Abstract

Men’s and women’s economic resources are important determinants of marriage timing. Prior demographic and sociological literature has often measured resources in narrow terms, considering employment and earnings and not more fine-grained measures of job quality. Yet, scholarship on work and inequality focuses squarely on declining job quality and rising precarity in employment and suggests that this transformation may matter for the life course. Addressing the disconnect between these two important areas of research, this study analyzes data on the 1980–1984 U.S. birth cohort from the National Longitudinal Survey of Youth 1997 to examine the relationships between men’s and women’s job quality and their entry into marital or cohabiting unions. We advance existing literature by moving beyond basic measures of employment and earnings and investigating how detailed measures of job quality matter for union formation. We find that men and women in less precarious jobs—both jobs with standard work schedules and those that provide fringe benefits—are more likely to marry. Further, differences in job quality explain a significant portion of the educational gradient in entry into first marriage. However, these dimensions of job quality are not predictive of cohabitation.

Introduction

Since the 1970s, economic opportunities in the United States for those with less than a college degree have deteriorated: earning power, job security, and jobs with good benefits have diminished, while precarious employment has become more prevalent (Fligstein and Shin 2004; Kalleberg 2009). Over the same period, entry into first marriage has declined precipitously, and marriage has become increasingly stratified by class, with more-educated men and women now more likely to marry than their less-educated counterparts (Ellwood and Jencks 2004; Goldstein and Kenney 2001; McLanahan 2004; Schneider et al. 2018; Wang and Parker 2014).

Although a great deal of social scientific research has examined the influence of employment and earnings on family formation and their role in explaining the stark differences across education groups in family formation, this literature has taken a narrow approach to measuring employment. Unemployment and low earnings clearly matter for family formation (i.e., Burstein 2007) and contribute to educational gaps in marriage (Harknett and Kuperberg 2011), but these measures do not capture important dimensions of job quality that have changed over time and are increasingly stratified by education.

Yet, it is precisely these aspects of employment contracts that are the focus of much scholarly and public discussion of the changing American economy (i.e., Kalleberg 2011; Steverman 2014). For instance, Hacker (2006) described the transfer of risk from large institutional actors, such as from employers to households and workers, evidenced in the erosion of employment benefit packages. Along similar lines, scholars have also recently called attention to another transfer of risk from employers to employees seen in the rise in on-call and variable schedules for hourly workers by which employers effectively transfer payroll risk from firm to worker (Boushey 2016; Lambert 2008). This erosion of job quality has been most pronounced for those with less than a college degree (Kalleberg 2009).

We use data from the National Longitudinal Survey of Youth 1997 (NLSY97; Bureau of Labor Statistics 2015) to examine the role of job quality in union formation and in explaining stark educational differences in marriage. First, we assess the extent to which job quality matters for entry into marriage and cohabitation. We also extend the rich existing literature concerned with gender differences in the relationship between employment and marriage by investigating whether job quality operates differently on marriage for men and women. Finally, we examine whether accounting for measures of job quality can explain educational gradients in entry into first marriage.

Theory and Prior Research

We first discuss how employment affects entry into marriage and cohabitation, for men and for women, and discuss educational gradients in both marriage and cohabitation and prior approaches to explaining these gradients. We then turn to the literature that focuses more squarely on job quality and union formation before integrating this previous theory and empirical research to derive a set of hypotheses about how job quality is likely to shape union formation, matter differently by gender for union formation, and account for educational gradients in union entry.

Men’s Employment and Entry Into Marriage

Sociologists and demographers have theorized that men’s economic resources are a mark of marriageability (Davis and Blake 1956; Edin and Kefalas 2005; Wilson 1987) and thus make men with such resources more attractive as potential partners and perhaps prompt men themselves to feel more ready to marry. Research has shown that men’s employment status and earnings are strongly related to transitions to first marriage (for comprehensive reviews, see Burstein 2007; Ellwood and Jencks 2004). Studies using a variety of data sets have consistently found that men are more likely to marry when they are employed (Harknett and Kuperberg 2011; Harknett and McLanahan 2004; Sassler and Goldscheider 2004; Sweeney 2002) and especially when they are employed full-time (Kuo and Raley 2014; Oppenheimer 2003; Oppenheimer et al. 1997; Schneider 2011; Shafer and James 2013). Similarly, men who earn more also have a higher risk of marriage (Clarkberg 1999; Kuo and Raley 2014; McClendon et al. 2014; Oppenheimer 2003; Oppenheimer et al. 1997; Schneider 2011; Schneider and Reich 2014; Shafer and James 2013; Sweeney 2002).

Women’s Employment and Entry Into Marriage

There has been some debate, though, regarding whether women’s work and earnings would operate similarly to men’s and increase entry into marriage (Oppenheimer 1988), or would instead grant women a degree of autonomy that would discourage marriage (Becker 1981). The weight of the research evidence suggests that in older cohorts, women’s employment and earnings might have acted as a deterrent or substitute for marriage because economically independent women were empowered to forgo marriage if that was their preference, high-achieving women were violating strong norms encouraging male breadwinners and female homemakers and were therefore less attractive on the marriage market, or nonmarriage created an imperative for women to work more and earn more (Burgess et al. 2003; Xie et al. 2003).

For more recent cohorts, however, women’s employment and earnings have come to be positively associated with entry into marriage. Research has found a trend toward gender convergence in the influence of employment and earnings on marriage (Clarkberg 1999; Lichter et al. 1992; Schneider 2011; Schneider and Reich 2014; Shafer and James 2013; Sweeney 2002). More recent work using the 1980–1984 birth cohort captured in the NLSY97 found that the positive relationship between women’s earnings and marriage persists (Kuo and Raley 2014; McClendon et al. 2014), as does the relationship between marriage and full-time employment (Addo 2014). However, although the weight of evidence suggests that women’s employment is positively associated with marriage, the strength of this association is generally weaker than for men (Burstein 2007; Shafer and James 2013).

Influence of Employment on Entry Into Cohabitation

Comparing marriage and cohabitation can bring the pathways through which employment influences union formation into sharper relief. If the main pathway through which employment affects union formation is that economic resources are needed to establish a joint household, then employment should be as influential for cohabitation as it is for marriage.

However, prior theory and research have considered marriage and cohabitation to be distinct institutions. Marriage has been seen as more fully institutionalized, meaning that marriage has a clearer and more rigid set of norms and expectations compared with cohabitation (Nock 1995). Because of the stronger normative expectations associated with marriage, including that marriage is more likely to come with the expectation of a long-term commitment, economic determinants are expected to play a larger role for marriage than for cohabitation.

Further, family scholars have theorized that contemporary marriage has a strong symbolic significance, as a capstone after economic milestones are achieved (Cherlin 2004; Edin and Kefalas 2005). Although couples could marry before they achieve economic stability, family scholars have argued that marrying with scant financial resources is stigmatized and perceived as risky because of the potential for economic insecurity to destabilize the relationship (Gibson-Davis et al. 2005). In contrast, financial strain appears to be at least one factor that precipitates cohabitation (Sassler 2004; Sassler and Miller 2011, 2017). Considering the connection between economic resources and cohabitation more broadly, Perelli-Harris et al. (2010) suggested that in many contexts, cohabitation really is an institution in response to economic insecurity.

Prior demographic research largely accords with this qualitative work. Although earnings and employment are important positive determinants of entry into marriage, the effects on cohabitation are much less pronounced. Several studies have found no association between employment and entry into cohabitation (Carlson et al. 2004; Manning et al. 2014; Raley 1996; Sassler and Goldscheider 2004). In this line of research, an important exception is the work of Clarkberg (1999), who found positive associations between earnings and cohabitation.

Prior research on gender differences in the association between employment and earnings and cohabitation is much more limited. In general, evidence suggests null effects of employment and earnings on cohabitation for women and men (Carlson et al. 2004; Manning et al. 2014; Raley 1996; Sassler and Goldscheider 2004), although again an important exception is Clarkberg (1999), who found that both men’s and women’s earnings positively predict cohabitation (vs. staying single) and that the effect is actually stronger for women than for men.

Educational Gradients in Union Formation

Marriage is strongly graded by educational attainment, with more highly educated individuals more likely to marry (Isen and Stevenson 2010; Thornton et al. 2007). Although this pattern has been true for men since at least the middle of the twentieth century, women’s educational attainment, at one time negatively associated with marriage, has now become a positive predictor (Goldstein and Kenney 2001; Schneider et al. 2018). In contrast, education is negatively related to cohabitation (Bumpass and Sweet 1989; Perelli-Harris and Lyons-Amos 2016; Sassler and Goldscheider 2004; Thornton et al. 1995). Education may be positively related to marriage because it is a signal of long-term economic prospects and stability (Sassler and Goldscheider 2004) or because education fosters a cultural approach to parenting that prioritizes investment in children through marriage (Lundberg and Pollak 2015). But, more simply, more highly educated men and women may also be more likely to marry because they possess the economic resources that have long been the normative prerequisites of marriageability (Ishizuka 2018).

In that case, the unequal distribution of economic resources by education in the United States has the potential to explain stratification in marriage entry along this axis of disadvantage. That is, one reason why men and women with a college degree may be more likely to marry than their less-educated counterparts is because education affords these individuals access to the economic resources that make them marriageable.

Job Quality and Union Formation

This body of empirical research has carefully shown that employment and earnings influence marriage and cohabitation and that those basic measures of work status may operate differently for men and for women. Both men’s and women’s employment and earnings have changed a great deal over the past several decades (i.e., Oppenheimer 1988; Wilson 1987), but these decades have also witnessed profound transformation in the quality of jobs beyond simply levels of employment or pay. Along several dimensions, job quality appears to have deteriorated and, importantly, stratified along the axis of educational attainment over the past several decades (Fligstein and Shin 2004; Kalleberg 2009).

Prior literature has identified several important dimensions of job quality. First, at the most basic level, the structure of compensation plays a large part in defining a “good job,” with salaried positions offering greater flexibility and often greater compensation than hourly positions. Being salaried offers the additional benefit of a fixed and predictable amount of earned income each month, whereas those paid hourly are subject to greater income volatility. Education is strongly related to compensation type: almost two-thirds of salaried workers have a college degree, and almost 80 % of hourly workers have less than a college degree (Brenan 2017).

Multiple job–holding is another indicator of job quality. Multiple job–holding has been shown to be motivated by the need for more income and to be an adaptive response to economic insecurity (Hipple 2010; Zangelidis 2014). Although those with lower levels of education can be expected to be more in need of supplemental income and more subject to economic insecurity, in fact, multiple job–holding appears more common for those with a college education (Lalé 2015). Multiple job–holding, then, is one notable exception to the typical pattern of association between higher education and better job quality.

Kalleberg (2011) noted that a defining feature of a good job is the provision of fringe benefits, such as health insurance, retirement accounts or pension, and paid time off. The availability of these benefits has declined for all workers over time, but especially for less-educated workers (Kalleberg 2011). For instance, employer provision of health insurance has declined, especially for those with less education (Farber and Levy 2000; Mishel et al. 2012). Retirement benefits have also changed substantially, shifting from defined benefit to defined contribution plans and eroding altogether as well (Hacker 2006). This change, too, has been stratified by education (Kalleberg 2009). Substantial stratification is also evident for paid time off: more-educated workers are much more likely to have paid vacation, paid sick days, and paid parental leave than their less-educated counterparts (Glynn et al. 2016).

In addition to nonmonetary compensation, job quality is also defined by work schedules. Scholars have long been concerned with nonstandard work hours (i.e., Presser 1999), but more recent research identified variable work hours as a key dimension of precarious work. Many hourly employees with low educational attainment now work varying numbers of hours each week, often with different starting and stopping times and on different days each week (Henly et al. 2006). This experience of unstable and unpredictable schedules is sharply stratified by education: many workers with higher socioeconomic status have more stable schedules or have more employee control over their work schedules—a desirable flexibility in contrast to instability (Schieman and Plickert 2008).

Finally, labor union membership may confer important economic benefits on workers in the near term and may also signal greater economic stability in the future (Schneider and Reich 2014). Although labor union membership has declined precipitously over the past several decades, approximately 11 % of workers remain union members (Dunn and Walker 2016). Those with some college or a college degree are more likely to be represented by a union than those with a high school degree only or less education, in keeping with the general pattern that higher educational attainment is associated with better job quality (Schmitt and Warner 2009).

In sum, by many measures, employment has become more precarious, and the educational divide in job quality appears pronounced and to have increased in many respects over time.

Consequences of Job Quality for Marriage and Cohabitation

Scholars of labor and employment have conjectured that the decline in job quality is likely to affect family formation and stability (Kalleberg 2009, 2011), and demographers have often noted that economic precarity may be a determinant of family formation (Cherlin 2015; Lichter et al. 2006). This perspective derives from the expectation that job quality may increase marriage through several pathways. First, to the extent that individuals with more economic resources—such as fringe benefits—are more marriageable, better-quality jobs may simply be more economically valuable and thus may increase marriageability. Second, in an era of increasingly precarious employment, having a job that offers fringe benefits or a stable schedule may serve as a marker of status and of the achievement normatively necessary for marriage (Cherlin 2004; Schneider 2011). Third, although income may convey information about current economic status, other aspects of job quality—such as fringe benefits or a union contract—may convey information about future prospects and economic stability that may be additionally valuable for marriage (Schneider and Reich 2014).

In contrast to marriage, prior literature suggests that employment—and thus, we would expect, job quality—is not strongly related to entry into a cohabitating union and that more economically precarious individuals may actually enter into cohabitation rather than into marriage either because they lack the normative prerequisites for marriage (Cherlin 2004) or because the economies of scale of coresidence are a partial solution to financial fragility (Sassler 2004; Sassler and Miller 2011).

These theoretical perspectives on why job quality might matter for marriage and might matter more for marriage than cohabitation are largely divorced from the particulars of given indicators of job quality. Rather, they suggest that holding a better job—as captured by a set of indicators—conveys information about one’s own and one’s potential partner’s long-term economic security and prospects.

But there are good reasons to expect that certain dimensions of job quality may matter more for marriage than for cohabitation. One reason is that the legal distinction between marriage and cohabitation makes some benefits more valuable to spouses than to unmarried partners. Specifically, employer-provided health and dental insurance is often structured in such a way as to directly benefit not just the employee but also spouses and dependents. Despite important exceptions for domestic partners (Polikoff 2012), this benefit will generally only have a mechanical transferability to married partners and children that might render it particularly valuable for marriage versus cohabitation. This idea finds credence in the literature on health insurance and divorce, where scholars have found evidence of “marriage lock” in which divorce is reduced because it would disrupt spousal health insurance (Sohn 2015).

Another reason is that some aspects of job quality are valuable only over a long time horizon. Specifically, the economic value of employer-provided retirement savings and of life insurance is generally realized only later in life, which is decades away from the time of marriages that we observe in the NLSY97. Although cohabitation could be a long-term relationship, demographic research makes clear that marriages are of significantly longer duration than cohabiting unions (Cherlin 2010) and that married partners expect to remain together for far longer than cohabitating partners (Nock 2005).

However, the demographic literature directly linking job quality with union formation is sparse. Using data from the NSLY79, Schneider and Reich (2014) found that union membership is a significant predictor of first marriage for men—a relationship they attributed to the health insurance benefits and longer job tenure that union membership can help workers secure. Drawing on data from the NLSY97, Kuo and Raley (2014) found that occupational autonomy is positively associated with marriage entry for women in their late 20s, but not for younger women and not for men, after controlling for earnings and employment. McClendon et al. (2014) found that paid parental leave is a significant predictor of marriage for women in the NLSY97 through round 13 (age 29). Piotrowski et al. (2015) took up the Japanese case and showed that employment in irregular jobs significantly reduces the risk of marriage entry relative to those working regular jobs, with the largest effects found for men.

Even less work has examined how job quality affects entry into cohabitation. Of notable exception, Oppenheimer (2003) used data from the NLSY79 and found that economic instability is negatively related to marriage but positively related to cohabitation. Similarly, Clarkberg (1999) found that having had more jobs and shorter job tenure are positively associated with cohabitation (although not with marriage) compared with remaining single. In a study measuring future earning potential, Xie et al. (2003) found that men’s expected future earnings predict entry into marriage but not into cohabitation. Further evidence that economic insecurity is a deterrent to marriage but not to cohabitation comes from research showing that credit card and student loan debt are associated with entry into cohabitation but with delays in marriage (Addo 2014).

In sum, these important exceptions aside, very little quantitative demographic research has evaluated how job quality shapes marriage and cohabitation entry. But given this forgoing work, we would expect the following:

  • Hypothesis 1: Individuals who hold higher-quality jobs will have a higher risk of marriage.

  • Hypothesis 2: Job quality will either negatively predict entry into cohabitation (compared with remaining single) or have a null relationship.

Prior research has shown that income and employment are positively related to marriage entry and, for at least the past several decades, have had a positive association with marriage for both men and women. Theory and a limited amount of prior empirical research also suggest that job quality matters for marriage for both men and women. We hypothesize that job quality should be positively related to marriage for both men and women.

However, particular aspects of job quality may have stronger associations with marriage for women than for men. One of these is having access to a standard work schedule. Women retain primary responsibility for domestic production and for childcare (Lyonette and Crompton 2014), and prior research has shown that nonstandard and unstable and unpredictable work schedules make it particularly difficult to fulfill those gendered responsibilities (Carrillo et al. 2017; Henly and Lambert 2014; Presser 2005). We might expect then that having a standard day schedule would have a stronger positive association with marriage for women than for men. In addition, we predict that family-friendly benefits—such as parental leave, childcare, and schedule flexibility—may be particularly valuable for women, given that women typically shoulder a disproportionate share of caregiving responsibilities.

  • Hypothesis 3: Both men’s and women’s job quality will be positively associated with entry into marriage. However, standard work schedules and family-friendly benefits will more strongly predict entry into marriage for women than for men.

Given the paucity of evidence that economic resources are significantly linked to cohabitation, we do not offer a hypothesis on how gender may interact with economic influences on cohabitation. Rather, we consider our comparison of the relationship between job quality and cohabitation for men and for women to be an exploratory analysis, for which any significant findings would need to be replicated in future research.

Finally, there are well-documented educational gradients in marriage entry and cohabitation and, as discussed earlier, there are also steep gradients in job quality by education. Higher educational attainment, especially a college degree, is associated with better job quality on all indicators with the exception of multiple job–holding. Higher educational attainment is also positively associated with entry into marriage and with relatively lower levels of nonmarital cohabitation. Job quality may confer both real and perceived economic security, and features of job quality such as fringe benefits can be considered economic resources. Given the robust literature showing that employment and earnings encourage marriage, we could reasonably expect differences in job quality by education to contribute to the higher rates of marriage for those with college education who enjoy the best-quality jobs. Therefore, we expect that accounting for a richer set of job quality measures will diminish educational differences in marriage rates.

  • Hypothesis 4: The greater likelihood of marriage for more highly educated individuals will be partially accounted for by differences in job quality.

Data and Methods

Data

We investigate the relationship between job quality and union formation using rounds 1–16 of the National Longitudinal Survey of Youth 1997 (NLSY97). The NLSY97 has followed a nationally representative sample of 8,984 youth born 1980–1984 (with an oversample of Hispanic or Latino and black youth) with annual interviews from 1997 to 2011 and biennial interviews thereafter.

The NLSY97 provides a unique opportunity to study the impact of job quality on marriage and cohabitation entry. Rich data on employment have been collected at fine-grained time intervals, allowing for an examination of the effects of labor market position that extends beyond employment and earnings to a range of job quality measures: fringe benefits, compensation structure, union membership, multiple job–holding, and schedule regularity. Although respondents in the sample have only reached their early 30s, most have already entered first cohabitation or first marriage (approximately 74 %).

Our sample includes all never-married/never-cohabited respondents over the age of 18 with nonmissing data through round 16 of data collection (retention rate at round 16 of the study is 79.5 %). Respondents enter the risk set at age 18 because labor market participation of youth is still quite low prior to exiting high school: only 49 % of 17-year-old NLSY97 respondents were employed, and only 58 % were employed or looking for work. The NLSY provides weekly employment data but monthly data on union formation and school enrollment, so our unit of analysis is the person-month. We draw on data from 4,162 men and 3,735 women. We observe a total of 336,535 person-months for men and 257,732 person-months for women before entry into either marriage or cohabitation. We also draw on a sample of 469,718 person-months for men and 402,448 person-months for women observed before entry into marriage (but that could be observed following a cohabitation).

Measures

Union Formation

We define two dependent variables that we employ in a competing-risks analysis of union formation: entry into first marriage directly from noncoresidential status and entry into first cohabiting union from noncoresidential status. The NLSY provides data on the calendar year and month in which each applicable respondent entered a first marriage or first cohabiting union. For the competing-risks analysis, we code a variable to be 0 for all person-months prior to union entry, 1 for the month in which a first entry to marriage from single status took place, and 2 for the month in which a first entry to cohabitation from single status took place, with respondents censored after the event has occurred. We also construct a dichotomous measure of entry into first marriage where respondents are not censored at cohabitation and are coded as entering marriage regardless of whether they transition from noncoresidential status or from a cohabitation. Individuals who neither marry nor cohabit are censored at last interview/observation.

Table 1 summarizes these transitions for the observed sample, separately for men and women. We see that 14 % of men and 16 % of women in the sample transitioned directly to marriage from a noncoresidential state. Substantially larger shares—27 % of men and 33 % of women—transitioned to marriage following cohabitation, and roughly 29 % of both men and women transitioned to cohabitation but were not observed to enter a marriage. The remaining 30 % of men and 22 % of women never married or cohabited in the observation period.

Employment and Job Quality

To capture variation in job quality, we construct a set of time-varying individual-level measures of economic characteristics. In each survey round, the NLSY collects a battery of information on all jobs held by the respondent between the current and the previous survey round. We use the start, stop, and gap dates provided by the NLSY97 for each job to create weekly measures of employment data. For weeks in which respondents worked at more than one job, job characteristics from the respondent’s “main job,” identified by the NLSY, are used.1 By averaging employment measures across weeks within each month, we collapse these weekly data to person-months in preparation to be merged with the monthly data on union formation. However, by design, these weekly (and monthly) arrays do not capture any short-term within-job variation in our measures of job quality. Within-person variation in job quality is observed only when respondents change jobs or when new information on a job is collected in a subsequent round of the NLSY.

Our basic measure of employment captures respondents’ earnings, where respondents who are not working have zero earnings. This measure of earnings is driven in part then by whether the young adults in our sample are employed. Because earnings and employment are therefore highly correlated (ρ = .95), we do not include a separate measure of employment, and we interpret the earnings variable as a measure of both employment and earnings. The value of earnings is defined based on the NLSY-calculated effective hourly rate of compensation, which includes earnings from overtime and performance pay. This hourly rate is multiplied by the number of hours worked to obtain weekly earnings. Earnings are inflation-adjusted to be expressed in 2013 dollars and logged.

We leverage the rich data on jobs in the NLSY97 to examine job quality along five dimensions: (1) multiple job–holding, (2) work schedule type, (3) compensation structure, (4) coverage by union contract, and (5) access to employer-provided benefits, including health insurance, parental leave, and flexible scheduling.

First, we use the detailed weekly job calendar data to construct a measure of whether the respondent holds more than one job at a time. Second, we code three types of work schedules that deviate from a standard day shift: (1) nonstandard shifts involving work in the evening, at night, or on the weekend, (2) a split or rotating shift, or (3) an irregular schedule. Third, the NLSY97 collects hourly pay rate data for those working at a job that pays hourly, and we use this variable to create an indicator of whether a job paid an hourly wage rather than a salary. Fourth, the NLSY97 also provides a dichotomous measure of whether a job is “covered by a contract that was negotiated by a union or employee association” that we use to define union jobs.

Finally, we examine the fifth dimension of job quality: access to employer-provided benefits. For each job listed by the respondent, the NLSY asks whether the employer provides a battery of fringe benefits: health, life, and dental insurance; paid and unpaid maternity leave; childcare; a flexible work schedule; a retirement plan; tuition reimbursement; and an employee stock ownership plan. We combine this series of dichotomous indicators for job benefits into a simple additive scale that has high reliability (Cronbach’s alpha = .90). This scale is referred to as the fringe benefits scale in our tables and discussion of results. We also estimate models that include each of the benefit items separately, as well as models that examine an abridged additive scale that sums what we term family-friendly benefits: paid or unpaid parental leave, childcare, and schedule flexibility. As with earnings, we capture benefits reported through the NLSY-identified main job.

Education

We use the NLSY97’s monthly educational history data to calculate a school enrollment indicator and a five-level variable that measures the respondent’s highest attained degree: completed less than 12 years of education, earned a high school diploma, attended some college but did not earn a degree, earned an associate’s degree, or earned a bachelor’s degree or more education. The results are substantively similar when we combine the categories of some college and associate’s degree.

Control Variables

Our analyses account flexibly for life course effects by including a linear and quadratic term for the age of respondents in months, calculated as the difference between a respondent’s birth month and the month of analysis. We also use dummy variables for the two recessions that occurred during the period of study—the recession from March to November of 2001 and the Great Recession from December 2007 to June 2009—to address any effect these economic shocks may have had on union formation. We also include measures of the number of children in the household and whether the respondent has recently experienced the birth of a child. Table 1 shows that almost one-half of men and 43 % of women reported a birth that preceded marriage or cohabitation. These time-varying characteristics are summarized for our analytic sample in Table 2.

We measure race/ethnicity using a four-category variable indicating whether a respondent identifies as black, non-Hispanic; white, non-Hispanic; another race non-Hispanic; or Hispanic. Finally, we control for family background using a measure of respondents’ mothers’ educational attainment (collected at baseline).

Method

We estimate two main statistical models. First, we estimate a discrete-time, competing-risks event history model of entry into first marriage or first cohabitation to assess the role that employment precarity plays in union formation. The model is estimated with multinomial logistic regression. Respondents are censored either when they enter first marriage directly from a noncoresidential status or when they enter a first cohabiting union directly from a noncoresidential status. The estimated coefficients are the risk of entry into each of the two union types relative to remaining in a noncoresidential state. We also present tests of the significance of differences in the estimated coefficients for entry into marriage versus cohabitation. Second, we estimate a discrete-time event history model of entry into first marriage, irrespective of whether the respondent transitions from a noncoresidential state or from a cohabiting union. In this model, respondents are censored only at first marriage—not at first cohabitation—and the model is estimated with logistic regression.

In both models, standard errors are clustered at the level of the respondent. All time-varying covariates, with the exception of age, are lagged in order to ensure that they are measured before the outcome events occur. In accord with our theoretical model, which assumes that sustained exposure to employment and job quality influence union transitions, we use 12-month averages of employment and job quality covariates. To ensure appropriate time-ordering with respect to union formation, we lag these predictors by 6 months, using average monthly employment and job quality data over the period 17 months to 6 months prior to the month the outcome is measured.2 Following NLSY97 official guidance, we do not weight the analyses (National Longitudinal Surveys of Youth n.d.).

Our analysis proceeds in three steps. In the first step, we focus on the association between our measures of job quality and union formation. Here, we begin by estimating the following model:
logpitupit0=α+μREC+γXi+δXit+βEit+δJQit.
1
We predict the log odds of union formation, either through first marriage or first cohabitation, with a discrete-time competing-risks model. The model includes the time-varying measure of earnings/employment (E) and measures of job quality (JQ), in addition to controls for time-invariant (Xi) and time-varying (Xit) individual socioeconomic and demographic characteristics and indicators of the two recessions (REC). Because the measures of job quality are highly correlated, we estimate the model five times, entering each of the job quality measures separately (along with the other covariates). We also estimate the models separately for men and women. We then repeat this same approach but with a model that does not treat cohabitation as a competing risk and simply focus on predicting the log odds of marriage, with a time-varying indicator of cohabitation status.3

In the second step of the analysis, we test for gender differences. We cannot easily test for differences in coefficients across separately estimated models or for interactions within a pooled model using the aforementioned logistic regression models (Mood 2010). While the competing-risks models rely on multinomial logistic regression, the models of entry into marriage from either noncoresidential status or cohabitation can be estimated with a linear probability model (LPM), which does allow for interpretation of interaction coefficients (Angrist and Pischke 2009). We then test for differences by gender in the association between job quality and entry into marriage using interactions between gender and job quality in the LPM.

Third, the same limitations of the logistic regression model that make it difficult to compare coefficients on separately estimated models for men and women also prevent us from easily comparing coefficients on the same measures across nested models. However, to test Hypothesis 4, we need to compare whether the coefficients on our set of indicators for educational attainment are significantly attenuated when we account for our measures of job quality. To make these comparisons, we use the method developed by Karlson et al. (2012), which allows us to make unbiased estimates of the extent to which our job quality measures mediate the association between union formation and education. Here, we assess how much the controlled total effect of education (adjusted for our controls as well as earnings/employment) is mediated by the inclusion of the full set of job quality measures. We present coefficients on education that summarize the total effect, direct effect, and indirect effect. We also show the percentage of the total effect that is attributable to the indirect pathway of job quality and tests of the statistical significance of that mediation.

Results

Job Quality and Entry Into First Marriage

We begin by examining the relationship between our measures of job quality and union formation. The key results are presented in Table 3, with each of the panels (numbered 0–5) containing a separate regression model specification, varying the key independent variable that is included. The first two columns show the estimates for men from the competing-risks models. When the measure of job quality significantly affects the risk of transitioning to marriage versus remaining in noncoresidential status (with cohabitation as a competing risk) or affects the risk of transitioning to cohabitation versus remaining in noncoresidential status (with marriage as a competing risk), the coefficients are marked with asterisks. We underline the coefficients when the association between entry into marriage versus cohabitation is significantly different. The third column shows how job quality is associated with entry into marriage from either noncoresidential status or from cohabitation. We present identical models for women in the fourth through sixth columns.

As predicted by Hypothesis 1, men’s job characteristics predict entry into marriage. Although neither multiple job–holding nor union coverage alters the risk of first marriage, we see that men who work a nonstandard evening, night, or weekend shift, or a split or rotating shift are significantly less likely to marry compared with men who work a standard daytime shift, when cohabitation is a competing risk.

We also see that, controlling for a set of background characteristics and for earnings/employment, the fringe benefits scale is positively and significantly associated with entry into first marriage (b = 0.6561, p < .01). Figure 1 (panel a) plots the predicted probability of marriage entry in a given person-month by the extent of men’s fringe benefits. Men who have the mean number of benefits have about a .0014 probability of transitioning to first marriage in a given month. Men with a standard deviation greater number of benefits are 15 % more likely to marry in a given month.

We also find support for Hypothesis 2. In Table 3, for men, none of the measures of job quality are significantly associated with entry into cohabitation (with marriage as a competing risk). Further, the differences between these nonsignificant coefficients for cohabitation and the significant coefficients for marriage are in fact significant in the case of nonstandard schedules and fringe benefits (denoted by the underlined coefficients).

In Table 4, across the various benefit types, we find some support for our expectation that health/dental insurance and retirement benefits would be more strongly related to marriage than cohabitation. For men, we see that having health insurance and dental insurance is significantly related to marriage but not to cohabitation and that the difference between the two is statistically significant. We see a similar relationship for having life insurance. The results for retirement do not fully accord with our hypothesis. Although retirement benefits are significantly related to marriage, they are also significantly related to cohabitation. The coefficient is substantially larger for marriage but is not significantly different from the association with cohabitation. Of all the specific benefits, only retirement and unpaid parental leave are positively related to cohabitation (vs. remaining in a noncoresidential state), while schedule flexibility is negatively related to entering cohabitation.

The last three columns of Table 3 present similar results for women. As predicted by Hypothesis 1, women who work nonstandard schedules are significantly less likely to marry (regardless of whether cohabitation is a competing risk), as are women who work split or rotating shifts (when cohabitation is not a competing risk). Similar to men, being paid hourly significantly reduces the risk of marriage for women.

We also find strong associations between fringe benefits and entry into marriage for women. Women whose jobs provide more fringe benefits, captured by higher values on the fringe benefits scale, have a significantly higher likelihood of transitioning into first marriage (b = 0.7728, p < .001) when we control for respondent characteristics and for earnings/employment. Similar to men, moving up by a standard deviation on the fringe benefits scale from the mean level of fringe benefits is associated with an approximately 19 % increase in the probability of first marriage in a given person-month (Fig. 1, panel b).

The evidence related to Hypothesis 2, on cohabitation, is perhaps more interesting for women than for men. First, Table 3 shows that although schedule type and benefits are significantly related to marriage, they are not significantly associated with cohabitation. Further, the differences between the significant association with marriage and the null association with cohabitation are themselves statistically significant for nonstandard schedule and benefits. Also notable is that although a split shift is negatively, but not significantly, associated with marriage for women (vs. staying in a noncoresidential state) and positively, but not significantly, associated with cohabitation (vs. staying in a noncoresidential state), having a split shift is a significant predictor of transitioning to cohabitation versus transitioning to marriage. The same is true of not having union coverage.

In Table 4, as for men, we also find some support for our hypothesis that having health and dental insurance and life insurance and retirement benefits would be more strongly associated with marriage than with cohabitation for women. We see that health and dental insurance benefits and retirement benefits are all positively and significantly related to marriage and that this positive association is significantly stronger than for cohabitation.

The full set of coefficients on earnings/employment from all models presented in Tables 3 and 4 are shown in Tables A1 and A2 in the online appendix, which show that earnings/employment is a consistent significant predictor of union formation net of job quality measures across model specifications. As a robustness check, we estimate all models in Tables 3 and 4 only on those who were employed in the prior year, and we find consistent results.

Gender Differences in Associations Between Job Quality and Union Formation

As reported in Tables 3 and 4, we find strong evidence of associations between job quality and marriage entry for both men and women and little evidence of associations between job quality and cohabitation for either men or women. Most of these differences in the power of job quality for predicting marriage versus cohabitation are statistically significant.

We also test Hypothesis 3 by assessing whether the positive associations between job quality and marriage were significantly different between men and women. We find no evidence of any interactions between gender and job quality. There are null effects for multiple job–holding for both men and women and although we hypothesized that nonstandard or irregular shifts might have more negative consequences for women’s marriage than for men’s, we find no statistically significant differences in the estimated coefficients by gender and no significant differences by gender in the coefficients on either fringe benefits or being paid a salary. Union membership is not a significant predictor of marriage in the models that are estimated separately by gender (Table 3), but when we pool the data, union coverage is positively and significantly associated with marriage for both men and women (results not shown).

Educational Attainment and Marriage Entry

We next turn to the relationship between educational attainment and first marriage and to the role of job quality in explaining any educational advantage in marriage entry. Table 5 presents the result of a decomposition analysis of the effect of education on marriage entry based on estimates from the competing-risks models (panels A and C) and the event history models that examine all transitions to marriage (panels B and D). Here, we use the Karlson et al. (2012) method to decompose the total effects of education (after conditioning on our controls and our measure of employment/earnings) into an indirect effect of job quality and the remaining direct effect. Panels A and B report the decomposition for men, and panels C and D for women. Although we use the language of “effects” in this section, we caution that these remain associational analyses.

Comparing across the top rows of panels A and B (the total effect), we see a strong and significant educational gradient in first marriage for men. The third row of panels A and B shows how much of the educational effect is accounted for by our five measures of job quality. These measures play a minor role in accounting for the differences in marriage between those with less than a high school degree and those who completed high school: just 1.5 % in the competing-risks models and 2.7 % in the marriage-only models, both of which are not significant. Job quality plays a somewhat larger role in accounting for differences between men with some college and less than high school—5.7 % and 6.5 %—a statistically significant share. However, job quality plays a much more important role in accounting for differences between men with an associate’s degree (10.4 % and 9.8 %) and those with less than high school. This attenuation is statistically significant at the p < .001 level. Job quality plays the largest role, however, in explaining the significant marriage advantage of men with at least a bachelor’s degree. Here, accounting for job quality explains roughly one-fifth of the advantage (p < .001). In supplemental models, we find that fringe benefits and job schedules are the primary drivers of this mediation.

Panels C and D of Table 5 present parallel results for women. As for men, educational gradients in entry into marriage are strong, as reflected by the increasing and significant coefficients across the top rows of both panels: the total effects. Although job quality does not significantly mediate the marriage advantage of women with a high school degree versus less than high school, we find a statistically significant mediation of the marriage advantage of women with some college, an associate’s degree, or a bachelor’s degree or more. As was the case for men, this attenuation is most pronounced for the most highly educated women: approximately 25 % of the advantage in the competing-risks model (panel C) and approximately 15 % in the models that allow for entry into marriage from cohabitation (panel D).

In all, Hypothesis 4 is strongly supported. Job quality substantially and significantly mediates the association between education and marriage, explaining on the order of 20 % to 25 % of the marriage gap between college-educated men and women and those with less than a high school education.

Discussion

Economic resources have long been appreciated as an important determinant of marriage, but prior research has mainly focused on narrow measures of economic circumstances—employment and earnings—and far less research has considered the influence of job quality. Yet, we know from research on the changing labor market that aspects of job quality—such as fringe benefits, work schedules, and whether workers are unionized or are paid hourly or salaried—are also important job features that affect financial security and work–life balance and have the potential to influence union transitions. However, very little prior research has examined this relationship. Our study takes advantage of rich data from the National Longitudinal Survey of Youth 1997, which follows the 1980–1984 birth cohort from their teenage years into their 30s, to address the question of how job quality matters for men and women’s transitions to marriage and cohabitation.

We find support for our hypothesis that better job quality encourages marriage. After taking into account employment status and earnings, we find that men and women employed in jobs offering more fringe benefits are significantly more likely to marry. Scholars and advocates have recently turned their attention to another dimension of precarious employment: variable work schedules (i.e., Henly et al. 2006; Lambert 2008). We show that split shifts, rotating schedules, and nonstandard schedules are negatively associated with marriage, providing rare quantitative support for the intuition that such practices have negative consequences for families. In all, our findings are consistent with the conclusions that poor job quality is an impediment to marriage and that job quality influences marriage entry above and beyond one’s level of earnings.

Prior demographic research has debated whether women’s economic resources would be positively or negatively related to marriage (Becker 1981; Oppenheimer 1988). Empirical work shows fairly conclusively that the association is positive for women in recent cohorts (Sweeney 2002), but the evidence has continued to show that men’s resources are more strongly related to marriage than women’s (Shafer and James 2013). We do not find evidence of that inequality in association with respect to job quality. Instead, women’s job quality appears to be just as strongly associated with marriage entry as men’s.

Recent decades have witnessed a striking divergence in marriage behavior across education groups, with declines in marriage concentrated among those without a college degree. The decline in labor market opportunities for those with less than a college education is a leading explanation for this divergence in marriage behavior across education groups. Here again, we find that job quality matters. The educational differences in entry to marriage are partially explained by differences across education groups in job quality as measured by fringe benefits, stable and regular schedules, and getting paid a salary. Those with higher levels of educational attainment are more likely to enjoy better-quality jobs (e.g., jobs with better fringe benefits and more stable schedules and earnings), and these job features help to explain why those with higher levels of educational attainment are more likely to marry. Specifically, we find that differences in job quality accounted for between 20 % and 25 % of the marriage advantage enjoyed by men and women with at least a bachelor’s degree over those with less than a high school diploma.

These differences in job quality matter much less for cohabitation. In competing-risks models, for both men and women, our measures of job quality—work schedule type, fringe benefits, and unionization—were simply not predictive of cohabitation (with the exception that women who are paid hourly are less likely to transition to either marriage or cohabitation). We do not find that low job quality is a barrier to cohabitation as it is for marriage, nor do we find much evidence that low job quality encourages cohabitation. In all, although precarious employment may slow marriage, it does not appear to significantly shape the formation of cohabiting unions.

The NLSY97 provides an opportunity to examine rich and nuanced job characteristics alongside standard measures of education and earnings and to estimate relationships between economic predictors that temporally precede marital and cohabitation outcomes. Nevertheless, men and women may alter their work effort and employment choices in anticipation of marital and cohabitation transitions. Therefore, a limit of our analysis is that we cannot take into account these anticipatory effects on economic circumstances, which may contribute to the relationship between economic predictors and marriage and cohabitation outcomes. If men and women increase their work effort and select into higher-quality jobs in anticipation of a planned marriage, for instance, then our models will overestimate the influence of job quality on marriage. Additionally, although we are able to extend the age range for analysis beyond that of previous research using the NLSY97, we still do not observe unions that form after age 34. Finally, we have drawn on a rich set of measures of job quality, but we would ideally be able to also capture contingent employment relationships, control over work tasks, or perceived job insecurity (Kalleberg 2009). We would also ideally be able to capture the “regular unpredictability” that appears to characterize many hourly jobs in terms of week-to-week variation in total hours worked and in work schedules (Schneider and Harknett forthcoming). The NLSY97 has added new measures of these emergent precarious labor practices in recent waves, but these measures are not available prior to 2011. Future research identifying data sources that capture such within-job instability could very usefully examine whether that form of precarity is associated with union formation for men and women.

In sum, we demonstrate that job quality matters for marriage for both men and women. We also find clear evidence that job quality plays a significant role in explaining educational divides in marriage entry, thereby lending credence to the predictions that the bifurcation of the labor market into “good jobs” and “bad jobs” helps to explain class differences in marriage behavior. Thus, understanding the mechanisms that sort individuals into bad jobs or provide ladders to better job opportunities is essential for understanding inequality in both economic and union formation domains.

Acknowledgments

We gratefully acknowledge grant support from the Washington Center for Equitable Growth (Award No. 39092) and the UC Berkeley Institute for Research on Labor and Employment. A previous version of this article was presented at the 2017 Annual Meeting of the Association for Public Policy Analysis and Management.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Notes

1

For employed person-weeks with no main job indicated, we designate the job with the largest number of hours as the main job.

2

We also check the robustness of the results to using a lag that measured the covariates at a point in time (rather than the 12-month average of the period 6–17 months prior). We test using lags that were 12 months prior and 6 months prior to the event month. These models show very similar results. However, for men, the fringe benefits scale is a weaker predictor of marriage when we use 12-month lags but a stronger predictor of marriage when we use 6-month lags compared with the preferred models. Additionally, for both the 12- and 6-month lag, the benefits scale coefficient predicts cohabitation entry more strongly than in the main models.

3

We also reestimate the models restricted to respondents who were cohabiting in the month prior to measurement of the dependent variable. The results are unchanged despite a large reduction in sample size: working a split/rotating shift remains negatively associated with marriage for women, hourly work remains negatively associated with marriage for men, and fringe benefits remain positively associated with marriage for both men and women. These predictors of first marriage entry are similar whether respondents enter marriage from cohabitation or not.

References

Addo, F. (
2014
).
Debt, cohabitation, and marriage in young adulthood
.
Demography
,
51
,
1677
1701
. 10.1007/s13524-014-0333-6.
Angrist, J. D., & Pischke, J.-S. (
2009
).
Mostly harmless econometrics: An empiricist’s companion
.
Princeton, NJ
:
Princeton University Press
.
Becker, G. S. (
1981
).
A treatise on the family
.
Cambridge, MA
:
Harvard University Press
.
Boushey, H. (
2016
).
Finding time: The economics of work-life conflict
.
Cambridge, MA
:
Harvard University Press
.
Brenan, M. (
2017
).
Hourly workers unhappier than salaried on many job aspects
(Gallup research report). Retrieved from https://news.gallup.com/poll/216746/hourly-workers-unhappier-salaried-job-aspects.aspx
Bumpass, L. L., & Sweet, J. A. (
1989
).
National estimates of cohabitation
.
Demography
,
26
,
615
625
. 10.2307/2061261.
Bureau of Labor Statistics, U.S. Department of Labor
. (
2015
).
National Longitudinal Survey of Youth 1997 cohort, 1997–2013 (Rounds 1–16)
.
Chicago, IL
:
National Opinion Research Center, the University of Chicago [Producer]
;
Columbus
:
Center for Human Resource Research, The Ohio State University [Distributor]
.
Burgess, S., Propper, C., & Aassve, A. (
2003
).
The role of income in marriage and divorce transitions among young Americans
.
Journal of Population Economics
,
16
,
455
475
. 10.1007/s00148-003-0124-7.
Burstein, N. (
2007
).
Economic influences on marriage and divorce
.
Journal of Policy Analysis and Management
,
26
,
387
429
. 10.1002/pam.20257.
Carlson, M., McLanahan, S., & England, P. (
2004
).
Union formation in fragile families
.
Demography
,
41
,
237
261
. 10.1353/dem.2004.0012.
Carrillo, D., Harknett, K., Logan, A., Luhr, S., & Schneider, D. (
2017
).
Instability of work and care: How work schedules shape child-care arrangements for parents working in the service sector
.
Social Service Review
,
91
,
422
455
. 10.1086/693750.
Cherlin, A. J. (
2004
).
The deinstitutionalization of American marriage
.
Journal of Marriage and Family
,
66
,
848
861
. 10.1111/j.0022-2445.2004.00058.x.
Cherlin, A. J. (
2010
).
Demographic trends in the United States: A review of research in the 2000s
.
Journal of Marriage and Family
,
72
,
403
419
. 10.1111/j.1741-3737.2010.00710.x.
Cherlin, A. J. (
2015
).
Labor’s love lost: The rise and fall of the working-class family in America
.
New York, NY
:
Russell Sage
.
Clarkberg, M. (
1999
).
The price of partnering: The role of economic well-being in young adults’ first union experiences
.
Social Forces
,
77
,
945
968
. 10.1093/sf/77.3.945.
Davis, K., & Blake, J. (
1956
).
Social structure and fertility: An analytic framework
.
Economic Development and Cultural Change
,
4
,
211
235
. 10.1086/449714.
Dunn, M., & Walker, J. (
2016
).
BLS spotlight on statistics: Union membership in the United States
(Spotlight on Statistics Report No. 9-2016).
Washington, DC
:
U.S. Department of Labor, Bureau of Labor Statistics
.
Edin, K., & Kefalas, M. (
2005
).
Promises I can keep: Why poor women put motherhood before marriage
.
Berkeley
:
University of California Press
.
Ellwood, D. T., & Jencks, C. (
2004
).
The uneven spread of single-parent families: What do we know? Where do we look for answers?
. In K. M. Neckerman (Ed.),
Social inequality
(pp.
3
70
).
New York, NY
:
Russell Sage
.
Farber, H., & Levy, H. (
2000
).
Recent trends in employer-sponsored health insurance coverage: Are bad jobs getting worse?
Journal of Health Economics
,
19
,
93
119
. 10.1016/S0167-6296(99)00027-2.
Fligstein, N., & Shin, T.-J. (
2004
).
The shareholder value society: A review of the changes in working conditions and inequality in the United States, 1976 to 2000
. In K. M. Neckerman (Ed.),
Social inequality
(pp.
401
432
).
New York, NY
:
Russell Sage
.
Gibson-Davis, C., Edin, K., & McLanahan, S. (
2005
).
High hopes but even higher expectations: The retreat from marriage among low-income couples
.
Journal of Marriage and Family
,
67
,
1301
1312
. 10.1111/j.1741-3737.2005.00218.x.
Glynn, S. J., Boushey, H., & Berg, P. (
2016
).
Who gets time off? Predicting access to paid leave and workplace flexibility
.
Washington, DC
:
Center for American Progress
.
Goldstein, J., & Kenney, C. (
2001
).
Marriage delayed or marriage forgone? New cohort forecasts of first marriage for U.S. women
.
American Sociological Review
,
66
,
506
519
. 10.2307/3088920.
Hacker, J. S. (
2006
).
The great risk shift: The new economic insecurity and the decline of the American dream
.
New York, NY
:
Oxford University Press
.
Harknett, K., & Kuperberg, A. (
2011
).
Education, labor markets and the retreat from marriage
.
Social Forces
,
90
,
41
63
. 10.1093/sf/90.1.41.
Harknett, K., & McLanahan, S. (
2004
).
Racial and ethnic differences in marriage after the birth of a child
.
American Sociological Review
,
69
,
790
811
. 10.1177/000312240406900603.
Henly, J. R., & Lambert, S. J. (
2014
).
Unpredictable work timing in retail jobs: Implications for employee work–life conflict
.
Industrial and Labor Relations Review
,
67
,
986
1016
.
Henly, J. R., Shaefer, H. L., & Waxman, E. (
2006
).
Nonstandard work schedules: Employer- and employee-driven flexibility in retail jobs
.
Social Service Review
,
80
,
609
634
. 10.1086/508478.
Hipple, S. F. (
2010
).
Multiple jobholding during the 2000s
.
Monthly Labor Review
,
133
(
7
),
21
32
.
Isen, A., & Stevenson, B. (
2010
).
Women’s education and family behavior: Trends in marriage, divorce and fertility
(NBER Working Paper No. 15725).
Cambridge, MA
:
National Bureau of Economic Research
.
Ishizuka, P. (
2018
).
The economic foundations of cohabiting couples’ union transitions
.
Demography
,
55
,
535
557
. 10.1007/s13524-018-0651-1.
Kalleberg, A. L. (
2009
).
Precarious work, insecure workers: Employment relations in transition
.
American Sociological Review
,
74
,
1
22
. 10.1177/000312240907400101.
Kalleberg, A. L. (
2011
).
Good jobs, bad jobs: The rise of polarized and precarious employment systems in the United States, 1970s–2000s
(American Sociological Association’s Rose Series in Sociology).
New York, NY
:
Russell Sage
.
Karlson, K. B., Holm, A., & Breen, R. (
2012
).
Comparing regression coefficients between same-sample nested models using logit and probit: A new method
.
Sociological Methodology
,
42
,
286
313
. 10.1177/0081175012444861.
Kuo, JC-L, & Raley, R. K. (
2014
).
Is it all about money? Work characteristics and women’s and men’s marriage formation in early adulthood
.
Journal of Family Issues
,
37
,
1046
1073
. 10.1177/0192513X14530973.
Lalé, E. (
2015
).
Multiple jobholding over the past two decades
.
Monthly Labor Review, April
. https://doi.org/10.21916/mlr.2015.7
Lambert, S. (
2008
).
Passing the buck: Labor flexibility practices that transfer risk onto hourly workers
.
Human Relations
,
61
,
1203
1227
. 10.1177/0018726708094910.
Lichter, D. T., McLaughlin, D. K., Kephart, G., & Landry, D. J. (
1992
).
Race and the retreat from marriage: A shortage of marriageable men?
American Sociological Review
,
57
,
781
799
. 10.2307/2096123.
Lichter, D. T., Qian, Z., & Mellott, L. M. (
2006
).
Marriage or dissolution? Union transitions among poor cohabiting women
.
Demography
,
43
,
223
240
. 10.1353/dem.2006.0016.
Lundberg, S., & Pollak, R. (
2015
).
The evolving role of marriage: 1950–2010
.
Future of Children
,
25
(
2
),
29
50
. 10.1353/foc.2015.0011.
Lyonette, C., & Crompton, R. (
2014
).
Sharing the load? Partners’ relative earnings and the division of domestic labour
.
Work, Employment and Society
,
29
,
23
40
. 10.1177/0950017014523661.
Manning, W., Smock, P., Dorius, C., & Cooksey, E. (
2014
).
Cohabitation expectations among young adults in the United States: Do they match behavior?
Population Research and Policy Review
,
33
,
287
305
. 10.1007/s11113-013-9316-3.
McClendon, D., Kuo, J. C.-L., & Raley, K. (
2014
).
Opportunities to meet: Occupational education and marriage formation in young adulthood
.
Demography
,
51
,
1319
1344
. 10.1007/s13524-014-0313-x.
McLanahan, S. (
2004
).
Diverging destinies: How children are faring under the second demographic transition
.
Demography
,
41
,
607
627
. 10.1353/dem.2004.0033.
Mishel, L., Bivens, J., Gould, E., & Shierholz, H. (
2012
).
The state of working America
(Economic Policy Institute Book, 12th ed.).
Ithaca, NY
:
Cornell University Press
.
Mood, C. (
2010
).
Logistic regression: Why we cannot do what we think we can do, and what we can do about it
.
European Sociological Review
,
26
,
67
82
. 10.1093/esr/jcp006.
National Longitudinal Surveys of Youth
. (n.d.).
NLSY97 sample weights and design effects
. Retrieved from https://www.nlsinfo.org/content/cohorts/nlsy97/using-and-understanding-the-data/sample-weights-design-effects
Nock, S. (
1995
).
A comparison of marriages and cohabiting relationships
.
Journal of Family Issues
,
16
,
53
76
. 10.1177/019251395016001004.
Nock, S. (
2005
).
Marriage as a public issue
.
Future of Children
,
15
(
2
),
13
32
. 10.1353/foc.2005.0019.
Oppenheimer, V. (
1988
).
A theory of marriage timing
.
American Journal of Sociology
,
94
,
563
591
. 10.1086/229030.
Oppenheimer, V. (
2003
).
Cohabiting and marriage during young men’s career development process
.
Demography
,
40
,
127
149
.
Oppenheimer, V., Kalmijn, M., & Lim, N. (
1997
).
Men’s career development and marriage timing during a period of rising inequality
.
Demography
,
34
,
311
330
. 10.2307/3038286.
Perelli-Harris, B., & Lyons-Amos, M. (
2016
).
Partnership patterns in the United States and across Europe: The role of education and country context
.
Social Forces
,
95
,
251
281
. 10.1093/sf/sow054.
Perelli-Harris, B., Sigle-Rushton, W., Kreyenfeld, M., Lappegård, T., Keizer, R., & Berghammer, C. (
2010
).
The educational gradient of childbearing within cohabitation in Europe
.
Population and Development Review
,
36
,
775
801
. 10.1111/j.1728-4457.2010.00357.x.
Piotrowski, M., Kalleberg, A., & Rindfuss, R. R. (
2015
).
Contingent work rising: Implications for the timing of marriage in Japan
.
Journal of Marriage and Family
,
77
,
1039
1056
. 10.1111/jomf.12224.
Polikoff, N. D. (
2012
).
“Two parts of the landscape of family in America”: Maintaining both spousal and domestic partner employee benefits for both same-sex and different-sex couples
.
Fordham Law Review
,
81
,
735
760
.
Presser, H. B. (
1999
).
Toward a 24-hour economy
.
Science
,
284
,
1778
1779
. 10.1126/science.284.5421.1778.
Presser, H. B. (
2005
).
Working in a 24/7 economy: Challenges for American families
.
New York, NY
:
Russell Sage
.
Raley, R. K. (
1996
).
A shortage of marriageable men? A note on the role of cohabitation in black-white differences in marriage rates
.
American Sociological Review
,
61
,
973
983
. 10.2307/2096303.
Sassler, S. (
2004
).
The process of entering into cohabiting unions
.
Journal of Marriage and Family
,
66
,
491
505
. 10.1111/j.1741-3737.2004.00033.x.
Sassler, S., & Goldscheider, F. (
2004
).
Revisiting Jane Austen’s theory of marriage timing: Changes in union formation among American men in the late 20th century
.
Journal of Family Issues
,
25
,
139
166
. 10.1177/0192513X03257708.
Sassler, S., & Miller, A. J. (
2011
).
Class differences in cohabitation processes
.
Family Relations
,
60
,
163
177
. 10.1111/j.1741-3729.2010.00640.x.
Sassler, S., & Miller, A. J. (
2017
).
Cohabitation nation: Gender, class, and the remaking of relationships
.
Berkeley
:
University of California Press
.
Schieman, S., & Plickert, G. (
2008
).
How knowledge is power: Education and the sense of control
.
Social Forces
,
87
,
153
183
. 10.1353/sof.0.0065.
Schmitt, J., & Warner, K. (
2009
).
The changing face of labor, 1983–2008
.
Washington, DC
:
Center for Economic and Policy Research
.
Schneider, D. (
2011
).
Wealth and the marital divide
.
American Journal of Sociology
,
117
,
627
667
. 10.1086/661594.
Schneider, D., & Harknett, K. (Forthcoming).
Consequences of routine schedule instability for worker health and wellbeing
.
American Sociological Review
.
Schneider, D., Harknett, K., & Stimpson, M. (
2018
).
What explains the decline in first marriage in the United States? Evidence from the Panel Survey of Income Dynamics, 1969–2013
.
Journal of Marriage and Family
,
80
,
791
811
. 10.1111/jomf.12481.
Schneider, D., & Reich, A. (
2014
).
Marrying ain’t hard when you got a union card? Labor union membership and first marriage
.
Social Problems
,
61
,
625
643
. 10.1525/sp.2014.12316.
Shafer, K., & James, S. (
2013
).
Gender and socioeconomic status differences in first and second marriage formation
.
Journal of Marriage and Family
,
75
,
544
564
. 10.1111/jomf.12024.
Sohn, H. (
2015
).
Health insurance and risk of divorce: Does having your own insurance matter?
Journal of Marriage and Family
,
77
,
982
995
. 10.1111/jomf.12195.
Steverman, B. (
2014
,
November
10
).
What the economy has done to the family
.
Bloomberg Business
. Retrieved from https://www.bloomberg.com/news/articles/2014-11-10/how-the-bad-economy-breaks-up-families
Sweeney, M. (
2002
).
Two decades of family change: The shifting economic foundations of marriage
.
American Sociological Review
,
67
,
132
147
. 10.2307/3088937.
Thornton, A., Axinn, W. G., & Teachman, J. D. (
1995
).
The influence of school enrollment and accumulation on cohabitation and marriage in early adulthood
.
American Sociological Review
,
60
,
762
774
. 10.2307/2096321.
Thornton, A., Axinn, W. G., & Xie, Y. (
2007
).
Marriage and cohabitation
.
Chicago, IL
:
University of Chicago Press
.
Wang, W., & Parker, K. (
2014
).
Record share of Americans have never married: As values, economics and gender patterns change
.
Washington, DC
:
Pew Research Center’s Social & Demographic Trends Project
.
Wilson, W. J. (
1987
).
The truly disadvantaged: The inner-city, the underclass, and public policy
.
Chicago, IL
:
University of Chicago Press
.
Xie, Y., Raymo, J., Goyette, K., & Thornton, A. (
2003
).
Economic potential and entry into marriage and cohabitation
.
Demography
,
20
,
351
367
. 10.1353/dem.2003.0019.
Zangelidis, A. (
2014
).
Labour market insecurity and second job-holding in Europe
. Unpublished manuscript,
Department of Economics, University of Aberdeen
,
Aberdeen, Scotland
. https://doi.org/10.2139/ssrn.2615268

Supplementary data