Behavior Bubbles

Further to my previous post on continuous measurement (i.e. frequency, duration, latency, inter response time, etc.) which can be read here, I shall dive into discontinuous measurement today!

Discontinuous measurement is defined as ‘measurement conducted in a manner such that some instances of the response class(es) of interest may not be detected’ (Cooper, Heron, & Heward, 2007). The length of the observation period is first decided before it is broken down in smaller equal intervals. Some examples are as follows: 1) a 1-hour session may be broken down into 3 minute intervals; 2) a 10 minute observation may be broken down into 1 minute intervals; and 3) a 5 minute observation may be broken down into 30 second intervals. This data is normally collected as a percentage the behavior occured (i.e. should the behavior be observed in 5 out of 10 of the intervals, the behavior occured 50% of the time).

Let us look at some types of discontinuous measurement and see some examples of further illustrate the concepts below.
_______________________________________________________

Whole Interval Recording
Whole interval recording is normally used to measure continuous behaviors which occur throughout the assigned interval. This can be used when behaviours occur so frequently that it is difficult to tell one instance from another instance (i.e. verbal stims, flapping hands, etc.).

In order for the behavior to be recorded as a positive (i.e. having occured), the behavior would have to be observed throughout the entire interval. Should the behavior only occur part of the interval, this behavior would be recorded as a negative (i.e. not occured).

As such, utilising this method of recording typically underestimates the overall occurrence of the behavior. Due to the nature of this recording method which underestimates the overall occurrence of behavior, whole interval recording should be utilised for behaviors you would like to increase (i.e. completing on a worksheet).

In this example shown in Figure 1, the behavior occured 16.7% of the time (i.e. 1/6 x 100 = 16.67%).

Partial Interval Recording
Partial interval recording looks at recording a positive (i.e behavior occured) should the behavior occur at any point during the interval. This is regardless of how any times the behavior was noted to have occured. A negative (i.e. not occured) would only be scored if the behavior was not observed throughout the interval. Due to the nature of this recording method, the occurrence of the behavior tends to be overestimated. As such, partial interval recording should be utilised more for behaviors you would like to decrease.

In this example shown in Figure 2, the behavior occured 66.7% of the time (i.e. 4/6 x 100 = 66.67%).

Momentary Time Sampling
Momentary time sampling looks at recording a positive (i.e. behavior occured) should the behavior be observed to occur at the end of the interval. A negative (i.e. not occured) would be scored even if the behavior was observed during the interval as long as it stopped before the interval ended. Momentary time sampling both over- and under-estimates behaviors. However, this method of data collection is useful as the observer would not have to constantly observe the client throughout the session.

In this example shown in Figure 3, the behavior occured 33.3% of the time (i.e. 2/6 x 100 = 33.33%).

_________________________________________________________
Hope the diagrams helped to make things much clearer to aid your understanding! Feel free to comment below to state other topics you would be interested in learning more about or brainstorm how you can utilise these measurement techniques!

References
Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis (2nd ed.). Upper Saddle River, NJ: Pearson.