Polymath1976
Polymath1976
Scribbled:
That is an interesting and fair point, John. Indeed, one might get strange results estimating a population average if one missed outliers (prostitutes, Wilt Chamberlain, etc.) when collecting data. (I allude to the possibility in my third footnote.) But keep in mind that these are not ordinary discrete data we are working with, since they refer to relationships between two elements (or edges between nodes, if you like). The key point is that a yes answer in the female group requires a yes answer somewhere in the male group and vice versa. Promiscuous people (women or men) would act as hubs in a network, but no matter how high their numbers, they could add at most one to anyone else's count. Missing such people in a random sampling is possible, but missing enough of them to create the type of disparity seen in this survey and others like it is tremendously unlikely. What I think is the case with the survey I looked at is that two phenomena occurred: 1) The polled subjects were less than completely honest. 2) The survey itself was poorly designed. It is possible that the second phenomenon was the most weighty in this case, but I am convinced that the first also played a significant role. Thank you for your thought-provoking comments!



