of intervals scored positive under each method. They do not mean that the behavior occupied exactly $25\%$, $75\%$, or $37.5\%$ of the observation time. In particular, whole-interval and partial-interval percentages should not be treated as precise measures of duration.

The scoring rule shapes the result: partial-interval recording is sensitive to brief occurrences, whole-interval recording requires sustained behavior, and MTS captures behavior only at scheduled instants. As the example shows, the same observation can produce very different percentages. A change in percentage across sessions may reflect a change in behavior, but it may also be influenced by the recording method, interval length, observation period, or consistency of scoring.

Planning reliable observations

Before collecting data, define the behavior clearly, specify the interval length, and state the exact moment for MTS observations. Use the same procedure across observations so that the method and its schedule are consistent.

Shorter intervals generally provide finer temporal detail, though the best practical choice depends on the behavior and the observer's ability to score reliably. If the goal is to know the exact number of occurrences or the total time spent in a behavior, use a suitable continuous measure, such as frequency or duration recording, when feasible. Interval sampling provides an estimate rather than a complete record.