Association for Behavior Analysis International

One of the challenges faced by parents is objectively measuring treatment outcome. Currently, measurement methods are too cumbersome or rely upon subjective information. The current investigation aims to use a simplified measurement system to assess treatment outcome for problem behaviors exhibited by individuals with DD. Direct Behavior Rating (DBR) is an evaluative rating completed by caregivers at the time the behaviors occur and includes multiple, continuous observations over a period of time that yields a repeated, objective, and quantitative measure of behavior. The specific aim of this project is to demonstrate that DBR is maximally sensitive to detect behavioral changes. The participants were four female college students. Each participant viewed 12, 5-min videos of a child with autism under baseline and treatment conditions. After viewing each video, the participant used DBR to rate the severity of the problem behaviors. Our primary analysis compared DBR to systematic direction observation (SDO) across baseline and treatment phases comparing changes in level, trend, and variability. The results showed good correspondence between DBR and SDO in terms of changes in level across phases, trends, and variability. These preliminary data highlight a promising approach to evaluating treatment outcome in settings in which SDO is difficult.

A variety of sequential analysis methods exist to quantify operant contingencies from observational data. Results of a recent simulation study indicated a modified event lag method may produce more accurate and interpretable contingency estimates relative to other standard approaches (Lloyd, Yoder, Tapp, & Staubitz, 2015). Whereas the simulation study modeled zero contingencies, the current study extends this research by applying the same methods of sequential analysis to observations with programmed non-zero contingencies. We used Sniffy the Virtual Rat software to generate observational sessions with programmed reinforcement contingencies, then superimposed fixed time schedules of reinforcement to model variation in contingency strength. We evaluated the degree to which each method of sequential analysis produced contingency estimates that (a) approximated programmed reinforcement contingencies and (b) demonstrated sensitivity to changes in contingency strength. Results indicated that event lag methods produced closer approximations of programmed contingencies and demonstrated greater sensitivity to changes in contingency strength relative to interval-based and time window methods. In addition, consistent with the previous simulation study, our results suggest that interval-based and time-window methods have the potential to produce negatively biased contingency estimates.

Advancing technology allows behavioral observation and analysis to move beyond summaries of partial-interval data, to methods that preserve the time sequence of events. With these changes come questions about analysis of indices of association available in different software packages for understanding contingencies between caregiver and participant behavior. Within the context of a larger study evaluating the impact medication changes had on challenging behavior of 11 individuals, weekly, one-hour direct observations were conducted over a period of 2.5 yrs. Videos were uploaded into The Observer XT and coded for the onset and offset of environmental conditions (e.g., location, noise level, number of people in the room), participant behavior (e.g., challenging and adaptive behavior), caregiver behavior (e.g., demands made, attention delivered, etc.). Data from observations were aggregated and analyzed using two tools: the lag sequential analysis function in Observer which provides unadjusted conditional probabilities, and the freeware Generalized Sequential Querier software which provides indices of association adjusted for base rates of occurrence of the event sequences in question (Yules Q). The results obtained from these tools produced different conclusions for the same events (e.g., Table 1). The benefits and challenges of interpreting results within a behavioral analytic contingency framework will be discussed.