So you see, gentlemen, that what we need in our continuing study of this subject, is a larger sample size.
Not at all what I said or meant.
A larger self selected sample would give equally meaningless results.
Melas pointed out a sample of attendees at an AA meeting would not give an accurate take on American drinking habits.
A sample of attendees at 20 different AA meetings would not give an accurate take on American drinking habits.
A sample of attendees at 2,000 different AA meetings would not give an accurate take on American drinking habits.
Sample only girls who show up at a free sexually transmitted disease clinic and you'll get a biased impression of overall behavior, no matter how many you sample.