Why babies die in D.C - part 2 - includes reply by original author

Public Interest, Fall, 1994 by Sanders Korenman, Nicholas Eberstadt

IN THE SPRING 1994 ISSUE of The Public Interest (Number 115), Nicholas Eberstadt argued that the high rate of low birth-weight births and infant mortality among African Americans in Washington, D.C., is due mainly to high rates of illegitimacy and other parental behaviors:

Impersonal social forces--material deprivation, joblessness, economic insecurity--cannot explain why one of the very richest black populations in America suffers from black America's very worst infant mortality rates. This perverse situation can, however, be explained in terms of dysfunctional or even pathological behavior by parents and adults--including parents and adults who happen to be neither poor nor poorly educated. Illegitimacy, welfare dependence, and the environment of violent crime mark out a vector of deadly risks to infants in Washington, D.C.

The Public Interest packaged Eberstadt's article with another by Charles Murray, who once again asserted that welfare causes illegitimacy. It is a small logical step to the conclusion that welfare, intended to help poor children, kills black babies. The list of Murray's critics is long and distinguished, but Eberstadt's thesis remains largely unchallenged. Yet, there is little if any evidence that nonmarital childbearing plays an important role in raising rates of low birth-weight (LBW) births among blacks in Washington, D.C., compared to blacks nationally and, by extension, among blacks compared to whites.

No one can deny that rates of LBW are higher for nonmarital than marital births, nor that high rates of LBW births contribute to high rates of infant mortality. Since African Americans have high rates of LBW births, they also face high infant mortality rates. But, of course, this does not mean that having a nonmarital birth increases the risk of an LBW birth. Women who have marital births differ from those who have nonmarital births in ways that do not directly result from marital status. Women who have marital births are likely to have grown up in more advantaged circumstances, and they and their children would face fewer health risks than less-advantaged women, even if they were to give birth out of wedlock. For example, the parents of black women who had marital first births (the infants' maternal grandparents) were 30 to 60 percent more likely to have completed education beyond high school, and the mothers themselves scored 50 percent higher on a test of academic aptitude (even among those who were tested prior to their first births), compared to black women who had nonmarital first births.(1) Differences in background characteristics between women who have births maritally and nonmaritally are even more dramatic for whites and Hispanics. Therefore, simple comparisons of LBW rates between marital and nonmarital births that ignore these background differences are likely to overstate the causal contribution of illegitimacy to high LBW rates.

In order to address the "correlation is not causation" problem, Eberstadt conducts a statistical analysis of LBW rates in 170 Washington census tracts. He finds that a higher rate of nonmarital childbearing in a census tract is associated with a higher LBW rate, even when he controls for the poverty level, education, racial composition, and median household income. However, he acknowledges that his evidence is flawed for two reasons. First, because his data refer to census tracts rather than individuals, interpretation is subject to the "ecological fallacy": what pertains to neighborhoods need not pertain to individuals. Second, the LBW rates he uses are for all births, not for births to African Americans.

BUT LET US for a moment sidestep the correlation versus causation morass and simply assume that differences in LBW rates between marital and nonmarital births are caused by illegitimacy. The question then becomes: How much do high rates of nonmarital childbearing contribute to the high LBW rates among Washington blacks compared to U.S. blacks overall? Although Eberstadt presents a careful and informative analysis of cross-neighborhood differences in LBW rates in Washington, D.C., he does little to answer this central question.

One might not be bothered by the use of an indirect approach if it were the best way to answer the question of interest. I believe it is not. In particular, there is a standard demographic technique for comparing rates across populations that differ according to some important characteristic--in this case, marital status. This technique--rate standardization--is appropriate for answering exactly the kind of question highlighted in italics in the previous paragraph.

The standardization "thought experiment" goes as follows. We know the LBW rate among U.S. blacks overall is 136 per thousand compared to 167 in Washington, D.C.(2) We also know that 68 percent of U.S. black births are nonmarital, as compared to 79 percent in Washington. If nonmarital childbearing among Washington blacks were to fall to the level of U.S. blacks overall, how much would the Washington black LBW rate decrease? The answer is: hardly at all. To see this, assume that the rate of LBW births for each marital status group remains the same in Washington (178 per thousand nonmarital births, 125 per thousand marital births) as the nonmarital proportion decreases from 79 to 68 percent. Then, the LBW rate among Washington blacks would fall to 161 per thousand (161 = .68 x 178 .32 X 125) from 167 per thousand, a decrease of six per thousand or about 3.5 percent. The inescapable conclusion is that higher rates of nonmarital childbearing account for virtually none of the high rates of LBW births among Washington blacks compared to U.S. blacks. Although reducing nonmarital childbearing in Washington, D.C., may be desirable for other reasons, it will not narrow appreciably the gap in LBW rates between Washington blacks and blacks nationally.