Mon, 26 Sep 2005 07:07:27 -0700 David Gale <gale@math.berkeley.edu>
Dan writes I don't get why replacing "sexual partners" with "spouses" makes any inconsistency clearer. It seems easy to believe that unmarried sexual behavior tends to be different from married.
If some one asserted that there were twice as many married men as married women in America most people (I hope) would say "Wait a minute." Not if each woman had two or more husbands... The spouse counting model is almost as obvious. Actually I'm told women more often then not outlive their husbands so in fact the survey should probably show the average (or median) woman has more spouses then the median man but surely not twice as many. It doesn't matter whether the coupling is marriage, copulation, or exchanging e-mails. If you could poll the whole world and get accurate response the only way getting a different number of pairings for the men and the women is of one member of a pair reports and the other doesn't. The sum, yes, but not the average. Even with totally accurate reporting, the average number per man could differ from the average number per woman, if there were more women in pairs than men. My main point is that it is not responsible for a highly official agency to publish statistics that they know cannot be correct. In the report's "highlights", also quoted in the NY Times, they make the statement about the median number of partners for men, 6 to 8, and women, 3.8 with no caveats. How many people will go to page 12 to find out that this is wrong?