Ah yes, some will say, but look how large and consistent those sex differences are—in aggression, nurturance, verbal skills, spatial abilities, and so on. Surely this strongly suggests (even if it can’t absolutely prove) that women and men have innately different talents—“beneficial differences” in the language of both CMBW and (some) CBE adherents. Everybody knows that men are from Mars and women are from Venus—at least on average. Really? Just how large and consistent are such differences? Just how much do (or don’t) those bell curves overlap for women and men? Because there is so much bad science journalism floating around about these matters (written by people of every political and religious stripe), more comments on social science methodology are in order.
I begin with what is known among social scientists as the “file drawer effect.” Since the time that psychology journals began publishing over a century ago, there has been a heavy bias against accepting studies on males and females that find no statistically significant sex differences. Even when we do a comprehensive literature review, our conclusions will be selectively tilted towards finding more, rather than fewer, sex differences because studies finding no effect for sex are likely to remain unpublished (thus ending up in the author’s file drawer).
My second—and more important—point has to do with the misunderstanding that continues to surround the term “statistically significant.” A research result that is statistically significant is not necessarily of practical significance. According to the most common tests of significance, if an obtained, average difference between two groups could have occurred fewer than five times out of a hundred “by chance” then it is deemed a “significant” difference. Even a tiny average difference between two groups—i.e., groups whose bell-curve scores overlap almost completely—may be “significant” in this statistical sense, whereas (because of the file drawer effect) a much larger average difference that “just misses” being statistically significant will not likely see publication, even though its potentially practical significance may be much greater.
As a result of such criticisms, a statistical technique called meta-analysis was developed to combine the results of many studies on sex differences in a given domain: aggression, verbal ability, or whatever. This technique differs from earlier ways of reviewing the literature, which simply gave equal weight to all studies examined and came to an “eyeball” or intuitive judgment as to whether reliable sex differences existed in a given domain.
Instead, meta-analysis allows us to ask, across many studies of sex differences of a certain trait or behavior, just how large that difference (known as “d”) is, or how far apart the tops of the two bell curves are (the tops representing the place where the male and female mean scores are). In other words, across many such studies, just how much do the male and female bell curves (or “distributions of scores”) overlap?
Even when an average effect size (or “d”) is 1.00 the range of scores within each sex is much greater than the average difference between the sexes. But even as large as 1.00 is almost unheard of. Most are in the range from 0.0 (no detectable difference) to 0.35 (a small difference)—and even the latter means that less than 5% of the variability of ALL the scores can be accounted for by the sex of the participants. This underlines my previous assertion: it is naive at best, and deceptive at worst, to make essentialist pronouncements about either sex when the range of scores within each sex is, for almost all traits and behaviors measured, much greater than the difference between the sexes.
The meta-analytic “d” for women’s versus men’s “empathy” scores based on self-report measures is around 1.00, in the direction of women being more empathetic than men. But when based on unobtrusive measures (i.e., studies where people do not know they are being measured for empathy), the meta-analytic “d” shrinks to about 0.05.
Meta-analyses can also be divided according to the particular era in which the studies were done. For example, a meta-analysis of studies of gender differences in verbal fluency done prior to 1973 (when gender roles were more rigidly dichotomized) found an overall, small effect size (d) of 0.23, in the direction of women scoring higher than men. A similar meta-analysis of studies done after 1973 found an effect size of 0.11.
Attempts to Evade These Findings: What do convinced gender essentialists (along with careless science journalists and trendy Mars-Venus advice book writers) do with such findings? The most common strategy is simply to ignore or distort them. Tendencies don’t sell books.
A less common strategy nowadays is to claim that, however much those gendered bell curves do—or can—overlap, we have to pull them apart as far as possible, in order to approximate God’s—or nature’s or optimal society’s—”true” purposes for males and females.
This was the approach taken by philosopher Jean-Jacques Rousseau in his 18th century educational treatise, Emile. Rousseau was convinced that “rational, active man” and “emotional, passive woman” were perfect complements for each other. If they were not trained to become “opposite sexes” there was no way they would be attracted to each other and be able to pair-bond for life.
It is not unheard of for theologians to have taken a similar stance. Abraham Kuyper did so in the early 20th century, claiming that “The woman can lend herself to study [of medicine and law] as well as the man,” but however much men’s and women’s capacities “naturally”’ overlapped, because women’s (not men’s) “position of honor” was by divine definition in the home, “whoever has man take his place at the cradle and woman at the lectern makes life unnatural.”
So far, the doctrine of separate spheres is not an official affirmation of CBMW’s gender hierarchicalists. But to the extent that their rhetoric overlaps with romantic Mars-Venus rhetoric, as it does on the shelves of many Christian bookstores, it is an issue to be reckoned with in many evangelical churches. And to the extent that the doctrine of separate spheres, combined with the doctrine of male headship, results in the social and economic disempowerment of women (as it has in both preindustrial and industrialized cultures) it does not comport well with biblical notions of justice.
This points to a third strategy. Some gender essentialists have reluctantly recognized that neither the Bible nor the natural or social sciences can come definitively to their rescue. Consequently, they take refuge in biblically and empirically questionable Jungian gender archetypes, and their precursors in Greek mythology and Eastern religions.
For example, Elisabeth Elliot, in her 1982 book Let Me Be a Woman warned female Christian readers that Eve, in taking the initiative to eat the apple, was trying to be like the “ultimately-masculine” God—as if God were somehow metaphysically gendered. She also appealed to the ancient Chinese concept of yin and yang to buttress her “Christian” argument for gender essentialism and gender hierarchy. Her brother Thomas Howard, in a 1978 article titled “A Note From Antiquity on the Question of Women’s Ordination,” frankly acknowledged that the Bible does not supply enough resources to justify talking about God or humans in terms of metaphysical, eternal gender archetypes. Undeterred by this, he invited his readers to consider the abundance of sexual imagery in pagan myths, and came to the conclusion that “a Christian would tend to attach some weight to this.”
Joan Burgess Winfrey is thus right, in chapter 25 of Discovering Biblical Equality (DBE), to express concern that “the church may once again opt for a Venus-Mars gender rubbish in the interest of cementing roles [or styles] and putting up divider walls.” Mars-Venus rhetoric gets virtually no support from the meta-analytic literature which, as we have seen, show almost complete overlap in the gendered distribution of traits such as nurturance, empathy, verbal skills, spatial skills, and aggressiveness. The romanticizing and/or rank-ordering of gender archetypes is biblically questionable whether it is done by gender-role traditionalists, by cultural feminists who reverse the hierarchy by valorizing the stereotypically feminine, or by evangelical writers who baptize the trendy Mars-Venus rhetoric with a thin, Christian-sounding veneer. More in keeping with both the biblical creation accounts of humankind and the overall findings of the social sciences is the bumper sticker which reads “Men are from Earth, Women are from Earth: Get used to it!”
Check out Part 1 of this series and join us next week for the final installment.