Fig. 2

Model predictions from the combined analysis of Study 1 (left column) and Study 2 (right column) for the posttest (top row) and the transfer test (bottom row). The multilevel logistic models predicted posttest or transfer test performance from two-way interactions between gesture group and verbal working memory span, and gesture group and visuospatial working memory span, with composite ACT score as a covariate, and random intercepts for participant and problem. For graphical purposes, performance was separately predicted from verbal and from visuospatial working memory capacities while holding all other variables in the model constant. Error bars represent the 95% confidence interval of the predictions