Cohort Trends in Union Dissolution During Young Adulthood

Abstract

Objective:

To determine whether recent birth cohorts of women experienced more union dissolution during young adulthood (18–25) than previous birth cohorts.

Background:

The union formation and dissolution patterns of young adult women in the U.S. have changed dramatically over the last twenty-five years. As a result, this life stage is demographically dense as women experience a bulk of relationship experiences, including coresidential unions and dissolution.

Method:

The author uses data on women’s marital and cohabiting dissolutions between the ages of 18–25 from the National Survey of Family Growth 1995, 2002, and continuous surveys 2006–2010 and 2011–2015 (N = 14,211). The sample is nationally representative. The author uses generalized ordinal logistic regression to examine the likelihood of dissolution during young adulthood across birth cohorts.

Results:

Women born between 1985 and 1989 experience more union dissolutions during young adulthood than women born between 1960 and 1979. However, the shift in cohabitation behavior accounts for this instability.

Conclusion:

Women’s relationship formation experiences in young adulthood are characterized by nonmarital relationships (cohabitation) and more union dissolution compared to older birth cohorts. Theoretical and empirical studies need to evolve to include this more complicated sense of young adulthood, union formation, and instability.

The patterns of young adult women’s union formation and dissolution in the United States have shifted dramatically over the past twenty-five years, making young adulthood (ages 18 to 25) a unique and dense (Rindfuss, 1991) period in the life course. The median age at first marriage was approximately 24 in 1985, yet today it nears thirty (Eickmeyer et al., 2017). However, while young American women may be delaying marriage more than ever, they are not delaying co-residential unions and are instead cohabiting in their early twenties (Manning, Brown, & Payne, 2014). While cohabiting unions in the U.S. last longer than those formed in the 1980s, they remain relatively temporary, especially compared to other industrialized nations (Kasearu & Kutsar, 2011). In the U.S., young adult women are more likely to experience the dissolution of these cohabiting unions than they are to marry their cohabiting partner (Lamidi, Manning, & Brown, 2015). Some women then go on to cohabit with another partner, and these higher-order cohabiting unions are prone to dissolution (Eickmeyer & Manning, 2018; Lichter & Qian, 2008). So, while the share of young adult women in the U.S. who have ever separated from or divorced their spouse has declined, the dissolution risk associated with cohabiting unions may result in a larger share of young adult women ever dissolving a union, despite the decline in separation or divorce.

The time between first cohabitation and first marriage represents an opportunity for women to experience multiple co-residential partnerships, and while young women are less likely to experience a divorce than earlier generations of women (i.e., those who experienced young adulthood in the 1980s), they are still ending cohabiting romantic relationships. These dissolutions may have implications for subsequent transitions into marriage and future relationship functioning and quality.

Using Cycle 5 (1995), Cycle 6 (2002), and continuous 2006–2015 National Survey of Family Growth (NSFG), this study documents trends in union dissolution during young adulthood across five-year birth cohorts spanning 1950 to 1980. This study determines not only the levels of instability but also the type of relationship instability experienced in young adulthood (cohabiting or marital). I contribute to research on union instability during young adulthood by determining whether recent birth cohorts of young adult women are more likely to experience multiple union dissolutions compared to women from prior birth cohorts spanning 1960 to 1989. Additionally, I consider whether cohabitation experience explains any differences in union dissolution between women born between 1985 and 1989, and earlier cohorts of women born between 1960–1964, 1965–1969, 1970–1975, 1975–1979, 1980–1984, and 1985–1989 net of sociodemographic characteristics, family background, or relationship history.

Background

Changes in U.S. Marriage and Divorce during Young Adulthood

Previous research indicates an overall retreat from marriage characterized not only by declining rates of marriage, but a delay in first marriage (Kennedy & Ruggles, 2014). 68% of women were married by the end of young adulthood in 1980; by 2015, this proportion had declined to 40% (Anderson, 2017). This corresponds to the rising median age at first marriage since 1980, from age 22 to age 27 in 2016 (Eickmeyer et al., 2017). These trends are one aspect of the change in marriage often attributed to the declining economic opportunities for young men, extended education, expanding women’s labor force participation, and rising incidence of premarital cohabitation (Lundberg & Pollak, 2013; Oppenheimer, Kalmjin, & Lim, 1997; Ruggles, 2015). Taken together, these marriage trends suggest that young adulthood is no longer characterized by entrance into a first marriage, and may be characterized by romantic alternatives.

Trends in marriage during young adulthood correspond to changes in divorce. In the 1980s, almost half of all divorces were concentrated within young adulthood, yet the divorce rate for young adult women declined by over two-fifths between 1990 and 2012 (Brown et al., 2014; Clark, 1995). This shift in divorce out of young adulthood may be attributed to the rising age at first marriage. Recent research indicates that the rising age at first marriage may be significantly related to the decline in divorce over time (Rotz, 2015). Divorce among young adult women, then, may be on the decline as young women delay marriage.

U.S. Trends in Cohabitation during Young Adulthood

Cohabitation is arguably a common experience among women during young adulthood in the United States. A majority of young adults have ever cohabited, and most marriages are preceded by cohabitation (Manning & Stykes, 2015). In some sense, cohabitation may have replaced marriage during young adulthood. However, even though cohabitation is more common, cohabiting unions are more likely to end in dissolution today than twenty-five years ago (Lamidi et al., 2015). While contemporary cohabitations in young adulthood last far longer than those formed in the 1980s (29 months compared to 18 months), these unions remain unstable and short-lived compared to cohabiting unions in other contexts (Lamidi et al., 2015; Kasearu & Kutsar, 2011).

An implication of this increased risk of dissolution of cohabiting unions is an opportunity for young adults to re-partner following a breakup. Young adulthood is characterized by romantic exploration, but not necessarily settling down, which may create an environment in which young women re-partner (Arnett, 2000). Overall, research indicates that the share of women cohabiting more than once (serially cohabiting) continues to grow among younger cohorts of women (Eickmeyer & Manning, 2018). These “serial cohabitations” may face an even higher risk of union dissolution (Lichter & Qian, 2008). As young adult women continue to cohabit, and have multiple cohabiting unions, we may see women’s young adult years characterized by more union dissolution than among previous birth cohorts.

Relationships in young adulthood, therefore, are becoming complex even before marriage. Rising relationship instability leads to more opportunities for young adults to repartner and form families before their first marriage (Payne, 2011). Likely these relationship experiences are consequential as these experiences accumulate. Precarious relationship experiences may manifest as relationship “baggage”, which interferes with building and maintaining intimate ties (Young, Furman, & Laursen, 2011). This accumulation of “baggage” may lead to unstable or unhealthy relationships. However, the dissolution of relationships and repartnering during young adulthood does not necessarily indicate maladapted behavior and may instead serve as a learning experience. These relationship experiences may contribute to relationship “competence” as young adults learn the ins and outs of committed relationships and bring knowledge of positive and negative relationship functioning into subsequent relationships (Young et al., 2011). That much of this learning is happening prior to marriage suggests that young adults could have a chance to experience more stable marriages than previous cohorts of women.

Cohabiting unions unions today are less likely to transition to marriage, even accounting for engagement (Guzzo, 2014). Thus, a cohabiting union may be less like a trial marriage or a step in the marriage process today and more likely to stand on its own for alternative reasons (Sassler, 2004). Given the trends in births to cohabitors, and the likelihood of a cohabiting union to be a stepfamily, these dissolutions may result in more tangible “baggage”, but any relationship, whether it ends or continues, offers young adults an opportunity to gain competence in relationship management and partnership decisions.

Correlates of Union Instability

Apart from the shifts in union formation, there are many additional correlates of union instability. I include demographic and socioeconomic indicators, family background characteristics, and relationship history. Trends in union instability differ by race and ethnicity as well as education attainment (Lundberg, Pollak, & Stearns, 2016; Raley, Sweeney & Wondra, 2015). Children who have experienced their parents’ divorce or lived apart from a two-parent family have a heightened risk of marital instability (Amato, 1996; Lyngstad & Jalovaara, 2010; Teachman, 2002). A family background characterized by socioeconomic advantage is associated with a decreased risk of divorce (Lyngstad & Jalovaara, 2010). An early age at first marriage or cohabitation is related to an increased likelihood of dissolution (Amato, 1996; Lamidi et al., 2015). Women who bear children prior to marriage are more likely to divorce (Amato, 2010). Women who have children prior to cohabiting are also at-risk of dissolving their union (Guzzo, 2017).

Current Study

Given recent shifts of young adult romantic relationships away from marriage and toward cohabitation, it is important to continue to assess the instability experienced during this life stage. I explore whether union instability has increased across six birth cohorts of women: women born between 1960–1964, 1965–1969, 1970–1974, 1975–1979, 1980–1984, and 1985–1989. I expect that women born between 1985 and 1989, women who most recently experienced young adulthood, amassed a greater number of union dissolutions between the ages of 18 and 25 compared to women born in birth cohorts prior to 1985. I expect this to be especially apparent among women born in the earliest birth cohort, 1960–1964. This may be in part because of the increased likelihood of cohabitation. If women born between 1985 and 1989 experience more instability during young adulthood, but this is accounted for by cohabitation, this suggests that cohabitation dissolutions have offset any declines in marital dissolution during this life stage and cohabitation, albeit a common part of young adulthood, may destabilize women’s romantic experiences during young adulthood. However, if net of cohabitation and key correlates associated with union dissolution, women born between 1985 and 1989 experience more instability compared to earlier birth cohorts, this suggests that union instability during young adulthood requires a more complex consideration, which I return to in my conclusion.

Data and Method

The National Survey of Family Growth (NSFG), designed by the National Center for Health Statistics (NCHS), is a nationally representative survey, which captures detailed retrospective information about women’s relationship formation and dissolution, including marriage and cohabitation. These retrospective reports include start and end dates for four marriages and premarital cohabitations, four non-marital cohabitations, and if applicable the current cohabiting union. The use of Cycle 5(1995), Cycle 6, (2002), and continuous 2006–2010 and 2011–2015 NSFG supports analyses of changing union dissolution experiences of 5-year, equal sized birth cohorts of women born between the years of 1960 and 1984.

Cycle 5 (1995) of the NSFG completed 10,847 interviews with women; Cycle 6 (2002) interviewed 12,571 women; the 2006–2010 continuous NSFG included interviews with 12,279 women; the 2011–2015 NSFG completed interviews with 13,342 women. All surveys include an oversample of Black and Hispanic women. Due to the lack of male respondents in Cycle 5, and the discordant cohabitation history questions between male and female respondents in the continuous surveys of the NSFG, only female respondents were included in the analytic sample.

Analytic Sample

I began with 42,069 cases from the five interviews between 1995 and 2015. I constructed generational birth cohorts using the respondent’s century-month birthdate and removed 9,498 women who were not born between years 1960–1989 (N = 32,571). Young adulthood is often defined as between the ages of 18–25 or 18–29 (Arnett, 2000; Fry, 2016). For the youngest cohort of women, an upper age limit of 25 may lead to a truncated marital history, as many of their marriages are at substantially later ages. With this limitation in mind, I defined young adulthood as between the ages of 18–25 in order to capture this broad range of birth years and to maximize the utility of the NSFG data. In order for women to have complete retrospective union experiences during this life stage, women needed to be at least age 26 at the time of the interview. This excluded 11,702 respondents, resulting in a sample size of 20,869. I excluded 6,639 women who did not form a coresidential union (marriage or cohabitation) between ages 18 and 25, as women must have experienced either a marriage or a cohabiting union in order to be at risk of a union dissolution. This specification resulted in a sample size of 14,230 women who ever married or cohabited between ages 18 and 25. Finally, I excluded women who were missing on information related to cohabitation or marital dissolutions (N = 19) for a final sample size of 14,211.

Dependent Variables

The outcome I examined was union instability during young adulthood. I summed the number of union dissolutions—both marital and cohabiting—experienced during young adulthood. Marital dissolutions include both separation and divorce. Few women experienced more than two union dissolutions between the ages of 18–25, so the variable was collapsed into none (0), one (1), and two or more (2).

Independent Variables

Six mutually exclusive birth cohorts were constructed from women’s birth dates, including the years of 1960–1964, 1965–1969, 1970–1974, 1975–1979, 1980–1984, and 1985–1989. In order to address recall bias (Hayford & Morgan, 2008), I restricted the birth cohorts to women who were between the ages of 18 and 25 within ten years prior to the interview year. For the earliest cohort of women, for example, interviewed in 1995, they would have to be born during or after 1960 to be included in the analyses.

The additional focal variable was cohabitation, indicated by a dichotomous variable indicating whether women had ever cohabited (1) or never cohabited (0). Never cohabited served as the reference category.

Controls

Those who identified as Hispanic, Latina, or of Spanish origin were coded as Hispanic. Else, women’s race and ethnicity were coded as non-Hispanic White, non-Hispanic Black, or non-Hispanic Other race. The reference group for race and ethnicity was non-Hispanic White. Respondent’s highest level of education, measured at the time of the interview, was coded into four categories: (a) less than high school, (b) high school diploma or GED, (c) some college, associate’s degree, (d) bachelor’s degree or higher. The reference category was a high school diploma or GED. The NSFG asked whether the respondent lived with two biological or adoptive parents from birth to age 18, and this variable was coded 1 for a two-parent family structure and 0 otherwise. The reference category was any family other than a two-parent family. Mother’s education was indicated by four categories: (a) less than high school, (b) high school diploma or GED, (c) some college, associate’s degree, (d) bachelor’s degree or higher. A high school diploma or GED was the reference category. The characteristics of the union included age at first union and whether a woman had an unpartnered first birth. The respondent’s age at first union is a continuous variable constructed from their age at first cohabitation or their age at first marriage. Whether the respondent had an unpartnered first birth prior to their first union was indicated by the timing of live births relative to the date of entering their first union (marriage or cohabitation). No unpartnered first birth was the reference category.

Analytical Approach

Descriptive statistics by birth cohort detail the distribution of the sociodemographic, family background, and relationship history characteristics of women overall, and those who ever formed a union during young adulthood (i.e. women at risk of experiencing union dissolution). These are followed by a description of union dissolution among young adult women by birth cohort and marriage or cohabitation experience. Following the descriptive statistics, I use generalized ordinal logistic regression to estimate the likelihood of experiencing higher union instability during young adulthood. The generalized ordinal logistic model assess the proportional odds assumption and to fit a partial proportional odds model (Williams, 2006). This model relaxes the proportional odds assumption for those variables that violate the proportional odds assumption while fixing those that do not, resulting in a model that is not as constrained as the proportional odds model but not completely freed from the assumptions as a multinomial logistic regression model (Williams, 2006).

The four models are nested: Model 1 includes only the indicator for birth cohort, assessing whether the odds of union instability are higher for the youngest birth cohort. Model 2 adds cohabitation experience to Model 1 in order to determine whether cohabitation accounts for union instability differences during young adulthood. Model 3 includes sociodemographic and family background indicators and Model 4 includes women’s relationship history to examine whether compositional differences explain the association between instability and birth cohort, and instability and cohabitation. Analyses were weighted to account for the complex sampling design of the NSFG.

Results

Descriptive Statistics

Results from Table 1 indicate that, among all women over the age of 25, the trend in dissolution during young adulthood is not straightforward. However, among women at risk of experiencing dissolution, union instability has increased across birth cohorts. Among women born between 1985 and 1989, nearly 30% dissolved a union compared to 19% of women born between 1960 and 1964.

Table 1.

Descriptive statistics

Birth cohort	b.1960–1964	b.1965–1969	b.1970–1974	b.1975–1979	b.1980–1984	b.1985–1989
	%| M	%| M	%| M	%| M	%| M	%| M
Among all women over age 25 (N=22,389)
Ever dissolved a union	21.2	19.6	22.8	25.6	25.3	19.2
Among women ever in a union (N = 14,211)
Frequency of dissolution
0	68.6	68.4	65.6	64.4	59.2	55.6
1	18.8	18.5	20.5	22.1	25.6	29.7
2+	12.7	13.1	13.9	13.6	15.2	14.7
Among ever married (N=9,213)
Ever divorced or separated	22.0	20.4	21.1	21.2	22.4	17.5
Among ever cohabited (N=9,635)
Ever dissolved a cohabitation	27.6	29.2	31.4	32.3	37.5	45.2
N	3,885	4,897	4,713	4,350	3,390	1,154

Note: All values are weighted; ever married and ever cohabited not mutually exclusive categories.

Source: 1995, 2002, 2006–2010, 2011–2015 NSFG

The next set of results specifically distinguishes between those who married and those who cohabited. Among women who were married during young adulthood, the share of women experiencing a divorce declined. The cohort trend in cohabitation dissolution is much more apparent: the share of cohabitors experiencing dissolution increased from 28% among women born in the early 1960s to 45% among those born in the late 1980s. The results from Table 1 indicate that women born between 1985 and 1989 appear not just more likely to experience union dissolution during young adulthood than their counterparts born between 1950 and 1979, but they experience a greater number of union dissolutions. This instability, however, appears to be driven more by cohabitation than by marriage, as the share of women divorcing during young adulthood has declined, but the share of women ending cohabitation has notably increased.

Table 2 presents the characteristics of women who formed a union during young adulthood. Consistent with previous research, the number of women cohabiting during young adulthood increased substantially across birth cohorts. What is most striking, however, is the 41% increase in the share of women who dissolved a union during young adulthood, from 31% to 44%. Additionally, the type of union dissolutions experienced across birth cohorts shifted, such that a vast majority of dissolutions during young adulthood for women born after 1965 were cohabiting. Only 9% of women born after 1985, compared to 17% of the early 1960s cohort, experienced only marital dissolution.

Table 2.

Descriptive statistics, women who ever formed a union between ages 18–25 (N = 14,211)

Birth cohort	b.1960–1964	b.1965–1969	b.1970–1974	b.1975–1979	b.1980–1984	b.1985–1989
Variable	%| M	%| M	%| M	%| M	%| M	%| M
Ever cohabited	51.7	59.0	66.8	72.9	77.3	79.0
Ever dissolved a union	31.4	31.6	34.4	35.6	40.8	44.4
Type of dissolutions
No dissolution	68.6	68.4	65.6	64.4	59.2	55.6
Only cohabiting	12.6	15.8	19.3	21.6	26.7	34.2
Only marital	16.9	14.1	13.2	11.8	11.8	8.7
Both cohabiting and marital	1.9	1.7	1.9	2.2	2.3	1.5
Race/ethnicity
Non-Hispanic White	75.7	68.0	65.7	63.4	59.6	64.5
Non-Hispanic Black	9.8	12.2	10.9	11.5	13.0	11.8
Hispanic	10.7	13.9	17.4	18.8	20.4	17.2
Non-Hispanic Other	3.9	5.8	6.0	6.2	7.1	6.5
Education
Less than high school	12.2	10.2	12.8	10.8	9.9	9.2
High school diploma or GED	35.4	33.5	28.0	27.8	27.8	26.4
Some college	31.9	30.2	28.6	32.2	32.5	36.7
Bachelor’s degree or higher	20.4	26.1	30.6	29.2	29.8	27.7
Family background
Raised with two biological/adoptive parents	67.5	63.6	61.4	56.5	56.6	51.9
Mother’s education
Less than high school	28.22	27.6	26.1	23.7	22.5	17.3
High school diploma or GED	44.9	41.8	37.0	35.1	32.1	35.6
Some college	15.6	16.9	20.0	23.4	26.9	22.9
Bachelor’s degree or higher	11.3	13.7	16.9	17.8	18.5	24.3
Relationship history
Age at first union (mean)	20.4	20.6	20.5	20.6	20.4	20.2
Unpartnered first birth	10.0	12.1	12.0	14.2	11.6	10.3
N	2,615	3,030	3,123	2,837	2,106	500

Note: All values are weighted.

Source: 1995, 2002, 2006–2010, 2011–2015 NSFG

The statistics presented in Table 2 also demonstrate a shift in the composition of young adult women who experienced coresidential unions across birth cohorts. The sample is predominantly White, but there is an increase in the proportion of non-White women over time. Women attain a higher level of education, and the proportion of women who grew up with two parents declined by 23%. The trend in mother’s education reflects higher educational attainment. Women formed their first union at approximately the same ages over birth cohorts, hovering around 20 years old. The share of women who had experienced an unpartnered first birth also remained static. It appears that the relationship experiences during young adulthood have shifted across birth cohorts, but many sociodemographic and family background characteristics have varied as well. Next, I turn to the multivariate analyses to account for all of these characteristics.

Multivariate Analysis

Table 3 presents the generalized ordered logistic regression models predicting union dissolution between the ages of 18 and 25. Again, models are nested to illustrate the role of cohabitation experience and additional correlates of union dissolution. These models present the odds ratios ratios (OR) of an increase in union dissolution, relative to none, during young adulthood. The baseline Model 1 includes just the indicator of birth cohort, and partially supports the hypothesis that, compared to women born between 1985 and 1989, women born between 1960 and 1979 are less likely to experience one or more union dissolution during young adulthood. Women born in the early 1980s, however, were not significantly different than women born after 1984. Relative to the latest birth cohort, being born in the early 1960s was associated with a 38% reduced likelihood of moving to higher levels of dissolution during young adulthood.

Table 3.

Generalized ordered logistic regression odds ratios predicting dissolution between ages 18–25 (N =14,211)

	Model 1		Model 2		Model 3		Model
Variable	OR	SE	OR	SE	OR	SE	OR	SE
Birth cohort (ref. = 1985–1989 [Millennials])
1960–1964	0.62***	0.08	0.91	0.13	0.88	0.12	1.00	0.14
1965–1969	0.62***	0.09	0.80	0.11	.0.81	0.11	0.92	0.13
1970–1974	0.71*	0.10	0.83	0.11	0.87	0.11	0.92	0.12
1975–1979	0.74*	0.10	0.80	0.11	0.83	0.10	0.91	0.12
1980–1984	0.90	0.12	0.93	0.13	0.99	0.13	1.05	0.14
Ever cohabited (ref. = never cohabited)			4.87***	0.31	4.40***	0.29	4.82***	0.33
Sociodemographics
Race/ethnicity (ref. = non-Hispanic White)
Black					0.90	0.06	1.13	0.09
Hispanic					0.61***	0.05	0.64***	0.05
Other					0.77*	0.09	0.82	0.12
Education (ref. = high school diploma)
Less than high school					1.21*	0.10	0.85	0.08
Some college					0.81***	0.05	1.00	0.07
Bachelor’s degree or higher					0.46***	0.03	1.01	0.08
Family background
Raised by two parents (ref. = no)					0.64***	0.04	0.79***	0.05
Mother’s education (ref. = high school diploma
Less than high school					0.85*	0.06	0.81**	0.06
Some college					1.05	0.07	1.07	0.08
Bachelor’s degree or higher					1.06	0.08	1.14	0.08
Relationship history
Age at first union							0.67***	0.01
Unpartnered first birth (ref. = no)							1.25*	0.12

^*p < 0.05

^**p < 0.01

^***p < 0.001

Source: 1995, 2002, 2006–2010, and 2011–2015 National Survey of Family Growth

Model 2 includes cohabitation experience to test whether cohabitation accounts for birth cohort differences in union dissolution. The results are clear. Cohabitation experience greatly increases the likelihood of union dissolution (OR= 4.87). With the inclusion of cohabitation, being born prior to 1985 is not significantly associated with union dissolution during young adulthood.

Model 3 includes sociodemographic and family background characteristics in order to determine whether these characteristics explain the association between cohabitation and union dissolution. The likelihood of dissolution for birth cohorts and ever cohabiting remained largely the same. These sociodemographic and family background characteristics, however, were associated with union instability. Relative to White women, Hispanic and non-Hispanic Other race women had a significantly lower likelihood of union dissolution during young adulthood. Women with less than a high school education were more likely to dissolve (OR = 1.21), while women with some college education or higher had a lower likelihood of dissolution. The likelihood of dissolution was significantly lower for women who were raised by two parents compared to those who were not.

Model 4, the final model, includes indicators of women’s relationship history. The association between cohabitation experience and dissolution is robust to the inclusion of these characteristics, and cohabiting during young adulthood remains not just significant, but great in magnitude (OR = 4.82). The association between sociodemographic characteristics and union dissolution largely mirror those in Model 3, with the exception of race and women’s education. Supplementary analysis (not shown) indicate that non-Hispanic Other race women are no longer less likely to experience union dissolution with the inclusion of the age at first union. The age at first union also accounts for the previous association between education and dissolution. As the age at a woman’s first union increased, her likelihood of dissolution during young adulthood was reduced by 33%. Having an unpartnered first birth increased the likelihood of instability by 25%.

The results in Table 3 support our hypotheses that Millennials experience more relationship instability during young adulthood than earlier birth cohorts of women, but that cohabitation experience accounts for this instability. To determine whether sociodemographic, family background, and relationship factors attenuated the relationship between birth cohort and union dissolution, supplemental analyses measured Model 4 without cohabitation experience (not shown). Without cohabitation experience, being born before 1975 is associated with a significantly lower likelihood of dissolution. The large odds ratios in Model 4 suggest that few women who experienced union dissolutions did not cohabit, and across all birth cohorts a majority of women who dissolved did cohabit, from 78% among women born in the early 1960s to 97% of women born between 1985 and 1989. This illustrates the increase in cohabitation across generations, but also supports the hypothesis that cohabitation experience largely drives the union instability experienced during young adulthood.

Conclusion

The goal of this study is to extend research on young adult union instability among a recent cohort of women. It appears that union dissolutions are increasingly common for American women. Experiences of divorce and cohabitation dissolution are often considered as separate phenomenon, yet this study illustrates that it is important to tell a story that considers all union dissolution, as relationship experiences have been rearranged to include more cohabitation and less marriage during young adulthood. Not only are more women born since 1985 forming any union during young adulthood than earlier cohorts, these are almost exclusively cohabiting unions, and the union instability experienced during young adulthood has increased. Results suggest, however, that birth cohort differences between women born in the1985–1989 birth cohort and preceding cohorts are explained by cohabitation experience. Divorce during young adulthood has declined among these women, but cohabitation dissolution has increased, and appears to drive the instability during young adulthood. This demonstrates that accounting for cohabitation is important in explaining union dissolution differentials between birth cohorts of women.

The sociodemographic composition of birth cohorts has shifted, and these changing compositional factors could temper the association between cohabitation and union dissolution. However, controlling for sociodemographic characteristics, family background, and relationship history does not account for the role of cohabitation. Cohabitors continue to have the highest risk of union instability during young adulthood, net of selection factors.

Young adulthood has become a period of development that garners explicit attention. It is a demographically unique and demographically ‘dense’ (Rindfuss, 1991) stage in individuals’ lives in which they attempt to lay a foundation for future relationships and family formations much more fervently than in adolescence. It appears that relationship trajectories have become far more complex for women born after 1985, than they were for earlier birth cohorts. The paths through young adulthood have become less clear and more diverse, and while this relaxation of roles and pathways may seem a benefit to young adults attempting to find their way, it ultimately may induce uncertainty and vulnerability (Settersten, 2012). The increasing instability of coresidential relationships during young adulthood, then, may be related to this uncertainty and transitory nature characterizing this life stage. Facing questions about which pathway to follow during this life stage, women may be choosing to end their relationships in the presence of uncertainty instead of remaining in an intact relationship or transitioning to marriage.

Experiencing relationship dissolution during young adulthood can have long-term implications. They may contribute to either relationship competence or relationship baggage that is carried into new relationships and contributes to patterns and habits of relationship functioning (Young et al., 2011). Prior evidence indicates that recent cohorts may face increased financial and psychological instability in the aftermath of these multiple dissolutions (Avellar & Smock, 2005; Kamp-Dush, 2013). They may become more likely to form stepfamilies as a result of ending and forming new cohabiting unions and bringing their children into new relationships, which can have implications for children’s well-being as well as adults (Guzzo, 2017).

While these results move our assessment of relationship instability forward, there are a few shortcomings. First, the NSFG does not contain couple-level data, or comparable questions between male and female questionnaires, so I am only able to analyze one side of the experience that takes two people, and I am unable to compare men and women’s experiences. A focus on men’s sociodemographic and risk profile in relation to the odds of overall dissolution, as well as charting their patterns over time, is warranted in future research. Second, the NSFG is cross-sectional and relies on retrospective reports of cohabitation. There is always the possibility of downward bias regarding cohabiting unions as time passes (Hayford & Morgan, 2008). Third, cohabitations are quite heterogeneous and the meaning of cohabitation may have shifted over time, which means that the implications of these dissolutions or exact start and end dates may vary across birth cohorts.

Finally, women in this study may have experienced even more union formations and dissolutions across their life course. The focus on young adults means that few marriages are included for recent cohorts. As women continue to age toward young adulthood, they may face unprecedented relationship instability. Birth cohorts develop unique life trajectories in response to their place in history, and I find evidence that indicates a pathway to adulthood more akin to a spiral staircase than an escalator—in other words, women’s relationship formation experiences in young adulthood are characterized by non-marital relationships and more union dissolution compared to older generations. Theoretical and empirical studies need to evolve to include this more complicated sense of young adulthood, union formation, and instability.