Figure 1.

Permutation and n-jackknife test for bivariate testing of performance-related changes. The example illustrates the concept of the testing procedure. Bootstrapping and n-jackknife of two-samples [(A): sample 1, (B): sample 2] from two conditions [(C) condition 1, (D) condition 2] were used. These data were sub sampled based on a combination of an n-jackknife and bootstrapping. The bootstrapping generates 12 bootstrap samples which contain random 75% of the trails and were used to determine variability in each data set (upper plots). In addition an n-jackknife resampling was used to partition the data into four quarters in order to check for stability during the experiment (lower plots). In total, the bootstrap data comprise 16 sub samples for each condition. To test whether differences between conditions were significant, a second permuted dataset was used. To this end, trials were permuted across conditions. After trial permutation, the same n-jackknife bootstrap procedure was employed on the permuted data.