Consistent Visual Analysis of Multielement Data: A Preliminary Evaluation

Abstract

Experimenters provided 33 graphical displays of hypothetical data depicted in a multielement experimental design to editorial board members of prominent, applied, behavior-analytic journals via an online survey. For each display, participants indicated (a) the presence or absence of experimental control and (b) the degree of experimental control (rated on a 1–100 scale). Each depiction varied systematically in (a) the number of data paths, (b) the number of data paths elevated above the control, (c) the mean difference between affected data paths and control conditions, and (d) the degree of variability within conditions. Correspondence among experts’ ratings of experimental control was high across all presented graphical displays, supporting the reliability of visual analysis as an evaluative tool for these designs.