On the epistemology of autocorrelation in applied behavior analysis

A number of earlier papers had suggested that autocorrelation is contained in the data for many behavior analyses. Recently, this position was challenged. Secondary analysis results showed that behavior analysis data are generally not autocorrelated. In this paper, an argument is advanced that the findings of no or little autocorrelation in behavior analyses data can best be described as inconclusive. It is further argued that the burden of proof is on the nonexistence rather than the existence of autocorrelation. To strengthen the conclusion of no autocorrelation, a number of methodological problems need to be minimized.