Evaluating progress in behavioral programs for children with pervasive developmental disorders: Continuous versus intermittent data collection
It is well documented that intensive behavioral treatment of early childhood autism can result in significant improvements in adaptive behavior. The typical teaching format in such programs is based on the restricted operant (i.e., discrete trial) in which the performance of an exemplar skill follows a clear instruction and precedes programmed reinforcement or error correction. Because of the often-intensive nature of behavioral treatment, it is not unusual for thousands of learning opportunities to be presented each week. There currently exists a professional debate regarding the frequency of data collection necessary in autism treatment programs. One side of the argument favors collecting data on every learning opportunity for a complete assessment of child performance. The other side favors intermittent data collection to facilitate more efficient instruction. Unfortunately, little published empirical evidence exists to inform the debate. Thus, the current study was designed to evaluate continuous (i.e., trial by trial) versus intermittent (i.e., first-trial only) data collection systems across a number of curriculum areas in behavioral treatment programs for children with pervasive developmental disabilities. In our study, 6 children were taught numerous exemplars in 2–4 curricular areas using established behavioral procedures. The exemplars within each curricular area were randomly assigned to one of the data collection conditions. Each condition was evaluated based on the number of sessions to reach a mastery criterion for an exemplar and the percentage correct score for that exemplar at a 3-week follow-up assessment. In general, our results indicate intermittent data collection resulted in slightly more efficient mastery in terms of number of sessions and minutes to mastery. However, continuous data collection appeared to result in slightly better maintenance of skills at a 3-week follow-up assessment. Thus, the data can probably best be interpreted as evidence that both continuous and intermittent data collection appear to similarly impact behavioral instruction. Although the experimental preparation employed in this study is not representative of all teaching circumstances, the data suggest that collecting data on only the first trial of a session might be a reasonable approach. If the margin of difference between the two data collection systems is as minimal as the data suggest, it may be a reasonable assumption that the intermittent data collection system would be less cumbersome, and more efficient while still accurately portraying data. Although the experimental preparation employed in this study is not representative of all teaching circumstances, our data suggest that collecting data on only the first trial of a session might be a reasonable tactic.