Unintended pregnancies and the resulting births, which occur disproportionately among unmarried and young couples, are of particular concern to policymakers and researchers. Although unintended pregnancy rates in the United States have declined in the past two decades, they are still high. In 1994, approximately 49% of all U.S. pregnancies were unintended; 54% of these ended in abortion.1 The prevalence of unintended pregnancies was particularly high among teenagers (78%), young adults aged 20–24 (59%) and never-married women (71%).
Empirical research has shown that unintended pregnancy is associated with the mother's mental health and with behaviors that may adversely affect infant and child health.2 For example, women who become pregnant unintentionally are more likely than others to have prenatal and postpartum emotional problems and to smoke heavily during pregnancy, and less likely to initiate prenatal care and to breast-feed after giving birth.
The findings on the effects of unintended pregnancy on children's outcomes, however, are mixed. Children born of unintended pregnancies do not have significantly lower birth weight or cognitive outcomes compared with their counterparts, when mothers' socioeconomic characteristics are controlled for.3 Children born of unintended pregnancies, however, tend to have more behavioral problems and worse relationships with their mothers.4 Furthermore, in research that divided unintended pregnancies into unwanted and mistimed, children born of unwanted pregnancies had the worst outcomes (e.g., low birth weight), whereas maternal behaviors and child outcomes did not differ between children born of mistimed and wanted pregnancies.5 In short, it seems that unintended, especially unwanted, pregnancies may generate substantial costs for mothers and children.
Previous studies of mothers' pregnancy intentions have consistently found that women who are unmarried, women with low educational attainment and women with low income are more likely than others to have mistimed or unwanted pregnancies.6 Race and age are important determinants as well: Blacks and young women are more likely than others to have unintended pregnancies.7 In addition, human behavior can be affected by the external environment. Some studies have found links between welfare assistance and teenage fertility,8 whereas others have shown the importance of family planning services, effective birth control methods and economic opportunities in reducing teenage fertility.9
Although much research has focused on women's characteristics and on assessing how community programs and public policies change women's behaviors, little attention has been paid to men, even though decisions about sexual activity and contraceptive use usually involve both men and women. One area that needs further investigation is the relationship between child support enforcement and the degree to which men take responsibility for their sexual activity. If government child support enforcement efforts are successful, men may be more likely to practice contraception or reduce sexual activity, which in turn may reduce unintended pregnancies and birth.
Even though child support enforcement has intensified during the past decades10 and the most recent welfare reforms were specifically aimed at reducing nonmarital births, only a few studies have examined the effects of child support enforcement on nonmarital births.11 Only one study has looked specifically at the association between perception of child support and men's sexual activity,12 finding that adolescent males who perceived a high likelihood of being required to pay child support as a nonresident father or who had a relative who pays child support are less likely than others to have multiple female partners and more likely to use contraceptives. These findings suggest that strict child support enforcement may be successful in deterring teenage pregnancies and nonmarital births.
Different variables have been used to measure the strength of state child support enforcement, including child support legislation, per capita child support expenditure, paternity establishment rate, child support receipt rate, average payment and collection effectiveness.13 Not surprisingly, findings have varied greatly in level of significance and in direction, possibly because of difficulties in accurately specifying child support enforcement and because effects have been assessed through individual measures. Successful enforcement of child support requires three steps: establishing a legal child support order, determining the award amount and collecting the payment. Moreover, effective child support enforcement requires strict child support legislation, sufficient expenditures and strong implementation. The beneficial effect of child support legislation might not be observed unless strong and effective enforcement is implemented.14
This article examines the association between state child support enforcement and pregnancy intention from men's perspectives, in the hope of shedding light on whether strengthening child support enforcement may not only increase the economic well-being of children in single- parent families, but also indirectly prevent unwanted pregnancies and births. Given that public attention has been particularly focused on nonmarital births and that associations between child support enforcement on pregnancy intention may be different for married and never-married men, the article specifically investigates child support enforcement and men's pregnancy intention by their marital status.
Data used in these analyses come mainly from the 14 waves of the National Longitudinal Survey of Youth (NLSY) between 1982 and 2002. The NLSY began in 1979 with a nationally representative sample of 12,686 men and women aged 14–22; the latest wave of publicly released data is from 2002. The 1982–2002 waves include data on whether men wanted a pregnancy, as well as detailed information on individual, family background and local labor market measures known to be associated with fertility behavior. With respect to state characteristics, child support enforcement data come from a national study by Huang, Garfinkel and Waldfogel;15 information on welfare policy was drawn from data from the U.S. House of Representative's Committee on Ways and Means.16
The pooled sample size for this study was 51,948, including 23,686 married men and 28,262 never-married men. Formerly married men were dropped from the analyses because the data do not indicate whether their most recent children were conceived with their ex-spouses or their current partners and child support enforcement may have different associations for currently and formerly married men.
•Dependent variable. In the 1982–2002 waves of the NLSY, all male respondents were asked whether they and their partner had had a child since their last interview or were currently expecting a child. Men who answered yes were asked, "Before the recent pregnancy, were you and your spouse/partner trying to have a baby or trying not to have a baby?" Possible responses were "trying to have a baby," "trying not to have a baby" and "neither." The first response was used as the measure of wanted pregnancy, and the second as the measure of unwanted pregnancy; the pregnancy intention of respondents choosing the third response was defined as undetermined.
•Explanatory variables. The child support enforcement index consisted of the summative rating of standardized scores from variables measuring numbers of state child support legislation, amount of expenditure on enforcement per case and degree of implementation performance (e.g., the proportion of eligible mothers with child support payment). The Cronbach's alpha of the index is 0.92. The child support enforcement index had a mean of –0.22 and a standard deviation of 0.80. A score below –0.22 signifies the child support enforcement in that state was less effective than enforcement in other states over time; a score above the mean signifies greater effectiveness.
The analyses controlled for a range of variables that have been associated with fertility measures (e.g., nonmarital births) in previous studies: age, race, education, employment status, religion, number of siblings, family structure at age 14, whether a foreign language was spoken in the home during childhood, mean local unemployment rate and the state's maximum welfare benefit (in 2001 dollars).17 Age, number of siblings, local unemployment rate and maximum welfare benefits were measured as continuous variables. Four dummy variables were created for race (white, black, Hispanic and other). Education was measured via three dummy variables (less than high school, high school and more than high school). Dummy variables were created for the remaining individual characteristics, with a value of one equaling employed, Catholic, intact family and use of a foreign language at home, respectively.
Because of their small number and ambiguous meaning, undetermined pregnancies were dropped from the analyses. Thus, the choices for the dependent variable, pregnancy intention, were wanted, unwanted and no pregnancy.
To compare individual choices, multinomial logit analyses were performed; individual, local and state characteristics were included as independent variables. A dummy for cohort effect was included to control for unmeasured cohort preferences, where zero equaled birth prior to 1962 and one equaled otherwise. To minimize biased estimates of child support enforcement and unwanted pregnancies, state and year dummies were included to account for unobserved characteristics between states and over time. For example, some states may have a more conservative attitude toward unwanted pregnancies than others, and this attitude may influence these states to implement strict child support enforcement to reduce unwanted pregnancies. A failure to measure such attitudes might result in an overestimation of the role of child support enforcement on unwanted pregnancies. With state and year fixed effects, the only way in which a state-level variable, such as child support enforcement, could be associated with the dependent variable would be through its changes within a state over time. Variables that are largely constant within states over time or that affect all states in a given year are subsumed within the state and year fixed effects.* Robust standard errors were used to account for the fact that some respondents are represented in the data more than once. Without this correction, the standard errors produced by the model would be underestimated.
Finally, simulations were created using regression results to estimate how men would have fared if they had experienced different forms of child support enforcement. Specifically, the child support enforcement index in this analysis improved from –0.76 in 1982 to 1.06 in 2002. A simulation was created were the child support enforcement index for all respondents was set at –0.76 and then multiplied by its estimated coefficient in multinomial logit analysis to calculate the prevalence of unwanted pregnancies that would have occurred if child support enforcement had not changed after 1982. These estimated unwanted pregnancy rates are then compared with the observed rates.
Overall, 17% of men reported that their spouse or partner had been pregnant since the last survey (Table 1): Ten percent reported wanted pregnancies and 4% unwanted; the pregnancy intention of 3% was undetermined. There were significant differences by marital status. Among married men, 18% reported a wanted pregnancy, and 6% an unwanted pregnancy; the proportions for never-married men were 2% and 3%, respectively. That is, 46% of pregnancies reported by never-married men were unwanted, compared with 21% of those reported by married men.
Men's personal, local and state characteristics differed by marital status, as well. Compared with never-married men, married men were older (mean, 31.2 years vs. 25.5 years) and lived in states with a higher child support enforcement index (0.0 vs. –0.4). In addition, greater proportions of married men than of never-married men were white (63% vs. 49%), employed (91% vs. 71%) and Catholic (33% vs. 30%); had more than a high school education (39% vs. 35%); had lived in a two-parent family at age 14 (74% vs. 65%); and had had a foreign language spoken in the home during childhood (23% vs. 21%). Never-married men, on the other hand, lived in states with higher mean maximum welfare benefits ($658 vs. $585).
In the multinomial logit regression analyses, the associations between men's pregnancy intention status and selected characteristics were examined. Child support enforcement, the main focus of this paper, was found to be marginally associated with unwanted pregnancy in two ways (Table 2, page 121). First, increased child support enforcement was associated with lower occurrence of unwanted pregnancy relative to no pregnancy (coefficient, –0.14). Second, increased child support enforcement was associated with lower occurrence of unwanted pregnancy relative to wanted pregnancy (–0.15). These findings suggest that strong child support enforcement may play a role in reducing the odds of unwanted pregnancy by reducing men's odds of being involved in a pregnancy and increasing their odds of experiencing a wanted pregnancy.
For individual characteristics, the results generally agree with previous empirical findings. The characteristics that were positively associated with wanted pregnancies rather than no pregnancy were being black, having greater than a high school education, having a greater number of siblings and having had a foreign language spoken in the home during childhood (coefficients, 0.03–0.21); being older and being never-married were negatively associated with wanted pregnancy rather than no pregnancy (–0.08 and –2.73, respectively). The characteristics that were positively associated with unwanted rather than no pregnancy were being black, Hispanic or employed, and having a greater number of siblings (0.03–0.95); older age, having greater than a high school education and being never-married were negatively associated (–0.08 to –1.51). Finally, the characteristics associated with unwanted pregnancies rather than wanted pregnancies were being black and being never married (0.73 and 1.22, respectively). In contrast, having more than a high school education and having had a two-parent family at age 14 were negatively associated with unwanted rather than wanted pregnancy (–0.60 and –0.16, respectively).
Given that public attention has often been focused on nonmarital births, and that the impacts of child support enforcement may differ by marital status, the multinomial logit regression analyses were reestimated by marital status (Table 3). For simplicity, only child support enforcement results are reported. For never-married men, strong child support enforcement was associated with lower occurrence of unwanted pregnancy relative to wanted pregnancy (coefficient, –0.36) and marginally associated with a higher occurrence of wanted pregnancy relative to no pregnancy (0.24). The findings suggest that for never-married men, increased child support enforcement may reduce the odds of unwanted pregnancy by increasing the chances of wanted pregnancy. For married men, increased child support enforcement was associated with decreased odds of unwanted rather than no pregnancy (–0.19), which suggests that strong child support enforcement may reduce married men's odds of unwanted pregnancies by increasing the chances of no pregnancy.
If child support enforcement had not improved over time, men's overall rate of unwanted pregnancy would have been 4.8%, rather than the observed 4.5% (Table 4). This suggests that improved child support enforcement is associated with an estimated 7% reduction in unwanted pregnancies over the study period (not shown). Likewise, the estimated reduction in unwanted pregnancies is about 8% for married men and 6% for never-married men.
The findings from this study suggest that the proportion of pregnancies that are unwanted is high. About half of pregnancies reported by never-married men and one-fifth of those reported by married men were unwanted. Strong child support enforcement was marginally associated with lower odds of unwanted pregnancy, and if child support enforcement had not improved over time, the predicted unwanted pregnancy rate would have been approximately 7% higher than the observed rate.
The finding of an association between child support enforcement and unwanted pregnancy suggests that strong enforcement would likely compel men to take more financial responsibility for nonmarital births, which might reduce the likelihood of unwanted pregnancies among men who do not want children. For men who want children, stronger child support enforcement should reduce nonmarital childbearing by making marriage more attractive.18
On the other hand, some might contend that stronger child support enforcement may make women more willing to give birth outside marriage because it would reduce the cost of raising children. However, the size of such an effect is likely to be small, at least among welfare recipients, because welfare policy prevents financial benefits from being collected if a mother is receiving welfare. For example, before 1996, a mother on welfare could keep only the first $50 of child support each month. If the father paid more than $50, the mother's welfare benefits were reduced accordingly. The 1996 welfare reform gives states the option of eliminating the $50 payment to mothers, and most states have taken up this option. The deterrent effect on men is shown to outweigh the possible incentive effect for women in this analysis.
The association between child support enforcement and unwanted pregnancy is also important given that current abstinence and sex education programs do not include information on child support.19 The disproportional focus on women's pregnancy intentions in public debates over teenage pregnancy, unintended pregnancy and nonmarital births neglects the significance of men and child support enforcement to fertility. Increased emphasis on child support may influence men into avoiding unintentional pregnancies and births. Studies have shown that unwanted pregnancies and births, particularly among adolescents and young adults, are likely to result in adverse social, economic and health outcomes for children and parents.20 The reduction in unwanted pregnancies and births that would probably result from taking greater responsibility might not only reduce costs associated with childbirth and child-rearing, but also generate opportunities for men and women that would affect their long-term potential.
The results of this study should be interpreted in the context of several limitations. First, the information on pregnancy intention is self-reported and is subject to intentional and unintentional reporting errors. Some men, especially never-married men, may desire to conceal information about their partners' pregnancies or births. Such intentional reporting errors would lead to an overestimate of the "no pregnancy" level. In addition, some never-married men simply might not be aware of or have accurate information regarding their partners' pregnancies and births. This unintentional reporting error would also lead to an overestimate of the prevalence of no pregnancy. Depending on the degree to which these reporting errors were present, they could have biased the estimated coefficients of child support enforcement. Second, child support enforcement may have simultaneous effects on marriage and pregnancy. In the face of strong child support enforcement over the survey period, some men and their pregnant partners—no matter what their pregnancy intention—may have married. This would have affected the observed marital status and biased the estimates from analyses by marital status. Third, the analyses show associations between pregnancy intention and the explanatory variables, but they do not illuminate the mechanisms underlying those associations. Still, the inclusion of cohort, state and year dummies enhances confidence in the connection between state child support enforcement and unwanted pregnancies.
Despite such limitations, this article advances our understanding of overlooked issues regarding men's perspectives on pregnancy intention.