Representativeness of direct observations selected using a work-sampling equation.
Work-sampling equations say you need near-continuous checks to catch brief behaviors, so real-world momentary time sampling will usually miss them.
01Research in Context
What this study did
Boudreau et al. (2015) asked a simple question: how many short behavior checks do you need to catch the real picture?
They used work-sampling math to pick observation moments for kids with autism and intellectual disability.
Then they compared those short samples to the full, continuous record to see what got missed.
What they found
The equation only gave an honest picture when staff watched almost non-stop.
At rates you can actually fit into a busy day, quick, low-duration behaviors slipped through the cracks.
In short, practical momentary time sampling under-counted brief acts even when the math said it should work.
How this fits with other research
Decasper et al. (1977) already warned that any interval method can mislead; Boudreau et al. (2015) now show the math behind that warning.
Iwata et al. (1990) found 15-second momentary samples worked fine in classrooms; the newer study says that win disappears when behaviors are short and rare.
Gardenier et al. (2004) told us to drop partial-interval recording for stereotypy; Boudreau et al. (2015) add that even momentary sampling needs far more checks than we thought.
Why it matters
If your treatment data feel off, the fault may be the tool, not the kid. Brief stims, rapid requests, or quick escapes can vanish from momentary samples unless you watch almost continuously. Use the equation numbers as a red flag, not a green light, and tighten your window or add continuous spot-checks when low-duration behaviors matter for decisions.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Cut your momentary sample window in half or run a five-minute continuous probe to double-check low-duration targets.
02At a glance
03Original abstract
Deciding on appropriate sampling to obtain representative samples of behavior is important but not straightforward, because the relative duration of the target behavior may affect its observation in a given sampling interval. Work-sampling methods, which offer a way to adjust the frequency of sampling according to a priori or ongoing estimates of the behavior to achieve a preselected level of representativeness, may provide a solution. Full-week observations of 7 behaviors were conducted for 3 students with autism spectrum disorder and intellectual disabilities. Work-sampling methods were used to select momentary time samples from the full time-of-interest, which produced representative samples. However, work sampling required impractically high numbers of time samples to obtain representative samples. More practical momentary time samples produced less representative samples, particularly for low-duration behaviors. The utility and limits of work-sampling methods for applied behavior analysis are discussed.
Journal of applied behavior analysis, 2015 · doi:10.1002/jaba.193