Outcome Assessment and Inference With the Percentage of Nonoverlapping Data (PND) Single-Case Statistic

Single-case experimental designs allow practitioners to conduct clinical outcomes research without the large samples and substantial resources required by randomized clinical trials. Single-case designs have been used to conduct outcomes research for many decades; however, the statistical measurement of treatment effect sizes remains an unresolved issue. The percentage of nonoverlapping data (PND) is one widely used statistic for effect size measurement of single-case experimental designs. Despite its limitations, PND is useful because it is easy to calculate and interpret. However, null hypothesis significance testing (i.e., the use of p values) is not currently feasible with PND because it has an unknown sampling distribution. A method to calculate p values for PND is introduced and discussed. An online calculator and statistical computing code are also made available to single-case investigators who wish to calculate p values for their data. Calculating PND and its associated p values may provide practitioners with valuable insights about their treatment outcomes when PND is used appropriately and its statistical assumptions are not violated.