Cohabitation has become increasingly accepted and normalized as part of the family system in Canada and has become the most common way to form a first union. The changing role of cohabitation in the family system is often understood as being driven by the ideational changes associated with the second demographic transition, but increasing international evidence indicates that this explanation is incomplete. Using nationally representative retrospective data from Canadians born between 1940 and 1979 from the 2011 General Social Survey, this study examines transitions out of first premarital cohabitation and fertility within these unions as two measures of the changing role of cohabitation. Across birth cohorts, Canadians are increasingly likely to use cohabitation as an alternative to marriage and less likely to use cohabitation as a short-lived prelude to marriage. These overall trends support the second demographic transition perspective. However, this study also finds that Canadians without a bachelor’s degree are far more likely to experience a birth within cohabitation and that their likelihood of transitioning to marriage has declined steeply across birth cohorts. This educational gradient in childbearing in cohabitation and the increasing educational differences in union transitions over time provide support for the diverging destinies thesis in Canada.
The institution of the family has undergone significant changes in Canada and other Western countries over the last century, including increased divorce rates, fertility declines, delayed marriage, increased diversity in family forms, increasing cohabitation, and extramarital fertility (Cherlin 2010). Marriage has been delayed and increasingly forgone, and nonmarital cohabitation has become more common as an accepted and normalized part of the transition to partnership (Bumpass 1990; Settersten and Ray 2010). Nonmarital cohabitation has become the most common way to form a first union in Canada, increasing from 2 % of first unions formed by Canadians born in the 1930s to more than 50 % for Canadians born in the 1970s (Le Bourdais and Lapierre-Adamcyk 2004). Research also suggests that cohabiting union formation has offset increases in median age at first marriage across Canadian birth cohorts (Wright 2016).
The rise of cohabitation is concomitant with ideational and structural changes that have taken place in Western countries over the past several decades, including increased individualism, secularization, and gender equality associated with the second demographic transition (Cherlin 2004; Lesthaeghe 2010). The second demographic transition (SDT) framework has become de rigueur as a way to understand broad changes in family behaviors, including the rise of cohabitation as an alternative family form. The SDT framework argues that changes in family behavior are primarily due to ideational shifts toward the fulfillment of higher-order needs of self-actualization and individual autonomy (Lesthaeghe 1995, 2010; Lesthaeghe and van de Kaa 1986).
However, a growing body of evidence suggests that family behavior changes are not as uniform as the SDT explanation suggests, but rather that trends in family behavior of the economically advantaged and disadvantaged are diverging in the United States (Cherlin 2010; Guzzo 2014; Kuo and Raley 2016; McLanahan 2004). For instance, Kuo and Raley (2016) found that Americans without a bachelor’s degree are increasingly less likely than those with a bachelor’s degree to legally marry their cohabiting partners over time. These studies suggest that ideational changes are not experienced uniformly across the population, leading to “diverging destinies” (McLanahan 2004) or a “pattern of disadvantage” (Perelli-Harris et al. 2010) in the ways that cohabitation is used in the partnership process across economic and social lines. However, educational differences in family behaviors among Canadians are less well understood (Kerr et al. 2006). Some studies found that the less-educated are more likely to cohabit rather than marry (e.g., Kerr et al. 2006), but others found no educational differences (e.g., Smock and Gupta 2002).
This study seeks to contribute to our understanding of the changing role of cohabitation in Canada, examining whether the SDT or the diverging destinies thesis more accurately explains trends over time. I use retrospective data from Canadians born between 1940 and 1979 from the 2011 General Social Survey to examine trends over time in cohabitation outcomes and fertility within cohabiting unions and to assess whether these trends are diverging across educational groups. How cohabiting unions end, either by transitioning to legal marriage or dissolution, and whether children are born into cohabiting unions are two important aspects of the changing meaning of cohabitation in Canada. Increasing educational differences in transitions out of cohabiting unions and fertility behaviors within cohabiting unions would provide evidence for the diverging destinies thesis in Canada.
Cohabitation has played a very different role in the partnership process at different times and in different places. Several scholars (e.g., Heuveline and Timberlake 2004; Kiernan 2001; Rindfuss and VandenHeuvel 1990; Smock 2000) have developed typologies of the role and meaning of cohabitation in the partnership process, generally using union duration, type of transitions out of cohabitation, and fertility behavior within cohabiting unions to classify the different roles cohabitation can play in the family system. The most differentiated typology was advanced by Heuveline and Timberlake (2004), who argued that cohabiting unions can take on six ideal types. Cohabitation is considered marginal in contexts with little social or institutional support, resulting in low incidence of cohabitation. Cohabitation emerges as a prelude to marriage or a stage in the marriage process when cohabiting couples use the union to assess the appropriateness of the match, or in a way akin to engagement. These unions transition to legal marriage relatively quickly and usually do not involve the birth of a child (prelude to marriage)—or if a child is born, legal marriage follows quickly (stage in the marriage process). Cohabiting unions that are an alternative to being single are short-lived and more likely to end in separation. When cohabitation becomes more prevalent, longer-lasting, less likely to transition to marriage, and more likely to involve children, it can be considered an alternative to marriage; with little selection into cohabitation based on individual characteristics, cohabitation can be said to have advanced to the final stage of being indistinguishable from marriage.
A variety of approaches have been used to examine the changing meaning of cohabitation, reflecting the diversity of factors that can be considered when exploring how cohabitation fits into the family system. Some studies examined the type of first union that people chose to enter, either cohabitation or direct marriage (e.g., Manning et al. 2014; Wright 2016). Others compared the characteristics of married couples and cohabiting couples (e.g., Kerr et al. 2006; Rindfuss and VandenHeuval 1990) or examined the level of institutionalization of cohabitation in various contexts (e.g., Beaujot et al. 2013; Nock 1995), the duration of cohabiting unions (Heuveline and Timberlake 2004), the prevalence of childbearing within cohabiting unions (e.g., Raley 2001), or the likelihood of different transitions out of cohabiting unions (e.g., Kennedy and Bumpass 2008).
American studies have shown that over time, fewer cohabiting unions are transitioning to marriage. Bumpass and Lu (2000) found that between 1987 and 1995, a larger proportion of cohabiting couples were dissolving their unions, and fewer were entering into marriage compared with cohabiting unions formed before this period. More recent research in the United States has found that this trend toward decreased risk of transitioning to marriage has continued. In 1995, 58 % of couples in first cohabiting unions transitioned into marriage by their third anniversary (Bramlett and Mosher 2002); in 2002, this decreased to 51 % (Goodwin et al. 2010). The most recent estimates are provided by Kuo and Raley (2016), who drew on the 2006–2010 National Survey of Family Growth to show that only 24 % of first premarital cohabiting unions formed between 2005 and 2010 in the United States transitioned into marriage by the end of the third year.
Fertility behaviors within cohabiting unions have also changed over time in the United States as cohabitation has become an increasingly acceptable context for childbearing (Cherlin 2010). In 1970–1974, 5 % of births were to cohabiting couples, which increased to 12 % of births in 1990–1994; increases in births to cohabiting couples accounted for 40 % of the increase in nonmarital births during this period (Raley 2001). By 1997–2001, 18 % of children were born to cohabiting parents, about the same proportion of children as were born to single mothers (Kennedy and Bumpass 2008). The increasing instability of cohabiting unions, combined with the changing fertility behaviors of cohabiting couples compared with married couples and single women, have led to arguments that in the United States, cohabitation is increasingly less likely to be a prelude to marriage (Seltzer 2000) but is also not becoming an alternative to marriage (Raley 2001).
In Canada, however, cohabiting unions are more stable than those formed in the United States (Le Bourdais and Marcil-Gratton 1996; Wu and Pollard 2000), and evidence of whether theses unions are becoming more or less stable over time is mixed. For instance, Wu and Balakrishnan (1995) found that first premarital cohabiting unions formed in Canada in the late 1980s were more likely to end in separation than those formed in earlier periods, although trends in the likelihood of marriage over time are less clear. On the other hand, Mills (2004) found that cohabiting unions have become more stable: 50.9 % of cohabiting unions formed by Canadians born in 1946–1950 transitioned into legal marriage, compared with 40.7 % of unions formed by the 1961–1965 birth cohort, and the older cohort had a higher probability of dissolving their cohabiting union than the younger cohort (Mills 2004). Fertility within cohabiting unions has also been increasing in Canada, especially in the province of Québec. In 1971, 2 % of Canadian children were born to cohabiting parents; by 1997, nearly one-half of children born in Québec and 15 % of children born in other provinces of Canada had cohabiting parents (Kerr et al. 2006; Laplante and Fostik 2015). Cohabitation trends in Canada have led Heuveline and Timberlake (2004) and Mills (2004) to argue that cohabitation has reached the stage of being an alternative to marriage in Canada, especially in the province of Québec, compared with being more of an alternative to being single in the United States or a prelude to marriage in Switzerland.1
Second Demographic Transition Perspective
One of the most popular explanations for the dramatic changes in family behavior, including the role of cohabitation, is the second demographic transition (SDT) framework (Lesthaeghe 1995; van de Kaa 1987). The SDT entails the changes in family behaviors that began in Western countries in the 1960s, including the rise of cohabitation, increasing extramarital fertility, and the deinstitutionalization of marriage (Lesthaeghe 2010). The SDT perspective argues these changes in family behaviors are primarily driven by cultural and ideational changes—fueled in part by sexual, contraceptive, and gender revolutions—that shifted focus to the realization of high-order needs, including self-actualization and individual autonomy (Lesthaeghe 2010; Lesthaeghe and Neidert 2006).
The original proponents of SDT drew from Inglehart’s theory of postmaterialism, which suggests that the highly educated will more strongly identify with values associated with the SDT, such as individualism and gender equality. Following the SDT, the highly educated are expected to be the forerunners in adopting different family behaviors, which will subsequently be diffused to the rest of the population (Lesthaeghe and Surkyn 2002; Perelli-Harris et al. 2010).
Diverging Destinies Perspective
A growing body of evidence suggests that the ideological changes at the center of the SDT perspective are conditioned by economic opportunity structures (e.g., Kuo and Raley 2016; McLanahan 2004). The likelihood of marriage in both Canada and the United States has been declining over time, but the decline in the United States has been much more dramatic among the less-educated, particularly those without a bachelor’s degree (Goldstein and Kenny 2001; McLanahan 2004). In a recent study, Kuo and Raley (2016) found that educational differences in the likelihood of transitioning to marriage from cohabitation increased between the early 1990s and late 2000s in the United States. The growing educational disparities in cohabitation outcomes are largely driven by the decline in marriage among those without a bachelor’s degree (Kuo and Raley 2016).
The less-educated in the United States and several European countries are also increasingly likely to have extramarital children (Cherlin 2010; Kennedy and Bumpass 2008; Perelli-Harris et al. 2010). In the United States, the proportion of births to cohabiting women increased from 6 % in 1980–1984 to 26 % in 2010–2014, but this increase was much more dramatic for those with less than a bachelor’s degree (Wu 2017). In 2010–2014, 42 % of births to mothers with less than a high school diploma occurred within cohabitation, compared with only 9 % of births to mothers with a bachelor’s degree (Wu 2017). Among births to women with a high school diploma or some college education, 34 % and 29 % (respectively) were within cohabiting unions (Wu 2017), suggesting that the fertility behavior of the college-educated within cohabitation is different from that of other educational groups.
Completing postsecondary education is increasingly required to achieve financial independence in both Canada and the United States (Berlin et al. 2010; Boothby and Drewes 2006), and many American studies have highlighted the increasing importance of financial security in facilitating the transition to marriage (Sassler and Goldscheider 2004; Sweeney 2002). An educational gradient in the type of unions that Canadians form exists; the less-educated are more likely to cohabit. However, educational differences are not as large as in the United States (Kerr et al. 2006), and past studies have found no differences in marriage transitions by education among older cohorts of Canadian cohabitors (e.g., Wu and Balakrishnan 1995). Whether more recent cohorts of Canadians are using cohabitation differently across educational lines remains an open question. If education stratifies cohabitation experiences in Canada, this would suggest that the diverging destinies paradigm may be a better explanation of the changing role of cohabitation witnessed in Canada and other Western nations.
The goal of this study is to examine trends over time in how first premarital cohabiting unions in Canada end, either through marriage or separation, and trends in fertility within these unions. This study also examines educational differences in these trends and explores whether educational differences in these family behaviors have increased over time. Answers to these questions will contribute to our understanding of the changing role and diverging trends of cohabitation in Canada and whether evidence exists for the diverging destinies thesis in Canada.
Data for the analyses are drawn from the 2011 Canadian General Social Survey (GSS), a cross-sectional survey conducted by Statistics Canada every year since 1985 with a specific thematic focus each year. The data for this study come from Cycle 25, the fifth and most recent GSS to focus on families. The 2011 GSS was conducted using computer-assisted telephone interviews between February and November 2011 and is representative of noninstitutionalized people aged 15 and older living in the 10 Canadian provinces. The survey includes detailed retrospective union and fertility histories collected through questions that ask about first and current marriages, whether these marriages were preceded by a common-law union, whether the respondent “ever lived with a common-law partner even if it was for less than 1 year,” first and current common-law unions, the dates and ages at which all of these unions started and ended, and the dates of birth of the respondent’s children.
The GSS includes a total sample of 22,435 respondents, but analyses are restricted to people born in Canada between 1940 and 1979 (n = 13,171) and are further restricted to respondents whose first union was a nonmartial cohabiting union, resulting in a subsample of 4,874. Only first premarital unions are examined because the risks for marriage and separation likely differ depending on whether individuals are in their first or subsequent cohabitation and whether they are in a cohabiting union following the dissolution of a marriage (Lichter and Qian 2008). Respondents who were born outside of Canada are excluded because all or part of their union histories may have occurred outside of Canada, which complicates the examination of changes in union formation in Canada. Fifty-eight respondents whose first cohabiting union ended through the death of their partner are excluded because age at partner’s death is not available, and 36 respondents who reported starting and ending their union in the same person-period are also excluded. Cases with missing data on age at the start of first cohabiting union, age at union dissolution, the type of union transition, or on any of the included covariates are excluded (n = 354; 7.4 % missing), resulting in a final analytic sample of 4,426.
The outcome of the respondents’ first nonmarital cohabiting union is coded into three categories: (1) union dissolved through separation, (2) union transitioned into legal marriage, or (3) union is intact at the time of the survey. A measure for childbearing within the cohabiting union is constructed to indicate whether the respondent gave birth to, or fathered, a biological child after the start of the union but before the union either dissolved or transitioned into legal marriage. Birth cohort is used to examine changes over time because the SDT links family changes to cohort succession (Lesthaeghe 2014) and is measured in decades: the 1940s (n = 201), 1950s (n = 963), 1960s (n = 1,414), and 1970s (n = 1,848).2 The education measure includes three categories: a high school diploma or less, some postsecondary education, and a bachelor’s degree or higher. Respondents who completed a two- or three-year college or technical program leading to a sub-baccalaureate diploma or degree, or those who attended but did not complete a four-year bachelor’s program, are included in the “some postsecondary education” category.
The analyses include controls for several measures associated with union outcome and childbearing including sex (female = 1) (Wu and Balakrishnan 1995) and age at union start (Guzzo 2009, 2014; Manning 2004). Age at union start includes four categories: unions formed during the respondent’s teen years (<20), early 20s (20–23), mid-20s (24–26), and late-20s or later (>26). Because cohabiting couples in Québec are less likely to transition to marriage and more likely to have children (Laplante and Fostik 2015; Wu and Balakrishnan 1995), a measure for whether the respondent was born in Québec or another Canadian province is included. As a sensitivity test, all analyses were also estimated separately for Québec and other Canadian provinces, given the established differences in family behaviors (Le Bourdais and Lapierre-Adamcyk 2004; Le Bourdais and Marcil-Gratton 1996; Pollard and Wu 1998; Turcotte and Belanger 1997). The results from these models can be found in the online appendix, and any regional differences are highlighted in the discussion of national-level results.
Discrete-time multinomial logistic regression models are used because event-history models take into account right-censoring (the fact that some current cohabiting relationships may transition into marriage or dissolve in the future, or that a child may be born within a cohabiting union after the date of the survey) (Allison 1984). A person-period data file was created with 201,365 person-period observations from 4,426 cohabiting unions. Cohabitors enter the risk set of union transition and extramarital fertility at the time of union formation. In the analysis of union outcomes, they exit at the time of (1) legal marriage, (2) union dissolution, or (3) survey date, whichever occurs first. In the analysis of extramarital fertility, respondents exit the risk set either (1) when they birth or father a child; or (2) when the union dissolves or transitions into marriage, or at the survey date if they are currently cohabiting.
Discrete-time models require the shape of the hazard (the duration dependence) be specified (Box-Steffensmeier and Jones 2004; Jenkins 2005). Rather than assume a theoretical shape of the hazard function, I use a piecewise constant to model the duration dependence. Dummy variables are used to group the units of union duration until separation, marriage, or the survey date into single years of duration (i.e., 0.1–0.9 years; 1–1.9 years up to 9.0–9.9 years), and union durations 10 years or longer are combined into one category. Within each category, the hazard rate is assumed to be constant but is allowed to vary across these duration categories as well as across categories of the union outcome. This approach has the advantage of allowing the shape of the hazard function to be determined empirically without burdening the model with dummy variables for every unit of time. These models assume that the hazards are proportional within each birth cohort such that the effect of each risk factor is constant over the duration of the cohabiting union within a given cohort, but that the effect of each risk factor can vary across cohorts.
Weighted descriptive characteristics of the analytic sample of Canadians’ first premarital cohabiting unions that are at risk of ending through separation or marriage and those at risk of having a child within a first cohabiting union are displayed by birth cohort. Next, multinomial logistic regression event-history models are estimated to examine the risks of separation and the risks of marriage, and logistic regression event-history models are estimated to examine the risks of an extramarital birth. In both sets of models, an interaction between education and birth cohort is estimated to examine whether the relationships between education and (1) union outcome and (2) fertility within cohabitation are moderated by birth cohort. All analyses are weighted to be representative of the population and to account for the clustering of observations within respondents in the person-period data file.
Table 1 displays the characteristics of the analytic sample by birth cohort and for the full sample of 4,426 respondents. Life table estimates (available by request) show that across birth cohorts, (1) a decreasing proportion of cohabiting unions transition to marriage by the end of the third year, (2) an increasing proportion of cohabiting unions remain intact into the fourth year, and (3) the proportion that separate has stayed relatively stable, suggesting that cohabiting unions in Canada are not becoming more unstable over time but are less likely to be used as a short-lived prelude to marriage for more-recent cohorts. Among first cohabitations formed by Canadians born in the 1940s, 51.3 % transitioned to marriage within the first three years, 35.9 % separated, and 12.8 % continued as cohabiting unions. For cohabiting unions formed by the 1970s birth cohort, only 41.5 % transitioned to marriage, 33.5 % separated, and 25.1 % remained in the cohabiting union at the end of three years. The proportion of first cohabiting unions that involve the birth of a child also increased across birth cohorts, ranging from a low of 13.2 % of cohabiting unions formed among the 1940s birth cohort to 25.5 % of unions formed among the 1970s birth cohort.
Table 2 presents results from multinomial logistic regression event-history models examining differences in the risks of separation and marriage relative to remaining in a first cohabiting union across birth cohorts. Significant differences between the risks of separation relative to marriage are shown in bold. Model 1 in Table 2 shows that the risks of dissolving a first cohabiting union relative to continuing to cohabit did not change across birth cohorts. The risks of marriage among cohabitors in their first union, however, generally decreased across birth cohorts. First cohabiting unions formed by respondents born in the 1950s were 48 % more likely to transition to marriage than unions formed by respondents born in the 1970s, and those formed by respondents born in the 1960s were 22 % more likely to transition to marriage than unions formed by the 1970s birth cohort. Model 1 in Table 2 also shows no educational differences in the risks of separation but significant educational differences in the risk of marriage. Overall, Canadians with a bachelor’s degree or higher were more likely to transition to marriage than their less-educated counterparts.
Model 2 of Table 2 includes interaction terms between birth cohort and education, which indicate that educational differences in transitioning to marriage have increased significantly across birth cohorts. To illustrate the changing educational differences, the relative log odds of marriage for each educational category in each birth cohort are displayed in Fig. 1. The carets (^) in Fig. 1 show no significant educational differences in the risks of marriage for cohabitations formed by the 1940s or 1950s birth cohorts, but educational differences became significant for cohabitations formed by those born in the 1960s (p < .05) and the 1970s (p < .001). Educational differences in more-recent birth cohorts are due to significant declines in transitions to marriage among Canadians with less than a bachelor’s degree (p < .001), while the relative risks of marriage among the university-educated did not change across birth cohorts. These results mirror those of Kuo and Raley’s (2016) analyses, suggesting that the same diverging destinies pattern in union transitions they found in the United States is evident in Canada.
Results from the same models examining Québec and other Canadian provinces separately show similar patterns (online appendix, Tables A1 and A2). Among people in Québec, educational differences in the likelihood of transitioning to marriage are statistically significant for only the most recent birth cohort, born in the 1970s. This educational difference is due to the significant decline in the likelihood that people in Québec with a high school diploma or less or those with only some postsecondary education transition to marriage across cohorts. In other parts of Canada, the educational differences are driven by the declining likelihood that people with some postsecondary education will transition to marriage; the likelihood of marriage for those with a high school diploma or less and for those with a bachelor’s degree or more was stable over time.
Fertility Within Cohabiting Unions
Table 3 presents results from discrete-time logistic regression models examining differences in the likelihood that a child is born into a first premarital cohabiting union across birth cohorts. Model 1 of Table 3 shows that childbirth within cohabiting unions increased across successive birth cohorts. Large educational differences in fertility are also evident: Canadians with a bachelor’s degree or higher are by far the least likely to bear or father a child within their first cohabiting union. The odds of having a child within a first cohabiting union are 2.1 times higher for someone with some postsecondary education and 2.76 times higher for those with a high school diploma or less compared with bachelor’s degree holders. Model 2 in Table 3 includes interaction terms between education and birth cohort, all of which are insignificant, indicating that educational differences in fertility within cohabitation are consistent across birth cohorts. The enduring educational differences in fertility behaviors in cohabitation across birth cohorts does not support the diverging destinies thesis, but the educational gradient is also inconsistent with the SDT perspective, which would suggest that the more highly educated would be the change-leaders as the cultural and ideational vanguard (Lesthaeghe and Surkyn 2002).
Educational differences in fertility behaviors within cohabiting unions are evident in models including only Québec (online appendix, Table A3) and including only Canadians from other provinces (online appendix, Table A4). As was the case for Canada overall, there are no significant birth cohort and education interaction terms in either of the region-specific models, indicating consistent educational differences in childbearing in cohabitation over time in Québec and other Canadian provinces. However, educational differences in the likelihood of bearing or fathering a child within cohabitation are much smaller in Québec than in other parts of Canada. Although the difference is smaller than in other parts of Canada, the less-educated in Québec are significantly more likely to use cohabitation as a context for childbearing.
Using nationally representative retrospective union and fertility histories from Canadians born between 1940 and 1979, this study examines two measures of the changing role of cohabitation in the family system—transitions out of first premarital cohabitation, and fertility within these unions—and whether educational differences in these two family behaviors have become larger across birth cohorts. The results provide some support for the SDT in that recent cohorts of Canadians are increasingly less likely to transition to marriage than older cohorts and increasingly more likely to experience the birth of a child within cohabitation. This suggests that recent cohorts of Canadians are less likely to use cohabitation as an alternative to being single or as a short prelude to marriage, and are more likely to use cohabitation as an alternative to marriage, especially in Québec.
However, the SDT suggests that the more highly educated would lead these trends because of their stronger identification with the value changes driving the SDT, including individualism and gender equality (Lesthaeghe and Surkyn 2002; Perelli-Harris et al. 2010). Yet, this study finds that in Canada, like in the United States (Kuo and Raley 2016), the less-educated are the drivers of the trend toward fewer cohabitations progressing to marriage. Educational differences in the likelihood of marriage are increasing across birth cohorts because of the declines in marriage among Canadians without a university degree among the 1970s birth cohort. Similarly, the less-educated in Canada are significantly more likely to have a child within cohabitation, which is consistent with the educational gradient found in Norway, Russia, and the United Kingdom (Perelli-Harris et al. 2010).
Less-educated Canadians are increasingly less likely to use cohabitation as a step in the marriage process and more likely to use it as an alternative to marriage, suggesting that marriage is reserved for only the most-educated. The educational gradient in these family behaviors may be due to a higher commitment to the values associated with the SDT among the less-educated rather than the more-educated, although this would be inconsistent with the SDT argument. Unfortunately, appropriate attitudinal measures are not available in the 2011 GSS. However, this explanation is unlikely given consistent evidence that the more highly educated identify more strongly with the values of individualism and gender equality (Weakliem 2002), two of the key drivers of ideational change according to the SDT (Lesthaeghe 2010).
The findings of this study align more closely with the diverging destinies perspective, suggesting that differential economic opportunities across educational groups have an important impact on cohabitation outcomes and fertility in cohabiting unions in Canada. In Canada, like the United States, the economy has transitioned away from manufacturing and primary and secondary industries toward a knowledge-based, service driven economy requiring a more highly skilled labor force, which has made it harder for young people without a university degree to achieve economic stability (Berlin et al. 2010; Danziger and Ratner 2010; Fussell et al. 2007). Simultaneously, evidence suggests that the economic requirements for marriage are rising (Smock et al. 2005) and that marriage is becoming a capstone to be achieved after economic stability is secured (Cherlin 2004). Thus, cohabitation is more likely to be a long-term alternative to marriage among the disadvantaged (Lichter et al. 2006), as evidenced by lower rates of marriage and higher rates of fertility within cohabiting unions. If Canadians with the fewest resources are forgoing marriage because of financial barriers, children born in these unions will also likely experience disadvantages in their social and economic outcomes, which will contribute to more divergence in the family patterns of the less-advantaged over time.
This study examines transitions out of cohabitation and fertility within cohabiting unions, yet other factors should be considered when determining the role of cohabitation, particularly whether partners pool resources and share assets (Hamplova et al. 2014), and the marital intentions and engagement status of the partners (Guzzo 2014; Manning et al. 2007). Measures of marital intentions and engagement status are not available in the existing data; future research is needed to further elucidate whether the patterns found here are also reflected in the decision-making processes within cohabiting unions. Future studies should also examine the duration of cohabiting unions after the birth of a child as well as the timing of cohabitation relative to the conception of a child, which have also been used to understand the role of cohabitation (Manning 1993; Lichter et al. 2014; Musick 2007). Future research should also control for the factors that affect selection into cohabitation rather than marriage as a first partnership, which have changed over time. For instance, Turcotte and Goldscheider (1998) found that among Canadian women born in the 1960s, 1970s, and 1980s, higher education was associated with directly marrying rather than cohabitation but that the opposite was true for previous birth cohorts. Controlling for this selection may result in even larger group differences in the role of cohabitation in the marriage process if more highly educated or older people are more likely to enter marriage directly.
The retrospective union histories used in this study allow for the analysis of first premarital cohabiting unions formed across four birth cohorts. However, cohabiting unions may be underreported, especially among older respondents who might be less likely to recall and report unions that were formed decades earlier (Hayford and Morgan 2008). This concern is mitigated by the detailed questions used in the GSS to illicit full union histories, including specific questions about whether first or current marriages were proceeded by cohabitation and whether the respondent has ever cohabited, even if for less than a year. As an additional test, I compared the proportion of first unions formed through cohabitation for each birth cohort using data collected in 2001 and 1984, when each birth cohort was younger, to test whether reporting of first cohabiting unions declined as the cohorts aged. There were no significant differences in the proportion reporting cohabitation as a first union across survey years. This finding is supported by the published work of Le Bourdais and Lapierre-Adamcyk (2004) and Burch and Madan (1986), who found similar proportions of each birth cohort of Canadians forming first unions through cohabitation as the present study using data from 2001 and 1984, respectively.
Despite its limitations, this study suggests that the role of cohabitation in the partnership process in Canada is further along Heuveline and Timberlake’s (2004) typology than cohabitation in the United States. Cohabiting unions in Canada are increasingly likely to be used as an alternative to marriage across cohorts: they are longer-lasting, less likely to transition to marriage, and increasingly involve childbearing. Moreover, the results suggest that the SDT explanation for changes in the role of cohabitation in Canada is incomplete and contributes to a growing body of similar evidence from the United States (e.g., Kuo and Raley 2016) and the UK and Europe (e.g., Perelli-Harris et al. 2010). The transition to marriage from a first premarital cohabiting union is becoming increasingly more difficult for Canadians without a bachelor’s degree, and cohabitation is much more likely to be used as a context for childbearing among those with less resources. The focus of the diverging destinies thesis on how economic opportunities structure and condition contemporary changes in family behaviors is a useful frame for understanding emerging educational gradients in the role of cohabitation in the lives of Canadians.
I thank Rachel Margolis, three anonymous reviewers, and the editors for their insightful comments and suggestions at various stages of this manuscript. A previous version of this manuscript was presented at the 2016 annual meeting of the Population Association of America.
At the national level, scholars have characterized cohabitation as an alternative to marriage. However, Le Bourdais and Lapierre-Adamcyk (2004) found that regional variation is so large that in the province of Québec, cohabitation is better characterized as indistinguishable from marriage, while in the rest of Canada, it remains a prelude to marriage. These differences are discussed in more depth throughout.
Sensitivity analyses were conducting using cohabitation cohort rather than birth cohort as the measure of change over time, and the results are consistent across these different specifications.