Introduction

Fertility rates in the United States have reached record low levels of 1728.0 births per 1000 women in 2018 (Hamilton et al. 2019). In light of these historical declines, recent research has explored the potential causes of low fertility in the United States, including parenthood postponement and the impact of the economic recession (Matthews and Hamilton 2016; Schneider 2015). However, an alternative explanation for this trend: a potential increase in intended childlessness, has received less attention. Is it possible that Americans, instead of postponing parenthood, are rather increasingly developing intentions to remain childless (hereafter ITRC)?

Existing empirical evidence on women indicates that ITRC are correlated with subsequent childlessness (Berrington 2004; Heaton et al. 1999; Rovi 1994). For instance, childless women who reported ITRC at age 24 were 4.5 times more likely to remain childless at age 45, and the association between childlessness intentions and permanent childlessness grew even stronger as women got older (Rybińska and Morgan 2019). An increase in the prevalence of ITRC at the population level could thus be associated with future increases in rates of permanent childlessness and further declines in fertility rates in the United States.

Reports of ITRC were rare in the United States in the twentieth century (Hagewen and Morgan 2005; Schoen et al. 1997). In 1998, 7.8% of women aged 18–39 reported intentions to remain childless, per authors’ own calculations using the Current Population Survey (Ruggles et al. 2017). Research from Europe has however documented an increase in childlessness intentions in the early twenty-first century in 12 countries, including the United Kingdom (Miettinen and Szalma 2014). Given the recent increases in ITRC in similar low-fertility contexts in Europe, in this research note we examine whether childlessness intentions might have also increased in the United States over the past two decades.

Importantly, men’s childbearing and childlessness intentions are largely omitted from childlessness studies (Blackstone and Stewart 2012) as well as childbearing studies at large (Almeling 2015). At the same time, most children in the United States are born to co-residing parents (Bumpass and Raley 1995; Bumpass and Lu 2000; Kennedy and Bumpass 2008) who make a joint decision about parenthood (for the United States, see e.g., Thomson et al. 1990; Thomson 1997). Consequently, both partners’ childbearing intentions might influence the decision about pregnancy and birth. A disagreement about parenthood between partners can lead to further childbearing delay, or even a decision to abandon plans for parenthood (for Italy: Testa et al. 2014). An additional contribution of our study is thus that we provide a rare look at childlessness preferences of American men.

Rising Prevalence of Intentions to Remain Childless: Explanations

Scholars who stress the impact of ideological changes on fertility argue that, increasingly, childbearing behavior is motivated by new ideas that place the individual and individual choice at the center of the unfolding life course (Lesthaeghe and van de Kaa 1986; van de Kaa 1987). Parenthood remains an important component of a meaningful life but in a context in which the decision to parent is optional, other life choices can also contribute to individual self-actualization (Giddens 1991). Simultaneously, social acceptance of intended childlessness is on the rise (Thorton and Young-DeMarco 2001; Chancey and Dumais 2009). In addition, costs of childrearing—in terms of financial resources as well as time investments—are increasing (Gauthier et al. 2004; Kornrich and Furstenberg 2013) which might lead young adults to perceive parenthood as more costly than the generation of their parents. Taken together, these cultural and economic changes might challenge the centrality of parenthood for individual life satisfaction. In line with these arguments, we hypothesize that ITRC are on the rise in the United States.

However, the prevalence of ITRC might also be affected by compositional changes in the structure of the United States’ population. ITRC are more prevalent among specific groups of women, such as older women, women with higher educational attainment, and never-married single women (Abma and Martinez 2006). In the United States, levels of college education and singlehood are rising (Ryan and Siebens 2012; Traister 2016). Thus, increases in the shares of these specific populations in the general population might explain increases in ITRC. Compositional changes in the racial/ethnic profile of the U.S. population could in contrast result in declines in ITRC, because minority women tend to have lower levels of ITRC than non-Hispanic white women (Lundquist et al. 2009; Sweeney and Raley 2014). Accordingly, the analyses will account for the age, race/ethnicity, educational attainment, and partnership status of the individuals in the sample.

Very few studies to date have specifically examined the interdependence of socio-demographic correlates and childlessness intentions among men (notable exceptions include Jacobson and Heaton 1991; Lunneborg 1999). However, previous research has documented gendered differences with respect to socio-demographic correlates of reproductive experiences (e.g., with respect to education in Norway: Kravdal and Rindfuss 2008). Thus, while our examination of men’s ITRC is mainly exploratory, we tentatively expect to observe some variation in factors associated with ITRC between men and women.

Data and Methods

We use pooled data from the National Survey of Family Growth (NSFG) cycles 2002, 2006–2010, 2011–2013, 2013–2015, and 2015–2017 to investigate the population-level changes in ITRC among American women and men. The NSFG is a repeated cross-sectional survey that is representative of the civilian, non-institutionalized population aged 15–44. A previous publication on this topic that focused on the United States by Hagewen and Morgan (2005) used the Current Population Survey (CPS, Ruggles et al. 2017), but the measure of childbearing intentions was discontinued in this survey in 1998. As a sensitivity test, we compare the NSFG trends for women with previous data from the CPS in the online Appendix.

We selected a sample of 31,739 women and 24,524 men aged 18–44 with complete data on the pertinent covariates (we excluded 630 records for women and 273 records for men due to missing data and imputations). If the respondents had no biological, adopted, or step-children, and reported no intentions of having children in the future, they were classified as reporting intentions to remain childless (ITRC). The respondents who already had children, or who were childless but said they intend to have children in the future, were assigned to the comparison group. We also classified uncertainty about future childbearing as not intending to remain childless. We believe that this approach generated a conservative estimate of childlessness intentions.

We present our results separately for men and women. First, we discuss estimates for the population aged 18–44, and then present multivariate logistic regression analyses accounting for age, race/ethnicity, partnerships status, and educational attainment. Race/ethnicity categories include non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic other. We distinguish between respondents with less than 12 years of education (equivalent to less than a high school degree), 12 years of education (equivalent to a high school degree), 13–15 years of education (equivalent to completing some post-secondary education), and 16 or more years of education (equivalent to completing a bachelor’s degree or higher degrees). Partnership status categories include married, cohabiting, separated single (due to the small sample size, this category includes all women who are divorced, separated, or widowed), and never-married single. All analyses account for the NSFG survey design. Descriptive statistics for the overall sample as well as by NSFG cycle are presented in the online Appendix.

Findings

At the population level, the prevalence of ITRC among women aged 18–44 increased from 8.3% in the 2002 cycle to 10.3% in the 2015–2017 cycle (Fig. 1). Bivariate logistic regression estimates indicate that this increase was statistically significant (Table 1, Model 1). However, in a multivariate context (Table 1, Model 2) the increase in the prevalence of ITRC among women is shown to be no longer statistically significant. Further analyses indicate that increases in the proportion of single women account for the association between NSFG survey cycles and ITRC (see the online Appendix).

Fig. 1
figure 1

Proportions of women and men aged 18–44 who intend to remain childless in the United States. National Survey of Family Growth cycles from 2002 to 2015–2017. Letters designate statistically significant difference at p-value <0.05 between the specific NSFG cycle and the reference cycle of: a2002; b2006–2010; c2011–2013; d2013–2015

Full size image
Table 1 Logistic regression estimates for intentions to remain childless, samples of women and men aged 18–44 in the United States. National Survey of Family Growth cycles from 2002 to 2015–2017
Full size table

For men, we observe a U-shaped tendency in ITRC: the prevalence of childlessness intentions declined from 11.8% in the 2002 cycle to 9.8% in the 2006–2010 cycle but increased steadily thereafter to 13.9% in the 2015–2017 cycle (Fig. 1). Because of this trend, we used the 2006–2010 cycle as a reference category for logistic regression modeling for men. In both bivariate and multivariate settings (Table 1, Models 3 and 4), we detect a dip in the odds of reporting ITRC between the 2002 and 2006–2010 cycles and an increase in the odds of reporting ITRC between the 2006–2010 and 2015–2017 cycles. In other words, in contrast to the findings for women, the increase in ITRC across NSFG cycles among men is not associated with changes in the socio-demographic composition of the population.

Models 2 and 4 show differences in ITRC across socio-demographic groups including a positive age gradient in ITRC and lower odds of reporting ITRC among minority respondents (compared to non-Hispanic White respondents). In addition, cohabiting, never-married single, and separated single respondents have higher odds of reporting ITRC compared to married respondents. These associations are observed across the subpopulations of men and women. However, educational differences are only present for the population of women: women with 16 or more years of education have higher odds of reporting ITRC compared to women with 12 years of education.

In the next step, temporal changes in ITRC within the socio-demographic subgroups are examined using marginal effects sourced from Model 2 for women and from Model 4 for men. We use the 2002 survey cycle as our reference point for describing the findings in this paragraph. There are however several non-linear associations, especially among men, which are marked with additional notations in Figs. 2, 3 and 4.

Fig. 2
figure 2

Marginal predicted probabilities of intentions to remain childless among women and men aged 18–44 in the United States by survey cycle and age group. National Survey of Family Growth cycles from 2002 to 2015–2017. Predicted probabilities represent marginal effects from a multivariate logistic regression including age, race/ethnicity, educational attainment, partnership status, and survey cycle indicators. Letters designate statistically significant difference at p-value <0.05 between the specific NSFG cycle and the reference cycle of: a2002; b2006–2010; c2011–2013; d2013–2015

Full size image
Fig. 3
figure 3

Marginal predicted probabilities of intentions to remain childless among women and men aged 18–44 in the United States by survey cycle and educational attainment. National Survey of Family Growth cycles from 2002 to 2015–2017. Predicted probabilities represent marginal effects from a multivariate logistic regression including age, race/ethnicity, educational attainment, partnership status, and survey cycle indicators. Letters designate statistically significant difference at p-value <0.05 between the specific NSFG cycle and the reference cycle of: a2002; b2006–2010; c2011–2013; d2013–2015

Full size image
Fig. 4
figure 4

Marginal predicted probabilities of intentions to remain childless among women and men aged 18–44 in the United States by survey cycle and partnership status. National Survey of Family Growth cycles from 2002 to 2015–2017. Predicted probabilities represent marginal effects from a multivariate logistic regression including age, race/ethnicity, educational attainment, partnership status, and survey cycle indicators. Letters designate statistically significant difference at p-value <0.05 between the specific NSFG cycle and the reference cycle of: a2002; b2006–2010; c2011–2013; d2013–2015

Full size image

We document a distinct upward trend in the probability of reporting ITRC among women and men aged 18–24: from 4.5% in the 2002 cycle to 7.8% in the 2015–2017 cycle for women and respectively from 5.8 to 8.8% for men (Fig. 2). We also observe increases in the probability of reporting ITRC among men aged 25–29 and declines among men aged 35–39. The findings further indicate that the probability of reporting ITRC increased among men with 13–15 years of education (Fig. 3). Finally, we observe increases in reporting ITRC for never-married single women (from 17.1% in the 2002 cycle to 21.9% in the 2013–2015 cycle) and for separated single men (from 6.2% in the 2002 cycle to 11.9% in the 2015–2017 cycle, Fig. 4). Because we detected no changes in the temporal ITRC trends within race/ethnicity subpopulations (in comparison to the 2002 NSFG cycle), we omitted these findings from the manuscript, and included them in the online Appendix.

Conclusion and Discussion

The purpose of this paper was to identify population-level trends in the prevalence of intentions to remain childless (ITRC) in the United States in the early decades of the twenty-first century. We used the National Survey of Family Growth cycles from 2002 to 2015–2017 to investigate ITRC among men and women aged 18–44.

We found increases in the prevalence of ITRC among women aged 18–44 from 8.3% in the 2002 survey cycle to 10.3% in the 2015–2017 survey cycle. To our knowledge, this study is the first to report rises in preferences regarding childlessness among women in the United States. The prevalence of ITRC among American men followed a U-shaped trend with a decline from 11.8% in the 2002 NSFG cycle to 9.8% in the 2006–2010 cycle and an increase to 13.9% in the 2015–2017 cycle. Because intentions to remain childless correlate with subsequent childlessness (Rovi 1994; Rybińska and Morgan 2019), increases in ITRC might indicate that childlessness rates will increase in the coming decades, further depressing fertility rates in this country.

In addition, distinct increases in ITRC are observed among young women (aged 18–24) and men (aged 18–24 and 25–29). As these young Americans are in the middle of exploring different life choices and of drawing up plans for adulthood, they might be especially attuned to current social norms regarding the importance of parenthood and acceptance of childlessness, as well as to the perceived costs of childrearing. Indeed, young Americans have recently verbalized concerns about the financial burdens of parenthood (Brinton et al. 2018). Are worries about the costs of raising children driving young Americans to choose childlessness? Future studies could explore the spread of intentions to remain childless among young adults with connection to social norms and economic conditions.

Finally, benefiting from data on ITRC among men, in this study, we uncover interesting differences in childlessness intentions among men and women. First, increases in the prevalence of ITRC among women are connected to the compositional changes in the socio-demographic profile of the population. Specifically, increases in the proportion of single women account for the higher prevalence in ITRC in the twenty-first century. However, the increases in the prevalence of ITRC among men are irrespective of compositional population changes. Second, while we observe a positive association between educational attainment and ITRC among women, we do not detect this education gradient in childlessness intentions among men.

These associations suggest that tensions between professional and family responsibilities might continue to play an important role in shaping permanent childlessness trends among women but not among men in the United States. Over the past decades, tensions between private and public social roles for women have increased due to women’s entry into the labor force (Gerson 1985; Hochschild 1989) and research indicates that the created work-family life conflict has disproportionately affected women with higher educational attainment (Blain-Loy 2005). Meanwhile, the expectations for men have changed to a lesser extent (Gerson 2010). Thus, women, especially college educated women, might continue to face a trade-off between education/professional career and marriage/parenthood while men are less affected by such considerations.

Given the observed similarities and differences in the ITRC trends and correlates between men and women, we encourage future empirical research into factors contributing to gendered socio–demographic variations in ITRC and ITRC development. Emerging theoretical approaches to men’s roles in biological and social processes of reproduction (Almeling 2015) offer compelling frameworks for such research. However, active data collection efforts are also necessary to fill the gap in the existing knowledge about men’s childbearing intentions. Measures of reproductive intentions could be for instance introduced to ongoing longitudinal social surveys, such as the 1997 National Longitudinal Survey of Youth, to complement information from cross-sectional sources.