Abstract
Objective:
This study documented change in the midlife first marriage rate for U.S. adults aged 40-59 between 1990 and 2019 and assessed the sociodemographic correlates of midlife first marriage formation for today’s women and men.
Background:
Median ages at first marriage are at record highs for women and men, signaling that marriage may be increasingly occurring at older ages. However, first marriage formation among midlife adults remains largely overlooked.
Method:
Data from the 1990 U.S. Vital Statistics and the 2010 and 2019 American Community Survey (ACS) were used to estimate change in women’s and men’s first marriage rates across age groups, with a focus on how the rate has changed for midlife adults. Average marginal effects (AMEs) were derived from logistic regression analyses that drew on the 2019 ACS to examine associations between sociodemographic factors and midlife first marriage formation for women and men.
Results:
Since 1990, the midlife first marriage rate has increased by 75% for women and 45% for men. The shares of women and men entering a first marriage who were aged 40-59 quadrupled between 1990 and 2019 (rising from 2% to 9% among women and from 3% to 12% among men). Some of the well-established predictors of first marriage in young adulthood operated uniquely for first marriage formation in midlife.
Conclusion:
Future research on first marriage formation should incorporate midlife adults.
Keywords: gender, marriage delay, middle-aged, union formation
Historically, the likelihood of marrying has declined precipitously with age, prompting scholarly speculation that record high levels of age at first marriage for men and women in the U.S. may foretell a retreat from marriage (Sassler & Lichter, 2020; Smock & Schwartz, 2020). Recent empirical research demonstrates that fewer young adults are getting married (Bloome & Ang, 2020), but the trend for midlife adults remains largely overlooked, a notable omission in an era of rising age at first marriage which may now stretch beyond young adulthood. Contemporary studies of first marriage formation rarely stray beyond the young adult stage of life (see Sassler & Lichter, 2020 for a summary), relying on upper age limits for marriage entry hovering around 40 to 45 (Bloome & Ang, 2020; Martin, Astone, & Peters, 2014). This approach is becoming less tenable given the steady ascent in the median age at first marriage, a signal that marriage is increasingly occurring at older ages.
Consequently, we shift the focus to first marriages that occur in midlife, a developmental phase of adulthood that begins at age 40 and ends by age 60 (Lachman, 2004). By evaluating change in the midlife first marriage rate, a measure of the number of first marriages in the past year per 1,000 never-married midlife adults, we assess whether the chances of entering marriage for the first time during midlife have increased over the past three decades. The climbing median age at first marriage suggests that first marriage entry is increasingly delayed, underscoring the utility of investigating midlife first marriage formation.
Combining U.S. data from the 1990 Vital Statistics and the 2010 and 2019 American Community Survey (ACS), we estimate change in women’s and men’s first marriage rates across age groups, with an emphasis on how the rate has changed for midlife adults. Additionally, we draw on the 2019 ACS to examine the sociodemographic correlates of midlife first marriage formation for women and men. By shedding new light on the scale and scope of first marriage entry during midlife, the findings from our study elucidate how the shifting timing and tempo of first marriage formation increasingly extends beyond the boundaries of young adulthood.
Background
Over the past half century, the U.S. first marriage rate has plummeted 60%, falling from 76.5 marriages per 1,000 unmarried women ages 15 and older in 1970 to just 31.3 per 1,000 in 2018 (Schweizer, 2020). The share of women ages 18-49 who have ever married dropped from 72% to just 57% between 1995 and 2020. For women ages 45-49, a decline is evident but comparatively modest: 93% were ever-married in 1995 versus 86% in 2020 (Carlson, 2020). The age distribution of first marriages has flattened considerably since 1960, reflecting rising rates of marriage beyond age 30 and falling rates for those under 30 (Eickmeyer, 2019). Roughly 49% of men and 40% of women are over age 30 at first marriage entry nowadays (Hemez, 2020).
The median age at first marriage has risen steadily from a low in the 1950s of about 20 for women and 22 for men to 28.1 for women and 30.5 for men in 2020 (Payne, 2021). The pace of the rise in the median age at first marriage has been accelerating, signaling more pronounced delays in marriage entry going forward. For women born in 1997, for example, only half are projected to marry by age 37.4 (Bloome & Ang, 2020), meaning roughly half will enter midlife never-married. This trend underlines the urgency of shifting the lens on first marriage formation toward midlife.
The institution of marriage is undergoing rapid change in ways that are likely to shape both the timing and tempo of marriage entry. The role of contemporary marriage in the life course is shifting as its practical significance wanes even as its symbolic value climbs. The blurring of spousal roles and decreased gendered specialization is one example of how the practical significance of marriage has declined. Another example is the viable alternatives to marriage, including cohabitation and singlehood. At the same time, the symbolic significance of marriage has soared. Marriage is now a capstone event that occurs once individuals have completed their education, secured a stable job, minimized their debt, and achieved some semblance of financial security, which increasingly may extend beyond young adulthood. The lengthening and winding road to adulthood may culminate in a more protracted path to marriage, proceeding more slowly than previous generations in terms of the timing of marriage entry (Addo, 2014; Arnett, 2014; Schneider, Harknett, & Stimpson, 2019).
The Current Study
Those who have not married during young adulthood should be increasingly likely to marry for the first time during midlife. A rising midlife first marriage rate aligns with the notion that Americans are delaying marriage entry, pushing it back to later ages as they labor to achieve the economic stability perceived as a prerequisite for marriage. The high economic bar for marriage coupled with its symbolic significance foretells pent-up demand for marriage in midlife that would be evidenced by a growing midlife first marriage rate. Alternatively, as an engine of inequality, marriage is increasingly out of reach for many, which would be indicated by a stagnant or declining midlife marriage rate. This pattern would be in line with both the declining centrality of marriage in family life and the growing economic divergence in marriage, which seems to be more often achieved by only the college-educated.
Despite the well-documented rise in the age at first marriage, to our knowledge researchers have not investigated the factors linked to first marriage formation during midlife. For this reason, we draw on the extensive literature on first marriage formation in general to guide our exploration of the correlates of first marriage entry during midlife, emphasizing demographic characteristics, economic resources, and disability status. Demographic factors include age, race-ethnicity, and U.S.- versus foreign-born status. For midlife adults, age should be negatively associated with first marriage entry. The racial-ethnic divide in marriage entry has widened in recent years with Blacks increasingly less likely to marry than Whites (Bloome & Ang, 2020; Raley, Sweeney, & Wondra, 2015). Marriage rates are higher, on average, among immigrant than U.S.-born adults (Brown, Van Hook, & Glick, 2008). Economic resources are key predictors of marriageability and include education, employment, and income (Sassler & Lichter, 2020; Sweeney, 2002). For both women and men alike, these resources should be positively associated with first marriage entry among midlife adults. Among young adults, persons with a disability are less likely to marry than their counterparts who do not have a disability (MacInnes, 2011). Still, whether the correlates of first marriage formation that are well-established among young adults hold for those in midlife remains uncertain. Midlife first marriage is an off-time event and thus it is possible that some of the covariates operate uniquely for this group. Although an in-depth examination of this possibility is beyond the scope of this brief report, we do conduct supplemental analyses to explore whether and how the correlates of first marriage formation differ for young adults versus those in midlife.
The processes undergirding decisions to marry are likely to vary between women and men and this gender gap arguably grows more pronounced with age as women exhibit higher rates of marriage than men at younger ages (Hemez, 2020), potentially resulting in disparate groups of never-married women and men in midlife. Similarly, the pool of eligible marriage partners diverges with age as men’s pool grows (they tend to marry younger women) while women’s pool shrinks (they tend to marry older men). Thus, we estimate age-specific first marriage rates and midlife first marriage entry separately for women and men.
Method
We drew on two data sources to document how the first marriage rate has changed over the past three decades across age groups. Age-specific first marriage rates for women and men in 1990 came from the 1990 U.S. Vital Statistics. National marriage rate statistics are not available for 2000 as the federal government discontinued the collection of marriage (and divorce) statistics from the states in the mid-1990s. We also performed original analyses of the 2010 and 2019 American Community Survey (ACS) to estimate age-specific first marriage rates for women and men in 2010 and 2019, including a midlife first marriage rate for those aged 40-59. Finally, we relied on the 2019 ACS to examine sociodemographic subgroup variation in the midlife first marriage rate as well as the correlates of midlife first marriage for women and men to determine which of today’s midlife never-married individuals are most likely to wed. Supplemental analyses were conducted to assess whether the associations between the covariates and first marriage differed for midlife versus young adults.
1990 U.S. Vital Statistics Report
The 1990 U.S. Vital Statistics Report included both the marriage rates and the number of persons who married in 1990 by 5-year age intervals for women and for men. This information was from the Marriage-Registration Area (MRA) sample of the District of Columbia and all states except Arizona, Arkansas, Nevada, New Mexico, North Dakota, Oklahoma, Texas, and Washington. The MRA was representative of 77% of marriages formed in the U.S. in 1990 and is the best available source of data on age-specific marriage rates for the time period. According to a state-level validation analysis performed by the Census Bureau, estimates produced by the U.S. Vital Statistics and the ACS were comparable (Elliott, Simmons, & Lewis, 2010). Prior to 1990, midlife first marriage was so unusual that the Vital Statistics did not produce estimates that would allow us to calculate the rate for 1980 or earlier decades (Clarke, 1995).
Age-specific First Marriage Rate.
To calculate the first marriage rate, we began by dividing the number of first married persons by the first marriage rate to obtain the number of persons at risk of first marriage. Summing the numbers who first married and the numbers at risk across age intervals by gender as appropriate and then dividing the numbers first married by the numbers at risk yielded the 1990 first marriage rates by ten-year age intervals, ranging from 18-29, 30-39, 40-49, to 50-59, along with the midlife first marriage rate for those aged 40-59.
2010 and 2019 ACS
The ACS is a nationally representative annual survey designed to gather information previously obtained through the census long-form sample, including demographic, economic, and social characteristics (U.S. Census Bureau, 2009). In 2008, the ACS began including marital history questions in response to the lack of other data on the incidence of marriage and divorce (Elliott et al., 2010). We used data from the 2010 and 2019 ACS to chart age-specific first marriage rates across ten-year age intervals (18-29, 30-39, 40-49, and 50-59). Also, we estimated a midlife first marriage rate for adults aged 40-59. In 2010, the ACS sample of 3,061,692 individuals included 879,717 people aged 40-59, of which 111,358 were at risk of first marriage (as defined in the next paragraph), including 49,400 women and 61,958 men. The 2019 ACS sample of 3,239,553 individuals included 825,574 people aged 40-59, of which 123,722 were at risk of first marriage, including 54,041 women and 69,681 men.
Age-specific First Marriage Rate.
The ACS asked respondents whether they got married in the past 12 months and the number of times they have been married. To calculate the first marriage rate for a particular age group, we divided the number of people who reported a first marriage in the past 12 months by the number at risk of a first marriage during the past 12 months. Those at risk of a first marriage included individuals who either remained never-married or married for the first time during the past 12 months. The ACS is the best data source for estimating the first marriage rate (Ratcliffe, Acs, Dore, & Moskowitz, 2008).
Correlates of Midlife First Marriage.
Demographic characteristics, economic resources, and disability status are related to marriage formation and are measured in the ACS. Demographic characteristics included age, race-ethnicity, and birthplace. For this portion of the analysis, age was coded in five-year intervals: 40-44 (reference category), 45-49, 50-54, and 55-59. This approach permitted a more fine-grained examination of the role of age among midlife adults (patterns were analogous when using ten-year age intervals of 40-49 and 50-59). Race/ethnicity distinguished among non-Hispanic Black, Hispanic, non-Hispanic Other (which included multiracial individuals as well as those who identified as a single race: Asian, American Indian or Alaska Native, Native Hawaiian and Pacific Islander, or some other race), and non-Hispanic White (reference category). U.S.-born was coded 1 for those born in the U.S. and 0 for individuals born outside of the U.S. (reference category). Economic resources included education, employment, and personal income. Education was categorized as less than a high school diploma, a high school diploma (reference category), some college, and a college degree or more. Employment captured employed full-time (at least 35 hours per week), employed part-time (1-34 hours per week), unemployed, and not in the labor force (reference category) during the past 12 months. Personal income measured the individual’s income over the past year from all sources and was categorized as less than $10,000, $10,000-$24,999, $25,000-$39,999, $40,000-$54,999, $55,000-$69,999, and $70,000 or more (reference category). Disability was coded 1 if respondents reported experiencing any of the following difficulties: hearing; seeing; concentrating, remembering, or making decisions; walking or climbing stairs; dressing or bathing; or doing errands alone and 0 otherwise.
Analytic Strategy
Our goals are to provide trend data and examine the correlates of midlife first marriage and thus our approach is primarily descriptive. We began by charting the age-specific first marriage rate by ten-year age intervals separately for women and men to assess how the first marriage rate changed over the past few decades. In supplemental analyses, we tracked the trend in the midlife (aged 40-59) first marriage rate and we also charted the changing age distribution of individuals entering first marriages in 1990, 2010, and 2019 to illustrate the growth in the share of persons entering a first marriage who are doing so in midlife.
Then, we performed a series of analyses of the 2019 ACS data to illustrate patterns of first marriage entry among today’s midlife adults. We began by tabulating bivariate statistics across demographic characteristics, economic resources, and disability status for those who entered a midlife first marriage versus remained never-married to facilitate comparison of the two groups. A supplemental table depicts the 2019 midlife first marriage rate for various subgroups to assess how the likelihood of entering a first marriage in midlife differed across sociodemographic indicators. Next, we estimated logistic regression models to determine how demographic characteristics, economic resources, and disability status were related to forming a first marriage in the past 12 months (coded 1) versus remaining never married (coded 0) separately for midlife women and men. To facilitate comparisons of coefficients by gender, we calculated (and show in the tables) average marginal effects (AMEs), which statistically account for the fact that women and men may have different residual variances (Breen, Karlson, & Holm, 2018). In supplemental analyses, we combined both young (aged 18-39) and midlife (aged 40-59) adults to assess whether the associations between the covariates and first marriage entry differed by life stage for women and men. And, a supplemental table shows the AMEs from multivariable logistic regression models predicting first marriage entry among young adults, which were performed for comparative purposes. All AMEs were computed at the observed values of the covariates for each observation. Mize, Doan, and Long (2019) recommended using seemingly unrelated estimation (the SUEST command in Stata) to compare AMEs across groups (e.g., women versus men or young versus midlife adults). However, the SUEST command in Stata does not support the jackknife method of variance estimation required when analyzing ACS data. Thus, we relied on the GSEM command in Stata (Stata Technical Support, personal communication, May 11, 2022). Our models yield correlational evidence only. In the ACS, first marriage entry occurred during the past 12 months, but the sociodemographic characteristics were measured at interview. All analyses of the 2019 ACS were conducted using replicate weighting techniques to generate robust standard errors because the ACS uses a complex sampling design (U.S. Census Bureau, 2009).
Results
Trends in Age-specific First Marriage
As shown in Figures 1a and 1b, the first marriage rate is now less age-graded than it was in 1990. For women (Figure 1a), the first marriage rate declined linearly with age in 1990. Among men (Figure 1b), the 1990 first marriage rate exhibited similarly high levels for those aged 18-29 and 30-39 followed by a sizeable drop off among men aged 40-49 that was even more pronounced for men aged 50-59. By 2010, the first marriage rate followed a curvilinear distribution across age groups, peaking among those aged 30-39 for women and men alike, which was consistent with the rising ages at first marriage during this period (Figures S1a and S1b). The flattening of the first marriage rate across age groups in 2010 also reflected a notable, sustained rise in the first marriage rate during midlife for both women and men which persisted through 2019. Strikingly, in 2019 the first marriage rate was nearly as high for men aged 40-49 (31.2 per 1,000) as men aged 18-29 (35.7 per 1,000), underscoring the growing prominence of midlife first marriage. In fact, the share of persons entering a first marriage that were in midlife (aged 40-59) quadrupled for both women and men between 1990 and 2019. For women, the shares rose from 2% in 1990 to 7% in 2010 and then to 9% in 2019 (Figure S1a). The pattern for men was similar, with the share at 3% in 1990, 10% in 2010, and 12% in 2019 (Figure S1b).
Figure 1a.
Women’s First Marriage Rate by Age Group 1990, 2010, and 2019
Figure 1b.
Men’s First Marriage Rate by Age Group 1990, 2010, and 2019
Within age groups, distinctive patterns of change unfolded over time. The first marriage rate declined precipitously among young women and men aged 18-29, falling from 86.5 to 52.3 to 46.3 per 1,000 and from 64.0 to 38.1 to 35.7 per 1,000, respectively, from 1990 to 2010 to 2019. For those in late young adulthood (aged 30-39), the first marriage rate trended differently for women versus men. For women aged 30-39, the first marriage rate ticked up slightly, from 59.9 in 1990 to 62.1 in 2010 to 65.2 in 2019. For men, the rate exhibited a curvilinear pattern, rising from 59.4 in 1990 to 64.9 in 2010 and then falling to 62.6 in 2019. Still, these shifts have been relatively modest and thus women and men in late young adulthood are not appreciably more likely to wed nowadays than they were a few decades ago. In contrast, the first marriage rate increased dramatically for midlife adults. This sharp upward trajectory in the midlife first marriage rate is evident across the full age spectrum of midlife with both the 40-49 and 50-59 age groups experiencing monotonic increases. Among adults in the 40-49 age range, the midlife first marriage rate nearly doubled for women, rising from 17.2 to 23.8 to 30.2 per 1,000 from 1990 to 2010 and 2019. For men aged 40-49, the rate increased almost 50%, climbing from 21.5 to 27.2 to 31.2 per 1,000 across 1990, 2010, and 2019. Notably, the rate of increase was greater for women than men and thus the groups experienced convergence over time. Whereas women’s rate lagged men’s by more than 4 points in 1990, by 2019 their rates were essentially the same. Among adults in the 50-59 age range, women and men experienced comparable gains over time. For 50-59 year old women, the rate more than doubled, rising from 6.2 in 1990 to 10.7 in 2010 to 15.0 per 1,000 in 2019. Among men aged 50-59, the first marriage rate nearly doubled, increasing from 8.6 in 1990 to 12.0 in 2010 to 16.9 per 1,000 in 2019. In short, the increase in the midlife first marriage rate since 1990 is not being driven solely by those in their forties. Even though the overall rates remained lower among those in their fifties than their forties, both groups experienced similar trajectories of growth in the midlife first marriage rate, underscoring the persistence of the trend across the entire age range. The midlife first marriage rate (aged 40-59) grew dramatically for women and men alike (Figure S2). For women, the midlife first marriage rate increased by about 75% between 1990 and 2019, rising from 13.66 in 1990 to 18.53 in 2010 to 23.81 per 1,000 in 2019. For men, the rate climbed from 17.34 to 21.36 in 2010 to 25.20 per 1,000 across 1990, 2010, and 2019, representing a 45% gain.
Correlates of Midlife First Marriage
Table 1 shows the characteristics of midlife individuals who got married for the first time in the past year compared with those who remained never-married, separately by gender. Across all demographic, economic, and disability indicators, those who got married significantly differed from those who did not, suggesting that midlife first marriage is a selective experience. The patterns were largely similar by gender and thus we describe them in concert as appropriate. Women and men who married in the past year were younger than individuals who did not marry. Among women, those who married were more often Hispanic (24.74% versus 18.71%) whereas those who remained never-married were disproportionately Black (29.63% versus 22.15%). Among men, the married were disproportionately Hispanic (24.07% versus 19.05%) and White men were overrepresented among the never married (53.07% versus 48.85%). Black and other race men were more similarly represented in the two groups. Among women and men who entered a midlife first marriage in the past year, nearly 30% were foreign-born. For those who remained never-married, only roughly 16% were foreign-born.
Table 1.
Percentage Distributions of Characteristics for Midlife Adults who Married for the First Time versus Remained Never-Married, by Gender
| Women |
Men |
|||||
|---|---|---|---|---|---|---|
| Married | Never-Married | Difference | Married | Never-Married | Difference | |
| Age | *** | *** | ||||
| 40-44 | 49.38 | 32.74 | 47.59 | 33.02 | ||
| 45-49 | 24.28 | 25.05 | 24.20 | 24.67 | ||
| 50-54 | 16.16 | 21.85 | 16.91 | 22.08 | ||
| 55-59 | 10.19 | 20.36 | 11.30 | 20.22 | ||
| Race and ethnicity | *** | *** | ||||
| White | 44.55 | 44.21 | 48.85 | 53.07 | ||
| Black | 22.15 | 29.63 | 18.80 | 20.88 | ||
| Hispanic | 24.74 | 18.71 | 24.07 | 19.05 | ||
| Other | 8.56 | 7.46 | 8.28 | 7.00 | ||
| Birthplace | *** | *** | ||||
| Foreign-born | 30.07 | 16.57 | 27.36 | 15.35 | ||
| U.S.-born | 69.93 | 83.43 | 72.64 | 84.65 | ||
| Education | **** | *** | ||||
| Less than high school | 14.06 | 14.20 | 13.14 | 17.71 | ||
| High school graduate | 22.70 | 27.23 | 28.16 | 34.71 | ||
| Some college | 27.28 | 29.01 | 27.62 | 25.96 | ||
| Bachelor’s degree or more | 35.97 | 29.57 | 31.09 | 21.62 | ||
| Employment | * | *** | ||||
| Not in labor force | 20.78 | 23.69 | 8.36 | 25.88 | ||
| Unemployed | 3.46 | 4.43 | 4.18 | 5.16 | ||
| Worked part time | 12.15 | 13.42 | 6.81 | 9.95 | ||
| Worked full time | 63.62 | 58.46 | 80.64 | 59.00 | ||
| Personal income | *** | *** | ||||
| <10k | 26.12 | 23.31 | 10.15 | 25.78 | ||
| 10-25K | 15.88 | 23.59 | 14.25 | 20.13 | ||
| 25-40K | 15.99 | 16.57 | 17.33 | 16.81 | ||
| 40-55K | 14.64 | 12.24 | 16.60 | 12.90 | ||
| 55-70K | 7.50 | 7.74 | 11.99 | 7.50 | ||
| >70K | 19.88 | 16.55 | 29.68 | 16.88 | ||
| Disability | *** | *** | ||||
| No disability | 90.48 | 80.86 | 91.65 | 79.97 | ||
| Has disability | 9.52 | 19.14 | 8.35 | 20.03 | ||
| Unweighted n | 1,319 | 52,722 | 1,708 | 67,973 | ||
p < .05
p < .01
p < .001
Column totals may exceed 100% due to rounding.
Data source: American Community Survey, 2019
The education distributions for women and men who married were higher compared with those who remain never married, and this differential was particularly wide among men. Among men who entered a midlife first marriage, 31.09% had at least a college degree versus only 21.62% among those who remained never-married. Full-time employment was especially common among those who married in the past year (63.62% for women and 80.64% for men) compared with those who remained never-married (58.46% for women and 59.00% for men). Income levels were higher for women and men who married in the last year, with large differentials observed for men. Roughly 42% of men who married were in the top two income brackets versus just 24% of men among the never married. Having a disability was uncommon among those who entered a midlife first marriage, hovering around 10% for women and men. Among those who stayed never married, roughly 20% reported having a disability.
Table 2 reports the AMEs from the logistic regression models predicting midlife first marriage entry separately by gender. The AME indicates the percentage change in the probability of entering first marriage for a given category relative to the reference category. The first column shows the model for women. On average, the probability of entering a midlife first marriage declined as age increased. Black and Other race women were significantly less likely than White women to have married whereas the probability for Hispanic women did not appreciably differ from that of White women. U.S.-born women were unlikely to marry compared with their foreign-born counterparts. Economic resources were largely unrelated to midlife women’s first marriage entry, net of other covariates. For instance, women with at least a college degree were no more likely to form a first marriage than their high school educated counterparts. Similarly, women’s employment status was not associated with midlife first marriage entry. The relationship between income and midlife first marriage entry was also weak. Only one significant association emerged between income and marriage with women earning $10,000-$24,999 less likely to have entered a midlife first marriage than women earning at least $70,000. The probability of marriage entry for women reporting a disability was lower than that of women who did not report having a disability.
Table 2.
Average Marginal Effects (Standard Errors) from Logistic Regressions Predicting the Likelihood of Midlife First Marriage, by Gender
| Women | Men | Cross-Model Difference | |||
|---|---|---|---|---|---|
| Age | |||||
| 40-44 (reference) | |||||
| 45-49 | − .010 (.002) | *** | − .009 (.002) | *** | |
| 50-54 | − .015 (.002) | *** | − .013 (.002) | *** | |
| 55-59 | − .024 (.003) | *** | − .018 (.002) | *** | |
| Race and ethnicitya,b | |||||
| White (reference) | |||||
| Black | − .008 (.002) | ** | .003 (.002) | *** | |
| Hispanic | − .004 (.003) | < .001 (.003) | |||
| Other | − .009 (.003) | ** | − .004 (.003) | ||
| U.S.-born | − .019 (.003) | *** | − .016 (.003) | *** | |
| Education | |||||
| Less than high school | − .001 (.003) | − .003 (.003) | |||
| High school graduate (reference) | |||||
| Some college | .003 (.002) | .003 (.003) | |||
| Bachelor’s degree or more | .005 (.003) | .005 (.002) | ** | ||
| Employmenta | |||||
| Not in labor force (reference) | |||||
| Unemployed | − .008 (.004) | .013 (.005) | ** | *** | |
| Worked part time | − .002 (.003) | .008 (.004) | * | ||
| Worked full time | < .001 (.003) | .014 (.004) | ** | * | |
| Personal incomea,b | |||||
| <10k | .006 (.004) | − .022 (.003) | *** | *** | |
| 10-25K | − .008 (.003) | * | − .015 (.003) | *** | |
| 25-40K | − .003 (.003) | − .013 (.003) | *** | * | |
| 40-55K | .002 (.003) | − .007 (.003) | * | * | |
| 55-70K | − .004 (.003) | − .002 (.003) | |||
| >70K (reference) | |||||
| Has disabilitya | − .013 (.003) | *** | − .006 (.003) |
p < .05
p < .01
p < .001
Data source: American Community Survey, 2019
Adjusted Wald test indicating a joint difference (p < .05) in likelihood of first marriage between women aged 18-39 versus 40-59 across the covariate using the SUEST command in Stata.
Adjusted Wald test indicating a joint difference (p < .05) in likelihood of first marriage between men aged 18-39 versus 40-59 across the covariate using the SUEST command in Stata.
Among men, age was negatively related to having entered a midlife first marriage. Notably, the likelihood of having entered a midlife first marriage did not differ across racial-ethnic groups. U.S.-born men had a lower probability of entering a midlife first marriage than foreign-born men. Economic resources were linked with midlife first marriage among men, net of other covariates. A college degree was positively associated with men’s entry into midlife first marriage. Relative to men who were not in the labor force, those who were either unemployed, working part-time, or working full-time were more likely to have married. Income was positively associated with men’s midlife first marriage entry. Men earning less than $55,000 were less likely than men earning at least $70,000 to have married. Disability status was unrelated to men’s risk of midlife first marriage.
For the most part, the covariates operated similarly for women and men (cross-model comparisons are reported in the last column of Table 2). However, Black women were significantly less likely to marry than Black men. And, among both the unemployed and the full-time employed, men were more likely to have wed than women. Among those in the lowest income bracket, women were more likely to have entered midlife first marriage than men whereas at middle income levels ($25,000-$39,999 and $40,000-$54,999) the opposite was true.
Supplemental analyses were performed to assess whether the covariates operated similarly for young (aged 18-39) versus midlife (aged 40-59) adults. First, we estimated models predicting first marriage for women and men aged 18-39 using the 2019 ACS to explore whether the sociodemographic covariates operated similarly among young adults (Table S2). Overall, the covariates operated as expected for young women and men. Second, we formally tested for differences in the effects of the covariates by life stage. Significant differentials by age group are shown in Table 2 using superscripts a and b for women and men, respectively. Several covariates were less strongly associated with first marriage among midlife than young adults. For instance, employment and disability status had weaker associations with marriage entry among midlife than young adult women. Racial gaps in first marriage entry were smaller in midlife than young adulthood for both women and men. Additionally, income was less strongly associated with women’s and men’s first marriage entry among the middle-aged, suggesting that the selection of the most economically advantaged into marriage may diminish in midlife.
Discussion
The steady ascent of the median age at first marriage in recent decades calls into question our persistent focus on marriage formation during young adulthood. It also foretells a life course shift in the timing and tempo of first marriage, signaling that first marriage entry may be increasingly occurring during midlife. Drawing on incidence rate data from the 1990 U.S. Vital Statistics and survey data from the 2010 and 2019 ACS, we estimated a midlife first marriage rate, tracked its change over time, and investigated its correlates today, to assess whether first marriage is gaining ground among the middle-aged.
The midlife first marriage rate has risen 75% for women and 45% for men since 1990. A midlife first marriage was exceedingly rare in the past, with only 2% of women and 3% of men marrying for the first time in 1990 being aged 40-59. Nowadays, 9% of women and 12% of men entering a first marriage are middle-aged. This shift is concordant with the sustained rise in the median age at first marriage, which is roughly 30 for women and men. It also indicates that first marriage is now less of an age-graded experience than it was in the past. The steep peak in first marriage entry during young adulthood that was evident in 1990 has flattened as fewer marriages occur among young adults and an increasing share occurs during midlife. The first marriage rate is higher for adults in early than late midlife, which presumably reflects age effects but also may be due to cohort effects. Regardless, the first marriage rate has risen across the entire spectrum of the midlife age range with larger percentage gains among those ages 50-59 than 40-49, signaling that for never-married middle-aged adults, marriage is increasingly viable.
Of course, this empirical evidence of an extended delay in marriage entry until midlife does not resolve larger questions about the extent to which marriage in the U.S. is in decline. The rise in the midlife first marriage rate does not fully offset the decline in the rate during young adulthood as the overall first marriage rate continues its steady descent over time. Rather, our study offers new evidence of growth in marriage delay. Stated differently, marriage is less often forgone among midlife adults than it was a few decades ago. Our study demonstrates using prospective data that the rate at which midlife never-married adults enter marriage is spiking. For some, marriage delayed until midlife is presumably marriage forgone, but that is much less true for today’s midlife never-married adults than those of a few decades ago. Increasingly, marriage is skewing towards older ages. Not only is the median age at first marriage at a historic high point, but so too is the midlife first marriage rate, supporting our hypothesis that adults are delaying marriage until older ages.
Our findings have the potential to reshape research on first marriage formation. In particular, we demonstrate the utility of expanding the age range of never-married adults examined in such studies. The strong ascent in the midlife first marriage rate calls into question the traditional boundaries characterizing research on marriage formation, including the narrow focus on young adults, which is arguably outmoded. First marriage entry has plummeted among those under age 30 and stalled among those aged 30-39.
Now that the experience of first marriage is less age-graded, its predictors and correlates also merit fresh scrutiny. We offer initial evidence that the antecedents of first marriage formation differ by life course stage. For example, although our analyses documented racial differentials in midlife first marriage among women, there were no appreciable racial differentials in the likelihood of first marriage among midlife men, which is a departure from prior work indicating widening racial gaps in marriage formation among younger adults (Bloome & Ang, 2020). In fact, several of the sociodemographic correlates of first marriage had distinctive effects for young versus middle-aged adults. The weaker effects of the correlates on first marriage among the middle-aged relative to young adults underscores the importance of deciphering the unique processes undergirding first marriage formation in midlife.
The shifting age distribution of first marriage is consonant with the patterns observed for divorce. Much like first marriage, divorce has plummeted among young adults even as it has risen among middle-aged and older adults (Brown & Lin, 2022, 2012; Cohen, 2019; Kennedy & Ruggles, 2014). Both first marriage and divorce are now less age-graded as rates continue to converge by age. These parallel trends in marriage and divorce point to what could be a transformational shift in family formation and dissolution patterns, which increasingly unfold during midlife rather than young adulthood.
Our study uncovered a notable rise in midlife first marriage and identified key correlates, but it also had some limitations. We were not able to directly address why this rise in the midlife first marriage rate occurred, although it would be challenging to identify existing data that would provide such evidence. We interpreted our results as showing an age effect, although it is possible that the increase we documented reflected cohort differences. By design of the ACS, covariates were measured after first marriage entry, precluding us from establishing temporal order. Our approach only documented correlational relationships and should not be construed as causal. Factors such as employment and income may have changed in response to marriage entry and thus our results should be interpreted with caution. Still, these factors operated in the expected positive direction. Likewise, there are other factors, such as having had a child, that may be predictive of first marriage formation in midlife that we were unable to measure (the ACS only captured children residing in the household at the time of interview). Additionally, premarital cohabitation presumably plays a role in midlife marriage formation with those cohabiting especially likely to formalize their union through marriage, but the ACS does not collect information on cohabitation experience. Regardless, the growth in midlife first marriage that we documented raises related questions about whether and how midlife cohabitation patterns have shifted in recent decades and whether such unions are more (or less) likely to eventuate in marriage nowadays.
By revealing the sharp rise in midlife first marriage, this study lays the groundwork for additional research on the predictors and consequences of this phenomenon. More broadly, it should reorient how scholars approach the study of first marriage formation, and lead to a new focus on first marriage during midlife. Despite unequivocal evidence of a steadily rising median age at first marriage in the U.S., the persistent focus on young adults in the union formation literature suggests that researchers have not adjusted the scope of their studies to reflect the shifting age distribution of first marriage. It is widely understood that first marriage rates are falling among young adults. Our study establishes that the first marriage rate is climbing among the middle-aged. The rise in midlife first marriage raises key questions about the subsequent stability of these marriages and their short- and long-term ramifications for individual health and well-being.
Supplementary Material
Acknowledgments
This research was supported in part by the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959). The authors thank Wendy D. Manning and Karen Benjamin Guzzo for their helpful comments on earlier versions of this manuscript.
References
- Addo FR (2014). Debt, cohabitation, and marriage in young adulthood. Demography, 51(5), 1677–1701. doi: 10.1007/s13524-014-0333-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bloome D, & Ang S (2020). Marriage and union formation in the United States: Recent trends across racial groups and economic backgrounds. Demography, 57, 1–34. doi: 10.1007/s13524-020-00910-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Breen R, Karlson KB, & Holm A (2018). Interpreting and understanding logits, probits, and other nonlinear probability models. Annual Review of Sociology, 44, 39–54. doi: 10.1146/annurev-soc-073117-041429 [DOI] [Google Scholar]
- Brown SL, & Lin I-F (2022). The graying of divorce: A half century of change. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences. 10.1093/geronb/gbac057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brown SL, & Lin I-F (2012). The gray divorce revolution: Rising divorce among middle-aged and older adults, 1990–2010. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 67, 731–741. 10.1093/geronb/gbs089 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brown SL, Van Hook J, & Glick JE (2008). Generational differences in cohabitation and marriage in the US. Population Research and Policy Review, 27(5), 531–550. doi: 10.1007/s11113-008-9088-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Carlson L (2020). Marriage in the U.S.: Twenty-five years of change, 1995-2020. Family Profiles, FP-20-29. Bowling Green, OH: National Center for Family & Marriage Research. doi: 10.25035/ncfmr/fp-20-29 [DOI] [Google Scholar]
- Cohen PN (2019). The coming divorce decline. Socius, 5, 1–6. doi: 10.1177/2378023119873497 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Eickmeyer KJ (2019). Age variation in the first marriage rate, 1990 & 2017. Family Profiles, FP-19-05. Bowling Green, OH: National Center for Family & Marriage Research. doi: 10.25035/ncfmr/fp-19-05 [DOI] [Google Scholar]
- Elliott DB, Simmons T, & Lewis JM (2010). Evaluation of the marital events items on the ACS. http://www.census.gov/hhes/socdemo/marriage/data/acs/index.html [Google Scholar]
- Hemez P (2020). Distributions of age at first marriage, 1960–2018. Family Profiles, FP-20-09. Bowling Green, OH: National Center for Family & Marriage Research. doi: 10.25035/ncfmr/fp-20-09 [DOI] [Google Scholar]
- Kennedy S, & Ruggles S (2014). Breaking up is hard to count: The rise of divorce in the United States, 1980–2010. Demography, 51, 587–598. doi: 10.1007/s13524-013-0270-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lachman ME (2004). Development in midlife. Annual Review of Psychology, 55, 305–331. doi: 10.1146/annurev.psych.55.090902.141521 [DOI] [PubMed] [Google Scholar]
- MacInnes MD (2011). Altar-bound? The effect of disability on the hazard of entry into a first marriage. International Journal of Sociology, 41, 87–103. doi: 10.2753/ijs0020-7659410105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mize TD, Doan L, & Long JS (2019). A general framework for comparing predictions and marginal effects across models. Sociological Methodology, 49, 152–189. doi: 10.1177/0081175019852763 [DOI] [Google Scholar]
- Martin SP, Astone NM, & Peters HE (2014). Fewer marriages, more divergence: Marriage projections for millennials to age 40. The Urban Institute. [Google Scholar]
- Payne KK (2021). Median age at first marriage, 2020. Family Profiles, FP-21-12. National Center for Family & Marriage Research. doi: 10.25035/ncfmr/fp-21-12. [DOI] [Google Scholar]
- Raley RK, Sweeney MM, & Wondra D (2015). The growing racial and ethnic divide in US marriage patterns. The Future of Children/Center for the Future of Children, the David and Lucile Packard Foundation, 25(2), 89–109. doi: 10.1353/foc.2015.0014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ratcliffe CE, Acs GP, Dore T, & Moskowitz DZ (2008). Assessment of survey data for the analysis of marriage and divorce at the national, state, and local levels. The Urban Institute, September. doi: 10.2139/ssrn.2206390 [DOI] [Google Scholar]
- Sassler S, & Lichter DT (2020). Cohabitation and marriage: complexity and diversity in union-formation patterns. Journal of Marriage and Family, 82(1), 35–61. doi: 10.1111/jomf.12617 [DOI] [Google Scholar]
- Schneider D, Harknett K, & Stimpson M (2018). What explains the decline in first marriage in the United States? Evidence from the panel study of income dynamics, 1969 to 2013. Journal of Marriage and Family, 80(4), 791–811. doi: 10.1111/jomf.12481 [DOI] [Google Scholar]
- Schweizer VJ (2020). Marriage: More than a century of change. Family Profiles, FP-20-21. Bowling Green, OH: National Center for Family & Marriage Research. doi: 10.25035/ncfmr/fp-20-21 [DOI] [Google Scholar]
- Smock PJ, & Schwartz CR (2020). The demography of families: A review of patterns and change. Journal of Marriage and Family, 82(1), 9–34. doi: 10.1111/jomf.12612 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.