I don't know which statistical analysis Kiefer would do, but we could estimate the the mean difference between bench press increase of the two programs within .573 standard deviations with a sample size of 10 with 95% confidence. With 20 it would be within .386 standard deviations.
Since this experiment will probably vary a large amount we probably will need 20 people on each side to get an accurate estimate. For most experiments, 10 would be enough, but I'm guessing that won't be enough, unless we control for training years, famililiarity with 5/3/1, familiarity with the bench press, and starting bench press. There are other factors, but I would think these would be the big ones.