Make random A & B samples
with items in each sample
simulated from populations whose means differ by

Each permutation resampling is done by:

randomly shuffling the combined set of 40 items contained in A and B,

dividing that randomized set into two groups the size of the A and B samples,

computing the mean value (i.e., the average) for each group,

subtracting one mean value from the other, and

using that difference-of-means value to increment the histogram.

A-B mean difference values for 250 sample permutations-20-15-10-50510152001020304050The red circle indicates the amount of difference between the means of the A and B samplesLower differences:54.0% of resamplesHigher differences:46.0% of resamples

The graph is split into two regions based on the original difference in the means of the A and B samples.
Each region is labeled with the percentage of the resamples for which the difference in the resampled means is on that side of the original difference.