Fig. 19From: Introducing hat graphsBias (calculated as percent overestimation) is potted as a function of graph type and whether there was no feedback (experiment 4) or there was feedback (experiment 5). Positive scores indicate a bias toward overestimating the depicted effect, and negative scores indicate a bias toward underestimating the depicted effect. The brims correspond to the no-feedback conditions and the crowns correspond to the feedback conditions. Error bars are ±1 SEM calculated from separate linear mixed modelsBack to article page