Academic Legal Writing - Eugene Volokh [90]
Be similarly cautious when people draw inferences from the general to the specific, as well as from the specific to the general. Just as behavior in San Francisco may not tell you much about behavior in the U.S. as a whole, so behavior in the U.S. may not tell you much about behavior in San Francisco. If the median homosexual male in the U.S. has had 10 sexual partners in his life (that seems to be the best data that I've found, based on surveys conducted from 1991 to 2002*), it doesn't follow that the median homosexual male in San Francisco has had the same number.
c. Extrapolating across time
Likewise, remember that most of the claims you read are based on data gathered at a particular time, often several years ago. Behavior patterns—sexual behavior patterns, crime rates, accident rates, and more—change over time. Inferring that a population is behaving the same now as it did ten or twenty years ago may be a mistake.
People have sex and commit crimes whatever the decade; but they may do it more in one decade and less in another. You may draw the inference that behavior today isn't materially different from what it was at some time in the past, but make clear to your readers that you're drawing this inference, and explain why it's sensible to draw it.
d. Extrapolating across populations
Finally, be especially cautious about inferences drawn from one subgroup of the population to another, or to the population as a whole. Consider, for instance, this quote from a scholarly-seeming and heavily endnoted book; all the material in the quote, including the ellipses, appears literally in the book, though I've omitted the endnote calls (which refer to citations that appear at the end of the book):
Research suggests there are qualitative and quantitative differences between patterns of homosexual and heterosexual activity. There is ample evidence homosexuals are likely to have significantly greater numbers of sexual partners than heterosexuals. Examples in the literature include studies showing “homosexual men ... reported a median of 1,160 lifetime sexual partners, compared with ... 40 for male heterosexual intravenous drug users”; “homosexual men had significantly more sexual partners in the preceding one month, six months, and lifetime (median 2, 9, and 200 partners, respectively), than the heterosexual subjects (median 1, 1, and 14 partners)”; and “homosexual patients are likely ... to have more partners ... than heterosexual patients.”
It is common in the literature to find homosexuals reporting median lifetime numbers of partners in excess of 1,000. One study reported the “median number of lifetime sexual partners of the [more than] 4,000 [homosexual] respondents was 49.5. Many reported ranges of 300-400, and 272 individuals reported ‘over 1,000’ different lifetime partners.” Another study reported:
[h]eterosexual patients from all risk groups reported considerably fewer sexual partners than did homosexual men, both for the year before onset of illness and for lifetime.... Homosexuals had a median of 68 partners in the year before entering the study, compared to a median of 2 for heterosexuals.... Homosexuals in the study had a median of 1,160 lifetime partners, compare to a median of 41 for heterosexuals in the study.
In another study of 93 homosexuals, the “mean number of estimated lifetime sexual partners was 1,422 (median, 377, range, 15–7,000).”
Sounds pretty remarkable: More than 1000 sexual partners for the median homosexual man—even more than the textbook I quoted above reported—which is to say that half of all homosexual men have had more than 1000 sexual partners. Naturally, one can debate what the legal consequences of this should be. But if it's true, or even if the more modest estimates of 200 or 377 are