Individuals' intentions to have a child consistently predict whether they eventually do so. According to an analysis of data from two rounds of the National Survey of Families and Households,1 the associations between respondents' intentions and their actual childbearing are strong and highly significant, even when life-course and background variables are controlled for. These associations decrease over an intentions continuum that goes from being very sure of intending to have a child to being very sure of not intending to do so: The odds of having a birth between the surveys fall from 7.2 among those who were very sure they would have a child to 1.8 among respondents who were initially moderately sure about not having a child (all compared with those who were very sure they did not intend to have a child). Only marital status has an effect comparable to that of childbearing intentions on the likelihood of a birth.
The data come from a national probability sample of individuals who were interviewed in 1987-1988 and again in 1992-1994. The investigators limited their sample to non-Hispanic white adults aged 16-39 who were fertile and nonpregnant (or whose partners were fertile and nonpregnant). The final sample consisted of 2,812 persons.
Thirty-seven percent of the overall sample had a child in the years between the two interviews. For the 2,768 persons who provided data at the first interview on the certainty of their intentions (1,538 women and 1,230 men), 67% affirmed they intended to have a child sometime (with 28% being very sure of it, 26% moderately sure and 13% not sure). The remainder said they did not intend to have a child (with 14% being very sure of that intention, 14% moderately sure of it and 6% unsure). The proportion having a child declined steadily across the intentions continuum, from 61% among those who had been very sure they intended to have a child to 11% among those who had been very sure they did not intend to have a child.
To assess how individuals' intentions about having children contribute to their actual fertility, the investigators conducted three sets of logistic regression analyses that measured the monthly likelihood of a conception leading to a live birth (or the likelihood of a formal adoption) in the period between the surveys. The first set controlled for three variables related to fertility intentions and timing--respondents' fertility plans (including their degree of certainty about those plans), how closely respondents' plans matched their spouses' plans and whether they planned to have a child within four years. The second set controlled for eight social and demographic variables--marital status, age, parity, school enrollment status, employment status, educational level, the respondent's mother's educational level and income. The third analysis considered all 11 variables simultaneously. The investigators also examined the data separately by marital status at the time of the initial interview and by gender among those who were unmarried at the first interview.
The results of the logistic regression analyses revealed statistically significant associations between each fertility intention and childbearing, with the magnitude of the association decreasing along the fertility intentions continuum (from being very sure about wanting a child to being moderately sure about not wanting one, with the category "very sure of not intending to have a child" serving as the reference group). For instance, among respondents who were married at the time of the first interview, the analysis accounting for all background and fertility-related factors revealed that childbearing was 7.2 times as likely among those very sure they would have a child as among those very sure they would not; the odds ratios declined to 5.4 among those moderately sure of having a child, to 2.2 among those intending to have a child but with very little certainty and to 1.8 among those moderately sure of not having a child. (This effect of intentions on observed fertility was consistently smaller, and not always significant, among respondents who were unmarried at the time of the initial interview; odds ratios ranged from 4.6 to 2.2, depending on the certainty of intentions.)
The likelihood of a birth also varied significantly by spouses' relative intentions about childbearing. Respondents whose spouse was more sure about having a child were twice as likely to do so (odds ratio, 1.9) as were respondents who had the same fertility plans as their partner. Correspondingly, when a spouse was less sure about having a child than the respondent, that likelihood was significantly lowered (0.8). Respondents' expectations of the timing of an upcoming birth had a more modest effect. Expecting to have a child within four years raised the likelihood of a birth in the first 29 months since the initial interview by 40%. Moreover, respondents who were no longer married by the time of their second interview were roughly 60% less likely than those who remained married to have had a child between the surveys.
In general, having already had a child significantly raised the likelihood of having another one, with fluctuations by the amount of time that had gone by since the most recent birth. For example, married respondents with three children whose last child had been born fewer than three years earlier were more than twice as likely as those who had never had a child to conceive between surveys (odds ratio, 2.2); however, parents of three children whose last child had been born 3-6 years ago were only half as likely as those who had no children to conceive between the surveys (odds ratio, 0.5).
The single background variable that had an even stronger effect on actual fertility than the respondents' intentions to have children, at least among men, was marriage. In an analysis based on respondents who were unmarried at the earlier interview, when all factors were controlled for, men who subsequently married were more than 10 times as likely as those who were still unmarried to have fathered a child (odds ratio, 10.4). Among initially unmarried women, marriage significantly raised the likelihood of conceiving a child by a factor of six.
The likelihood of a birth was significantly elevated among initially unmarried female respondents who intended to have a child (3.1-3.8); odds were also elevated among initially unmarried women if they had had one or two children 3-6 years ago (1.7-2.8) or if they had had three children more than six years before their last child (4.7). Moreover, the timing of an expected birth had a moderate effect on the likelihood of a conception among these women, as those who said they expected to have a baby within four years were 2.3 times as likely as those who did not to have conceived in the first 29 months after their initial interview. However, the odds of a birth between surveys among women who were unmarried at their first interview were significantly lower if they had had some college education or had graduated from college than if they had had a high school education only (odds ratios of 0.6 and 0.5, respectively).
Among men who were unmarried at the time of their first interview, those who were very or moderately sure they intended to have a child, as well as those who were unsure about not intending to do so, had significantly elevated odds of fathering a child compared with men who were very sure they did not intend to have a child. Men younger than 25, those whose only child had been born 3-6 years ago and those who had not completed high school all had elevated odds of fathering a child (1.7-2.8), while male respondents who had a partial or complete college education had significantly reduced odds of doing so (0.4 and 0.5, respectively).
According to the investigators, the only life-course variable that approached fertility intentions in importance as a predictor of fertility behavior was marital status. The researchers suggest that intentions were less predictive of subsequent fertility among unmarried than married respondents because these intentions, which may be contingent on marriage, were expressed without the knowledge of a partner's preferences. However, even the fertility intentions of unmarried persons independently predicted whether they would have a child. Therefore, the investigators conclude that "recognizing the predictive power of fertility intentions would encourage a healthy redirection of fertility research toward the dynamic interaction between the individual and society."--L. Remez
1. Schoen R et al., Do fertility intentions affect fertility behavior? Journal of Marriage and the Family, 1999, 61(3):790-799.