Meta-analysis for Single-Subject Research

Scruggs, Mastropieri, and Casto (this issue) advocate the use of the percentage of nonoverlapping data points (PND) to summarize and synthesize single-subject research. We take issue with the PND method, which reduces functional analyses to a one-number summary, for the following reasons: (a) the most relevant and credible evidence from single-subject designs lies in the patterns of change across time; (b) summaries based on PND may miss vital idiosyncracies in behavior within and across studies; (c) as practiced previously, syntheses of literature based on PND may misrepresent procedural facts and outcomes; and (d) as practiced previously, syntheses based on PND attempt, inappropriately, to draw conclusions about the relative merits of broad categories of intervention.