Fig. 4.

Analysis of the amount of data required to obtain a significant role switch cost effect in each experiment. Mixed effects model comparisons (for models with and without Role as a factor) that were identical to those reported for each experiment were calculated on data from each block, cumulatively (i.e., for cumulative block number on the x-axis, each block number also includes data from all previous blocks, e.g., the data point for block number 30 represents a statistic calculated using models with data from blocks 1–30). The dotted line in each plot indicates the chi-square value required for a level of significance of p < .05 for that experiment’s model comparison. Blue points indicate significant chi-square values. (Points for some data subsets do not appear because models using those subsets did not converge). These plots indicate that fewer blocks of trials would have been sufficient for detecting the role switch cost in each experiment (in some cases, such as in Experiments 1a and 2, we would have detected the switch cost with as little as one-tenth of the data).