Fig. 4.

Baseline-adjusted mean accuracy stratified by treatment, trial difficulty, and sex (A) and fruit and vegetable (FV) intake (B). The pairwise tests showed significant differences between the placebo and blueberry groups in males on both switch trials (p = 0.018) and non-switch trials (p = 0.031). In addition, accuracy was greater in low FV consumers on switch trials (p = 0.028) as well as non-switch trials (p = 0.025). The values are evaluated at a baseline accuracy of 72.8 for switch trials and 87.5 for non-switch trials. Error bars represent standard error of the mean. EMM estimated marginal means * denotes p < 0.05