Behavioral observation

Easily simulate thousands of single-case designs

Earlier this month, I taught at the Summer Research Training Institute on Single-Case Intervention Design and Analysis workshop, sponsored by the Institute of Education Sciences’ National Center for Special Education Research. While I was there, I shared a web-app for simulating data from a single-case design. This is a tool that I put together a couple of years ago as part of my ARPobservation R package, but haven’t ever really publicized or done anything formal with.

ARPobservation

Simulate systematic direct observation data.

New article: Four methods for analyzing PIR data

My article with Daniel Swan, “Four methods for analyzing partial interval recording data, with application to single-case research” has been accepted for publication in Multivariate Behavioral Research. In an extension of my earlier paper on measurement-comparable effect sizes for single-case studies, this article provides some approaches to estimating effect sizes from single-case studies that use partial interval or whole interval recording to measure behavioral outcomes. The full abstract is below. Preprint and supporting materials are available.

New article: Alternating renewal process models for behavioral observation

My article with Chris Runyon, titled “Alternating renewal process models for behavioral observation: Simulation methods, software , and validity illustrations” has been published in Behavioral Disorders. The abstract is below. Postprint available here. All of the examples in the paper are available in the R package ARPobservation. Direct observation recording procedures produce reductive summary measurements of an underlying stream of behavior. Previous methodological studies of these recording procedures have employed simulation methods for generating random behavior streams, many of which amount to special cases of a statistical model known as the alternating renewal process.

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.

ARPobservation now on CRAN

Version 1.0 of the ARPobservation package is now available on the Comprehensive R Archive Network. This makes it even easier to install. Here’s the package description: ARPobservation: Tools for simulating different methods of observing behavior based on alternating renewal processes ARPobservation provides a set of tools for simulating data based on direct observation of behavior. It works by first simulating a behavior stream based on an alternating renewal process, given specified distributions of event durations and interim times.

New article: Measurement-comparable effect sizes for single-case studies of free-operant behavior

My article “Measurement-comparable effect sizes for single-case studies of free-operant behavior” has been accepted at Psychological Methods. Postprint and supporting materials are available. Here’s the abstract: Single-case research comprises a set of designs and methods for evaluating the effects of interventions, practices, or programs on individual cases, through comparison of outcomes measured at different points in time. Although there has long been interest in meta-analytic technique for synthesizing single-case research, there has been little scrutiny of whether proposed effect sizes remain on a directly comparable metric when outcomes are measured using different operational procedures.

To what extent does partial interval recording over-estimate prevalence?

It is well known that the partial interval recording procedure produces an over-estimate of the prevalence of a behavior. Here I will demonstrate how to use the ARPobservation package to study the extent of this bias. First though, I’ll need to define the terms prevalence and incidence and also take a detour through continuous duration recording. Prevalence and incidence First off, what do I mean by prevalence? In an alternating renewal process, prevalence is the long-run proportion of time that the behavior occurs.

ARPobservation: Basic use

The ARPobservation package provides a set of tools for simulating data generated by different procedures for direct observation of behavior. This is accomplished in two steps. The first step is to simulate a “behavior stream” itself, which is assumed to follow some type of alternating renewal process. The second step is to apply a procedure or “filter,” which turns the simulated behavior stream into the data recorded by a given observation procedure.

Getting started with ARPobservation

UPDATED 5/29/2014 after posting the package to CRAN Here are step-by-step instructions on how to download and install ARPobservation. For the time being, ARPobservation is available as a pre-compiled binary for Windows. For Mac/Linux, you’ll have to download the source from Github. Download and install R. R is free, open-source software that is used by many data analysts and statisticians. ARPobservation is a contributed package that runs within R, so you’ll need to get the base software first.