Delaying a Family: Determinants and Characteristics of Temporary Childlessness in the US

William Mosher, National Center for Health Statistics, Centers for Disease Control and Prevention
Gladys M. Martinez, National Center for Health Statistics, Centers for Disease Control and Prevention

Most childlessness in the United States is thought to be temporary: that is, most women who are currently childless expect to have 1 or more children at some time in their lives. The popular media debate frequently whether the expectations of “temporarily” childless women are realistic, while sources such as the CPS and Census tell us relatively little about the context in which these expectations are reported. Yet it is possible to learn much more about temporary childlessness in the US from the 1995 National Survey of Family Growth (NSFG)—a large, reliable national sample, based on detailed in-person interviews with 10, 847 women 15-44 years of age. In this paper, we use the NSFG to look in detail at women 18-39 years of age who had no births but expected 1 or more at some time in the future. This analysis will emphasize data items that are not in most other data sources. These data will also allow us to assess the context and probable quality of the birth expectations of these women: how many births these women will have in the future and when. The NSFG’s response rate in 1995 was 79 percent and the average interview lasted 100 minutes. (W. Mosher, 1998) This very rich set of individual-level data was enhanced in recent years with a contextual data file, containing data on a very wide range of characteristics of the block groups, census tracts, counties, and states in which the respondents lived in 1990 (at the census date), in 1993, and in 1995 (at the interview date). The analyses in this abstract use the contextual data for the address in 1990, to make certain that the contextual variables were measured before the interview. While some studies have attempted to project or forecast permanent childlessness in the US and other developed countries, and a few have profiled the relatively small proportion who expect to remain childless---the voluntarily and involuntarily childless (Mosher and Bachrach, 1982; Abma and Martinez, forthcoming)—very few studies have attempted to shed light on the large and demographically important group of temporarily childless women in the US. They are expecting millions of births: will they have them? If so, when? As an outcome measure, we can look at more than the responses to a single question on births expected in her lifetime: we can look at the number of births she expects to have in her lifetime, the number she expects in the next 5 years, how sure she is about those expectations, and whether her husband or partner disagrees with her intentions. This analysis looks at 5 categories of determinants and characteristics, using cross-tabulation, logistic regression, and Hierarchical Linear Modeling. (Raudenbush and Bryk, 2001; Kreft and De Leeuw, 1998). First, we look at the woman’s individual and family history: whether she comes from an intact, 2-parent family; her mother’s educational attainment; the religion in which she was raised, and her own educational and work history. We also run the analyses separately for women 18-29 and 30-39 years of age. Second, we look at her relationship status, starting with the following categories: (1) never married and never cohabited with a man; (2) ever married or ever cohabited but not currently; (3) currently cohabiting; and (4) currently legally married. For those who are not currently married, we look at whether they expect to marry. Third, we look at current factors that may often affect childbearing intentions: if she is married, whether her marriage is her first marriage or a remarriage; whether she has adopted children or step children; her current household income; her current labor force status (working full-time, working part-time, going to school, or other); attitudes toward marriage, the family, and children; and how important religion is in her life (very, somewhat, not important). Fourth, we look at fertility prospects: Her current contraceptive method, if any, and how consistently it is used; her report of her fecundity and infertility status; whether she (or her partner) has ever sought medical help for infertility; and how frequently she is having intercourse. Fifth, we look at contextual variables. There is recent evidence that contextual variables do affect temporary childlessness. A recent analysis (Mosher, Deang, and Bramlett, forthcoming 2002) based on the NSFG Contextual data file shows that overall, about 34 percent of women 18-39 are childless but expect to have 1 or more children eventually: 35 percent of white women, 28 percent of black women, and 29 percent of Hispanic women. That analysis showed that the prevalence of temporary childlessness varies strongly by certain contextual characteristics, for Hispanic, white, and black women. For example, Figure 1 shows that, in block groups with low median family incomes (the lowest third), 26 percent of women were temporarily childless; in block groups with high median family incomes (the highest third), 46 percent were temporarily childless. This same pattern was found for Hispanic women (24 and 45 percent temporarily childless); for white women (28 and 46 percent temporarily childless); and for black women (25 and 48 percent temporarily childless). Equally striking, however, is that the percent temporarily childless does not differ significantly by race/ethnicity (i.e., for Hispanic, white, and black women) within each category of neighborhood-level median family income. This suggests that once we control for neigborhood-level SES, there are no differences by race/ethnicity in fertility intentions. Similar patterns are found by other contextual measures of high SES, such as the percent of adults who are college graduates, and the percent of employed adults who work in professional or managerial jobs. Further, a multivariate, Hierarchical Linear Model (using the SAS procedure PROC MIXED and the GLIMMIX macro) for a similar dependent variable suggests that a contextual (county-level) measure of median family income is independently associated with current childlessness, controlling for age, race, Hispanic origin, and individual-level household income. (Mosher, Deang, and Bramlett, forthcoming, table 3). Finally, we point out that, in approximately 2 years, it will be possible to use data from the 2002 NSFG to update and enhance the analysis. REFERENCES J Abma and G Martinez. 2002. Childlessness Among Older Women in the U.S.: Trends, Work Experience, and Gender Role Attitudes Paper presented at the Annual Conference of the Population Association of America, Atlanta, Georgia. I Kreft and J DeLeeuw. 1998. Introducing Multilevel Modeling. Thousand Oaks, CA. Sage Publications. W Mosher. 1998. Design and operation of the 1995 National Survey of Family Growth. Family Planning Perspectives 30 (1): 43-46. W Mosher and C Bachrach. 1982. Childlessness in the United States: Estimates from the National Survey of Family Growth. Journal of Family Issues 3 (4): 517-543, December 1982. W Mosher, L Deang, and M Bramlett. 2002 (or early 2003). Community Environment and Its Relationship to family size, contraception, and selected measures of women’s and infants health. Vital and Health Statistics, Series 23, No. 23, forthcoming about January of 2003. SW Raudenbush and AS Bryk. 2001. Hierarchical Linear Models: Applications and Data Analysis Methods, Second Edition. Thousand Oaks, CA: Sage Publications.