E Technical Notes on the Recalculation Exercise: Supplement to Chapter 3
Member, Committee on Unintended Pregnancy
How much change in children's family contexts could potentially be achieved if there were no childbearing derived from unintended pregnancy? Unwanted and mistimed pregnancies figure differently in this estimation. Births unwanted at conception* would, by definition, simply not occur, whereas births that were mistimed at conception would occur later—some to parents who are married and some to parents who are unmarried.
Births Unwanted at Conception
Births that are unwanted at conception are those that occur to women who say, prior to pregnancy, that they did not want to have any more children. The avoidance of such births would have a surprisingly large effect on the proportion of children born to unmarried mothers. It is little appreciated how large a proportion of childbearing among unmarried women is to women of higher parity, to women over age 25, and to women who want no more children.
It is necessary to adjust for some possible misreporting of mistimed births as unwanted births. Although the estimation of "unwantedness" is quite good in the aggregate, this is a category of births for which misclassification seems to
result in some bias. For example, it seems unlikely that many women who had their first child when unmarried and at a young age and who reported a birth as unwanted actually intended to remain childless in the future. In this exercise, first births to women under age 25 that were reported as unwanted at conception were recoded as mistimed. Although the resulting difference in estimated reduction of unwanted births is small (approximately 2 percent), it seems the most appropriate way to proceed.
For this exercise, it is necessary to begin with estimates based on births in 1986–1988, the most recent years for which there is complete information from the National Survey of Family Growth (NSFG). With these data and the above procedures, it is estimated that 28 percent of all births in 1986–1988 were either unwanted births (10.6 percent) or wanted births to unmarried women (17.6 percent). When the unwanted births are removed from the denominator (as would be the case in the absence of childbearing derived from unwanted pregnancy), 19.9 percent of the wanted births are to unmarried women. Hence, the avoidance of unwanted births in 1986–1988 would have resulted in a reduction in the percentage of all births that were either unwanted or to an unmarried mother, from 28 to 20 percent.
One further step is necessary, however, given the continued increase in the proportion of births that are to unmarried women. About one-quarter of all births in the 1986–1988 period were to unmarried women and the proportion is at present more likely one-third. In the recalculation, the relationships observed in the 1986–1988 data are applied to the current level, although this is a somewhat conservative procedure since levels of unwanted fertility also appear to have increased since that time. Under these assumptions, it is estimated that currently 27.2 percent of all births are wanted births to unmarried women, and (again) 10.6 percent are unwanted births. Removing unwanted births from both the numerator and the denominator implies a change from 38 to 30 percent of all births being either unwanted births or wanted births to unmarried women.
Births Mistimed at Conception
Births that are mistimed are those births that were wanted at some time in the future but that occurred sooner than they were wanted as a result of contraceptive misuse, nonuse, or failure. The effect of mistimed fertility cannot be properly estimated by simply observing the distribution of births intended at conception, since this would incorrectly reallocate births away from groups with higher proportions of unintended births, such as women with less education. For this reason, a logit regression is used in the recalculation to estimate the potential reduction from the elimination of mistimed fertility, since this procedure allows the distribution of births on other variables to be kept constant. Whether wanted births occurred outside of marriage (for births occurring 1986–1988 as reported
in the NSFG) was predicted in a logit equation that included the mother's race, education, and parity, as well as whether the birth resulted from an unintended pregnancy. The resulting coefficients were then used to estimate the proportion of wanted births that occurred outside of marriage, first with all variables set at their observed means and then with all other variables set at their means and the "unintended" variable set to zero. The first simulation closely replicates the observed proportion of births to unmarried women; the second simulation is 24 percent lower than the observed proportion of births to unmarried women. Hence, this analysis implies that, in addition to the reduction in the number of children born to unmarried women through the avoidance of unwanted births noted above, about a quarter of the wanted births that occurred to unmarried women would be delayed until after marriage. This would reduce the proportion of births estimated to be wanted births to unmarried women from 27.2 to 20.7 percent of births.
In sum, then, instead of the likely current level of 38 percent of all births that are either unwanted births (10.6 percent) or are wanted births to unmarried women (27.2 percent), the avoidance of unintended fertility could potentially reduce this proportion to the 21 percent that are intended births to unmarried mothers—a 45 percent reduction overall. Of course, the complete elimination of unintended fertility is an unrealistic goal, but the experience of other countries makes it likely that a serious effort could move the United States substantially in that direction—with proportional improvements in the well-being of future generations.