Wanted: PIR data
Partial interval recording (PIR) is one method for recording data during systematic direct observation of a behavior. While a convenient method, PIR has the key drawback that it systematically over-states the prevalence of the behavior under observation. When used in single-case research to measure changes in behavior resulting from intervention, the systematic bias in PIR data can lead to deceptive results, such as inferring that an intervention reduces the prevalence of a problem behavior when in fact the opposite is true.
With my student Daniel Swan, I am currently working on developing methods for analyzing partial interval recording data that take its systematic bias into account. Some of these methods can be used with session-level summary PIR measurements (i.e., the percentage of intervals with the behavior), which are easily extracted from published single-case graphs. See here for the paper describing these methods.
We are now turning our attention to methods that use the finer-grained, interval-by-interval PIR data to obtain better estimates of the prevalence and incidence (frequency per unit time) of the behavior. For instance, if the observer uses 15 s partial interval recording, with 5 s for recording, for a 20 min session, this is a total of 60 intervals, for each of which the presence or absence of the behavior is recorded. The methods we’re working on make use of the full set of 60 ordered data points from the session. The general idea our work is similar to the post-hoc correction techniques proposed by Suen & Ary (1986), but we think we can greatly improve on their proposal.
To fully validate the methods we are developing, we need to test them out on real-world data. If you, dear reader, have access to PIR data and would be willing to share it with us, I would love to hear from you. We are looking specifically for:
- Fine-grained (interval-by-interval) PIR data collected in real research contexts, such as single-case studies or observational studies involving students with behavioral disorders, children with autism-spectrum disorders, etc.
- Alternately, continuously-recorded behavioral observation data (e.g., as collected through MOOSES, the Direct Assessment Tracking Application, or ProCoderDV) that we could then convert into PIR data.
- Along with either type of behavioral observation data, a brief (or lengthier) description of the participant(s) whose behavior was measured and the context in which the measurements were collected.
We can work with data in whatever format you might be willing to provide–whether that means photo-copied, paper observation forms, an Excel workbook, or a bunch of ProCoderDV data files. In return for sharing data, we will share with you the examples that we develop based on the data, which could also provide a basis for further collaboration. If you are interested in seeing your data analyzed and helping to advance this methodological work, please contact me.
- Alternating renewal process models for behavioral observation: Simulation methods and validity implications
- ARPobservation now on CRAN
- New article: Measurement-comparable effect sizes for single-case studies of free-operant behavior
- To what extent does partial interval recording over-estimate prevalence?
- ARPobservation: Basic use