Question: The best example I can think of were a nonparametric test methods would be more beneficial is if we try to measure the average number of days in fighter pilot training before the typical male and female trainee dropout occurs. The use of a histogram would be used and from that we would see the distribution of females and determine their mean number of days in training before dropping out. The distribution, median and means for males then could be compared to those of the female pilot trainees. I expect they would be significantly different. Based on that using the assumption of normality associated with a parametric test would probably not work. I'm interested in feedback on this example to see if you agree.