Statisticians are people who explain ballistics to a dog so he can have a full understanding of its flight path in order to catch it. Scientists are dogs who just want you to throw the darn ball. I concluded this after wading through half a dozen papers written by statisticians, for statisticians, on the estimation of effect size in multiple ANOVA. Just throw the darn ball!

Three people were driving in a car along a country road. The fields around were barren until they all saw a black sheep in profile upon a low hill. "Ah," said the statistician, "all the sheep in this area are black". "On the contrary," said the ecologist, "we can only conclude one sheep in this area is black." "Perhaps not," said the mathematician. "All we can logically conclude is that one side of one sheep in this one area is black." The three drove on in silence.

Or the physics major and math major who walked into room where there were two naked people of the opposite sex on the other side. They were each told they could approach the person, but only go half way each time. The math major left saying it wasn't worth it has only going half way each time they'd never arrive. The physics major stayed, knowing that they'd get close enough together for practical purposes.

The importance of statistics is very well understood in the scientific community, but not well understood by the public.

When I try to explain to full professors why Dunnett's multiple t is appropriate and Tukey's is not for a given design, their glazed expressions lead me to believe that the importance of statistics is not that well understood in the scientific community, either. And trying to get them to believe that ANOVA is not the right tool to use for each and every possible experiment is quite a bit of work, too.