A brief report on a comparison of time-sampling procedures.
Check your time-sampling plan against continuous data before you trust the numbers.
01Research in Context
What this study did
The author lined up different time-sampling plans head-to-head. Each plan watched the same short clips of a teacher’s behavior.
Goal: see which quick, intermittent method best matched the gold standard of continuous recording.
What they found
Some sampling plans tracked the true picture. Others missed large chunks of behavior.
The paper warns: pick a plan, then test it. Don’t trust any pattern until you check its accuracy.
How this fits with other research
Skrtic et al. (1982) extends this idea into busy classrooms. They built a one-minute sheet that lets you watch 18 behaviors across several kids at once while still hitting over 90% agreement.
Field et al. (2001) pushes further. They show that real-time, moment-by-moment recording clarifies fast changes within a session—something any interval plan can blur.
Lloyd et al. (2018) echo the same tune decades later. They compared ways to measure contingency and again found event-based beats interval-based unless you tune the interval method to the schedule.
Why it matters
You probably can’t watch every second. Before you settle on 10-s partial, 15-s momentary, or any other quick plan, run a short accuracy check against continuous data. Five minutes of double-coding can save you from weeks of misleading graphs.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Code one session twice: once with your usual interval sheet and once continuously, then compare.
02At a glance
03Original abstract
Much behavioral research has been based on an observation method in which an observer continuously time-sampled the occurrence of certain responses, typ- ically making a notation of occurrence or nonoccur- rence every 10 sec. Research and training settings often had the resources to allow such recording for every subject under study. Applied settings, by con- trast, often could not afford an observer for every subject under study, and at present, even research and training settings find their funding inadequate for continuous time-sampling. Thus, intermittent timesampling becomes necessary, and the question arises: how accurate is recording in each of the various possi- ble patterns of intermittency? This brief report uses the teacher-training model to examine some quantita- tive comparisons.
Journal of applied behavior analysis, 1974 · doi:10.1901/jaba.1974.7-623