By: France-Pascale Ménard
McGill Sociological Review, Volume 2, April 2011, pp. 59-76
Over the last decades, families in Canada have experienced tremendous transformations. We have witnessed a reduction in marriage rates, along with a significant rise in the divorce rate and in the rate of common-law unions, leading to rising conjugal and family instability. In this study we investigate factors, such as parental separation, school enrollment and region of residence, that influence the risk of first union dissolution in Canada for common-law unions and for marriages that are or are not preceded by cohabitation. We examine to what extent these factors affect differently the stability of the different union types. We use proportional hazards models based on retrospective data from Statistics Canada’s General Social Survey conducted in 2006. Interestingly, we find that the presence of a child, even of a stepchild, increases the stability of common-law unions, as it brings greater commitment to the relationship. As for marriages, only the more traditional family setting, in which children are conceived after the formation of the union, decreases the risk of dissolution.
Since the 1960s, Canada has seen major socio-demographic shifts that have had dramatic impacts on the institution of marriage. Among other changes, there has been a substantial decline in marriage rates as young adults are delaying marriage and as adults of all ages are slowly adopting common-law unions in the place of marriage. Therefore, cohabiting unions are becoming the new family setting for a rising proportion of children, particularly in Québec, where family transformations have been more pronounced. Among more recent cohorts, cohabitation in Québec has emerged as a substitute for marriage, while in the rest of Canada it is considered as a legitimate premarital step leading to marriage. In 2006, a third of couples in Québec were in common-law unions as opposed to 18% of all Canadian couples (Milan, Vézina, and Wells 2007). Consequently, more couples have children outside of marriage; in 2000, close to 60% of new registered births in Québec were to unmarried mothers, while a third of such births were recorded for Canada as a whole (Le Bourdais and Lapierre-Adamcyk 2004). In addition to being less popular, marriage has become less stable. In 2003, 35 years after the adoption of the Divorce Law, over a third of all marriages in Canada ended before reaching their 30th wedding anniversary (Statistics Canada 2005 cited in Clark and Crompton 2006). Further, cohabiting unions, even when children are involved, tend to be less stable than marriages (Le Bourdais, Neill, and Marcil-Gratton 2000). Premarital cohabitation is no insurance against separation since the stability of marriages preceded by cohabitation appears to be lower than for marriages not preceded by cohabitation (Le Bourdais and Marcil-Gratton 1996 cited in Beaujot 2000). As a consequence of conjugal instability, an increasing number of children experience parental separation (Bumpass and Lu 2001; Seltzer 2000), and are affected by post-separation economic vulnerability (Bumpass, Raley, and Sweet 1995). Therefore, conjugal instability raises questions regarding the effects of changing parenting contexts on children, the economic circumstances of their families, and family values and obligations (Bumpass and Lu 2001). It is thus crucial to understand the factors that influence the risk of union dissolution for cohabiting unions and marriages. In this study we will focus our attention on the dissolution of first unions, both to simplify the analyses and because of the higher risk of marital failure among remarriages than first marriages (Beaujot 2000; Bumpass et al. 1995; Clark and Crompton 2006). More particularly, we will compare the factors that influence the risk of separation by the type of union.
Determinants of Union Dissolution
The study of changes within the family has been of great interest among social theorists, and extensive research has been conducted to understand the progression of divorce and separation. Studies have observed higher rates of union dissolution among cohabiting unions and marriages preceded by cohabitation than among direct marriages, suggesting that the experience of cohabitation can have a negative effect on marital stability. This explanation suggests that the more unstable nature of cohabitation undermines conjugal commitment and increases approval of divorce (Axinn and Thornton 1992; Hall and Zhao 1995). Another possible explanation for the higher rate of conjugal disruption among cohabiting unions than marriages, preceded or not by cohabitation, suggests that people who cohabit have demographic, socioeconomic, cultural, or personality characteristics that make them more prone to separation (Axinn and Thornton 1992; Le Bourdais et al. 2000; Liefbroer and Dourleijn 2006). One such factor that tends to increase the risk of union dissolution is parental divorce, which also influences the choice of union (Bumpass et al. 1995; Hamplová 2003; Liefbroer and Dourleijn 2006). It has been shown that individuals from divorced homes are less likely to take on long-term commitments and marry, and hence are more inclined towards cohabitation than marriage (Hamplová 2003). They are also more likely to experience the dissolution of their first union.
In contrast, the presence of children in a first union tends to decrease the risk of union dissolution for both cohabiting unions and marriages. In fact, having children may completely change the way cohabiting couples define their relationship. Cherlin (2004) argues that the post-modern ideology of individualism and self-interest, associated with the rising incidence of cohabitation, is less prevalent when children are involved because the issue of commitment comes into consideration. However, having children before entering into a first union is associated with higher risk of union dissolution than is having children after the start of the union (Le Bourdais et al. 2000; Wu and Balakrishnan 1995). Similarly, the presence of stepchildren of the first union may put an additional strain on the relationship, increasing the risk of separation (Hall and Zhao 1995). In addition, the age at the start of a first union tends to significantly influence the risk of separation (Raley and Bumpass 2003). People who enter a first union at an early age face a greater risk of union dissolution than people who postpone entry into their first union. Another significant effect is the period effect, measured by the time period when a union forms.It is a reflection of changes in expectations of marriage, and family values, i.e. from a family-oriented to an individually-based set of values. The period effect has been shown to influence the risk of union dissolution as union instability has increased over time (Wu and Balakrishnan 1995). Nevertheless, as cohabitation becomes more common, its influence on union stability is assumed to become less pronounced with later cohorts of union formation (Le Bourdais et al. 2000).
Level of education is one of the main economic factors that tend to influence the risk of union dissolution. If we consider education to be a measure of earning potential, financial independence, and socioeconomic status, we would expect higher educational levels to be associated with a greater risk of union dissolution, possibly because of more liberal attitudes toward divorce and separation (Beaujot 2000). Similar hypotheses can be made regarding employment status. The employed, particularly women, are more likely to separate given their greater financial independence. For men, however, the effect of employment is associated with a decreased likelihood of separation (Brines and Joyner 1999; Smock and Manning 1997). Another important aspect of education is the influence of school enrollment on the risk of union dissolution. As part of the life course approach, it is expected that individuals finish school before starting a first union. The transition into adulthood tends to be closely linked with the end of full-time education, the start of a professional career, and the beginning of a first union when young adults are ready to make a long-term commitment (Hamplová 2003). Heterogamy in partners’ education and earnings is also likely to raise the risk of union disruption, especially among common-law couples, as found in Brines and Joyner’s (1999) study on conjugal stability.
It has been shown that cohabiting couples tend to hold more liberal and modern values and attitudes than married couples with no prior cohabiting experience (Axinn and Thornton 1992). Cohabiting couples tend to be less religious, less inclined to marry, and more accepting of divorce. In this sense, religious attendance is associated with union stability, especially when bound by marriage (Clark and Crompton 2006). Therefore, the effect of religious observance, even if infrequent, is likely to lower the risk of union dissolution. In addition, living in an urban setting is associated with higher dissolution rates than rural settings since urban areas tend to propagate liberal and individualistic values (Liefbroer and Dourleijn 2006). The region of residence, which tends to be strongly correlated with the choice of living arrangements among first unions (Laplante 2006; Mongeau, Neill, and Le Bourdais 2001), also appears to affect patterns of union dissolution, with conjugal instability being greater in Québec than in the rest of Canada. However, region of residence is expected to have a different effect on union stability for cohabiting unions and marriages; the effect of cohabitation is expected to be less pronounced in Québec where cohabiting unions have become the norm, while being more important in the rest of Canada where cohabitation is still more marginal (Le Bourdais et al. 2000).
Accordingly, in this study, we examine which of the aforementioned demographic, socioeconomic, and cultural characteristics significantly affect the risk of union dissolution for cohabiting unions and marriages. We assess the risk of dissolution separately for marriages preceded or not by cohabitation, and common-law unions, as opposed to analyzing the risk of dissolution for all types of union together, as is done in most studies. This allows us to analyze whether these factors affect marriages and cohabiting unions differently. We start with the assumption that common-law unions face a greater risk of separation. We then investigate to what extent the cohabiting experience leads to higher rates of union dissolution among current and former cohabiting couples who have subsequently married their partners, in contrast to couples who married directly.
Data and Methods
Our study on the first union dissolution in Canada is based on data from the 2006 Statistics Canada General Social Survey (GSS,cycle 20). This cross-sectional survey asked 23,608 respondents, 10,351 men and 13,257 women, aged 15 years and over living in private households in the ten Canadian provinces to provide information about the chronology of marriages and common-law unions. It asked every respondent about the date (month and year) of entry into and exit from their relationships along with a wide array of other demographic, socioeconomic and cultural characteristics. This study focuses on the risk of separation of the first union in Canada. By definition, all respondents who had never entered a marital or nonmarital cohabiting union at the time of the interviews, which were conducted in four waves from June to October 2006, are excluded from the sample, reducing the initial sample to 19,608 respondents. Moreover, all subsequent unions are discarded from the analysis. For determining which unions to keep, the rank of unions is established based on the chronology of marriages and common-law unions. Cohabiting couples are referred to as common-law partners, i.e. partners who live together in a conjugal relationship. Only unions that were formed after 1970, i.e. two years after the adoption of the Divorce Law, are analyzed. Also, only individuals aged 25 to 64 are kept in the analysis because they are the ones most likely to have committed themselves to a marital or common law union (Le Bourdais et al. 2000; Mongeau et al. 2001).
This study is based on the life-course approach, which takes into account the timing of events in a person’s life, including changes related to the family, school, and work that individuals go through during their lifetime. The order of events is particularly important as life events must be analyzed in relation to preceding but connected events. The life-course approach calls for event-history data, which retrospectively computes the events in the life cycle and their timing (Blossfeld and Rohwer 2002 cited in Hamplová 2003). Analyses are performed using the risk of dissolution of the first union as the dependent variable, and the dates (month and year), if available, of entry into and exit from the union. Union stability is estimated based on the timing of entry into a first union. Thus, we consider whether the union had started by marriage or by cohabitation, and in the latter case, whether it was followed by marriage or not. In order to do so, a time-varying covariate for union type is included to account for the different modes of entry into conjugal life (direct marriage or cohabitation), and the state of the relationship: cohabitation outside of marriage, marriage after prior cohabitation, and marriage without prior cohabitation (Liefbroer and Dourleijn 2006) [Note 1]. This time-varying covariate is constructed by comparing the date of the start of the union and the date of marriage. Therefore, if cohabiting couples marry, they move from common-law unions to marriage with prior cohabitation, on the date of their marriage. This variable allows us to test whether the experience of cohabitation is associated with a higher risk of union dissolution among marriages preceded by cohabitation than direct marriages. As for the end of a union resulting from causes other than death, the date of separation is used, i.e. when partners stopped living in the same household, rather than the date of the official divorce, except for a small minority of respondents who divorced without prior separation. In fact, the timing of divorce rarely corresponds to the end of cohabitation because of the length of the legal process (Raley and Bumpass 2003). For those whose marriage ended with an immediate divorce or annulment without a period of legal separation, the time of divorce or annulment is used (Clark and Crompton 2006). Finally, for those whose first unions never dissolved by the time of the survey, they are censored at the date of the interview, meaning that the period of observation ended while no event occurred. Similarly, respondents who experienced the death of their partner are censored at the date of their partner’s death, without having gone through the separation or the divorce of their first union.
In our study, we include several demographic, socioeconomic and cultural covariates, which have been shown to influence the risk of union disruption in previous studies, and different models are run accordingly. In terms of demographic variables, a time-constant variable is constructed, indicating whether respondents experienced the separation or divorce of their parents before the age of 18. Another time-constant age variable is included, indicating whether respondents entered into a first union before the age of 25. We have created a time-varying covariate to measure the effect of the presence of at least one biological or adopted child on the respondent’s first union. This variable accounts for the timing of the child’s conception (or adoption), whether it preceded the start of the union, or followed it. We account for the presence of at least one stepchild on the respondent’s first union with a time-constant covariate. To account for the timing of entry into a union, we use a time-varying covariate covering four time periods (1970-1979; 1980-1989; 1990-1999; 2000-2006). We use entry into union instead of birth cohorts to capture the duration of the union as well as the context in which unions are formed and evolve (Le Bourdais et al. 2000).
The highest completed level of education is used as a proxy for human and cultural capital as well as socioeconomic status (Hamplová 2003) in order to measure the influence of educational attainment on first union dissolution (Raley and Bumpass 2003). For the respondents who had completed their studies by the time they entered into their first union, we consider their highest educational level at the time of the survey (less than high school; high school degree; post-secondary education; university degree). For those who had not completed their studies by the time they entered into their first union, a time-varying covariate is constructed as a function of their highest level of studies ever completed, and their age at the start of their first union. Moreover, the influence of gender on first union dissolution combined with educational attainment is included in order to test their interaction effect. School enrollment is measured as a time-varying covariate, indicating when the respondents completed their studies. Finally, we include a time-varying covariate that measures the effect of whether respondents had started a first work episode by the time they entered their first union, whether they had worked without interruptions, and if they had had a first work interruption, whether they had started a second work episode. Once again, the influence of gender on first union dissolution combined with employment status is included in order to test the interaction of gender and work status. Unfortunately, given data limitations, it is not possible to test whether heterogamy in partners’ education or earnings is likely to raise the risk of union disruption because questions on partners’ education and earnings in the GSS cycle 20 are limited to current partners. Several time-constant cultural variables are also included, such as religious attendance (not at all, infrequently, at least once a month), region of current residence (Québec, rest of Canada), and size of current community (rural vs. urban Canada according to Statistics Canada’s definition of an urban area [Note 2]), to assess the differences in first union dissolution by union type.
Kaplan-Meier survival functions are performed to evaluate the progression of union instability in Canada and estimate the proportion of first unions that ended within 35 years of formation by region of residence, and cohort of union formation for the three different states (time-varying) of relationship: cohabitation outside of marriage, marriage with prior cohabitation, and marriage without prior cohabitation. Two separate sets of Cox proportional hazard models are carried out to determine which factors among demographic, socioeconomic and cultural covariates significantly affect the risk of union dissolution, by union type (time-varying): the first is marriage, with or without prior cohabitation, and the second is cohabitation. Each of them measures hazard rates, i.e. the risk that an individual will experience a first union dissolution, given the set of above explanatory variables. A relative risk above 1 implies a higher rate of union dissolution than the reference category, while a risk below 1 implies a lower rate. Both sets of models test the effect of each of the covariates, starting with the background variables, adding the economic variables, then the cultural variables, and finishing with the interactions between gender and economic variables. For both sets of models, the best-fitted model includes all three groups of variables, excluding the interaction terms since none of them were statistically significant at a 10% significance level. All of the abovementioned analyses are performed using a weighting factor provided by Statistics Canada for analyses at the person level. Each respondent is thus representative of several others in the Canadian population. Moreover, bootstrap weights, also provided by Statistics Canada, are used with the best-fitted models to make replicate-based variance estimations, whose estimations are reported subsequently. Finally, several tests are run to examine whether the proportionality assumption, which is key to the Cox proportional hazard model, is violated, indicating the presence of time-dependent covariates. Since we already include several time-varying covariates, accounting for the interaction of these variables with time, we do not find any additional non-proportional effect.
Table 1 provides a description of the independent variables included in the analyses of first union dissolution by union type (state of the relationship) at the time of the survey. Therefore, individuals who entered their first unions as common-law partners, but moved into marriage, are in the marriage with prior cohabitation category. We notice major differences between union types, especially within the demographic and the cultural categories, which suggests that individuals in common-law unions have characteristics that differ from those of people who marry.
Starting with the demographic variables, the difference between the average age at the start of first unions between types of union is negligible, averaging from 22.6 years among those married with prior cohabitation to 23.9 years for individuals in common-law unions. The majority of individuals did not experience the separation of their parents before the age of 18 years old (91%). Nevertheless, the proportion of individuals living common-law (21%) who did experience a parental separation is almost threefold that observed among individuals who moved into marriage without prior cohabitation (7%), which corroborates previous research that shows that individuals from divorced homes are more inclined towards cohabitation than direct marriage. The proportion of individuals in common-law unions who did not have children (71%) is much higher than those who ended up moving from cohabitation into marriage (57%) or entered marriage directly (42%)[Note 3]. However, the proportion of respondents living common-law (9%) who reported having stepchildren is greater than that of individuals who entered marriage directly (2%). Finally, the proportion of individuals who formed their first unions in the 1970s (76%) were more likely to marry directly than those in younger cohorts (49%) as common-law union has become a common mode of entry into conjugal life. In fact, a little over half of respondents who entered into their first unions in the early 2000s chose cohabitation.
Continuing with economic variables, the average educational level reached upon entering a first union is only slightly higher among individuals who entered marriage directly. This could be due to the fact that a higher proportion of individuals (30%) who cohabited first were still enrolled in school when they started living common-law compared to those who married directly (19%). As for employment status, the difference between types of union is negligible.
Finishing with cultural variables, region of current residence seems to influence the choice of entry into conjugal life since more than half of individuals currently living in Québec at the time of the survey started their conjugal life with nonmarital cohabitation compared to less than a third of those living outside of Québec. As for the size of the current community, it does not appear to influence the choice of union. Finally, individuals who cohabited first tend to attend religious services much less regularly than those who married directly. In fact, 55% of individuals in common-law unions and 53% of those who married after prior cohabitation don’t attend any religious services as opposed to a third of individuals who married directly.
Before looking into the determinants of first union dissolution, we examine conjugal changes experienced during the last decades in Canada. Table 2 shows the progression of nonmarital cohabitation as a mode of entry into conjugal life across cohorts. In Canada as a whole, direct marriage has lost ground over the years as the main mode of entry into conjugal life. It remains, nonetheless, quite prevalent in Canada outside of Québec. In Québec, the progression of cohabitation has been particularly strong in the last thirty years, as more than two thirds of individuals who entered a first union during the early 2000s chose cohabitation over direct marriage as opposed to a little less than half of individuals living in the rest of Canada.
Survival Analyses: Kaplan-Meier Survival Functions
Figures 1 and 2 illustrate the Kaplan-Meier survival functions for first unions in Québec and in the rest of Canada, by union type (time-varying covariate). These graphs allow for a few interesting observations to be made. They show that individuals in common-law unions have a very high risk of union dissolution across Canada; within ten years of cohabitation, well over half of common-law unions had dissolved. Interestingly, the rate of dissolution is higher in the rest of Canada (77%) than in Québec (63%), where unmarried cohabitation is more prevalent. These rates are more than twice those for marriages with prior cohabitation (26% in Québec and 24% in Canada) and direct marriages (20% in Québec and 19% in Canada). Within the first ten years of formation, individuals who lived in Québec at the time of the survey and married directly face a similar risk of experiencing a separation than individuals who lived in the rest of Canada. However, after ten years of marriage, they face a greater risk of dissolution; within thirty years of marriage, the proportion of first marriages that dissolved is close to half in Québec and around 40% in Canada. As for marriages preceded by cohabitation, the risk of separation is greater than for direct marriages across Canada, regardless of union duration. This supports the explanation according to which the experience of cohabitation, given its unstable nature, plays a predominant role in union dissolution. In the rest of Canada, the rate of dissolution for marriages with prior cohabitation accelerates after twenty years compared to that of Québec, suggesting that the effect of cohabitation may indeed be less pronounced in Québec than in the rest of Canada, where cohabiting unions are still less common.
Figures 3 and 4 illustrate the Kaplan-Meier survival functions for first unions of all types in Québec and in the rest of Canada, by cohort of union formation. These graphs show that union dissolution has considerably increased since the 1970s across Canada. However, the progression of union instability has slowed down, especially in the rest of Canada. The rate of dissolution within ten years of formation was only slightly higher in Québec (27%) for individuals who entered their first unions in the 1970s than for those living in the rest of Canada (24%). However, the gap between the two regions of Canada has expanded with younger cohorts; it was 44% in Québec for the 1990s cohort compared to 32% in the rest of Canada. This suggests that the progression of common-law unions in Québec is strongly associated with the progression of union instability observed here, although it might have slowed down with the youngest cohort of union formation. It is, however, too soon to draw conclusions in this regard.
Survival Analyses: Cox Proportional Hazard Models
Table 3 presents the results of the two separate sets of Cox proportional models that investigate which factors among demographic, socioeconomic and cultural covariates significantly affect the risk of the first union dissolution of marriages preceded or not by cohabitation, and of common-law unions in Canada [Note 4]. It shows clear differences in the way these factors influence the risk of union dissolution between the two types of union.
Table 3 replicates what was previously observed in figures 1 to 4 Kaplan-Meier survival functions. First, it suggests that union type has a significant effect on the dissolution of the first marriage, where the risk of separation is 42% higher for Canadians that married after cohabiting than for those who married directly. Therefore, the nature of cohabitation itself appears to be associated with a greater risk of separation among marriages preceded by cohabitation than direct marriages, all else being equal, as shown in figures 1 and 2. Moreover, table 3 indicates a clear progression of conjugal instability over time, as observed in figures 3 and 4. For married individuals, there is a steep progression of the risk of separation from one cohort to the next, which supports the hypothesis according to which there has been a shift in peoples’ values and expectations of marriage. For individuals who were in common-law unions, there is no significant difference in the risk of dissolution from the 1970s cohort to the 1980s. However, there is a significant increase from the 1970s to the 1990s and to the 2000s. As for the region of current residence, table 3 shows that individuals who were cohabiting and whose region of current residence is Québec have 19% less risk of union dissolution than those living in the rest of Canada, as shown in figures 1 and 2. This finding supports our initial assumption according to which the influence of cohabitation should be less pronounced in Québec since cohabitation is more common and accepted than in the rest of Canada. The region of residence also appears to be statistically significant among married individuals. Living in the rest of Canada reduces the risk of marital dissolution by 20%, which aligns with the fact that Québec has a higher divorce rate than the rest of Canada, as observed in figures 3 and 4.
Other factors are shown to affect the risk of union dissolution. Table 3 assesses the effect of having experienced parental separation before age 18, which is found to be statistically significant for both types of union, leading to 48% higher risk of separation among direct and indirect marriages, and 20% higher risk among common-law unions. The experience of parental separation before age 18 seems to have a stronger effect on marriages than on common-law unions, which may be due to the influence that this experience already has on the choice of union. Furthermore, our results confirm the hypothesis according to which the risk of dissolution decreases with the age at union formation. The risk of union dissolution is higher when individuals enter a first union before the age of 25, and is only slightly higher among marriages (21%) than common-law unions (17%).
Table 3 considers the effects of school enrollment, employment status, and the highest completed level of education on conjugal life. We observe that school enrollment increases the risk of separation by 32% for marriages and 36% for common-law unions. These results confirm the initial expectation that individuals who are still in school have less stable lives, and are not yet ready to make a long-term commitment. However, the influence of educational level on union disruption is not as clear; for marriages, having a high school degree increases the risk of separation by 18% compared to not having completed high school, but further schooling doesn’t significantly affect the risk of dissolution. For common-law unions, having pursued some postsecondary education reduces the risk of dissolution by 4%, but having completed a university degree increases that risk by 12%. Being employed increases the risk of dissolution by 22% among married individuals. However, it has only a small effect among individuals who were cohabiting. These last results seem to indicate that the effect of financial independence plays a more important role on the risk of marital dissolution than on disruption among common-law unions.
Our results show that the effect of religious observance on union stability is much stronger when partners are bound by marriage, as assumed earlier; the predicted risk of marital dissolution is 44% lower among those who attend religious services on a regular basis, and 20% lower among those who attend infrequently, compared to those who don’t attend at all. For individuals who were cohabiting, the effect of religious attendance remains significant although less important. Moreover, living in a rural area lowers the risk of union dissolution for marriages by 9% and common-law unions by 11%, which supports our initial assumption regarding the influence of community size on the risk of union disruption.
We observe that the timing of the conception (or adoption) of a first child affects the risk of first marriage dissolution; when conceived (or adopted) before the start of the union, the presence of a child increases the risk of separation by 31%, but when conceived (or adopted) after, it reduces the risk of separation by 34%. However, the timing of conception (or adoption) does not appear to matter as much in the case of cohabiting unions; when conceived before the start of the union, the presence of a child decreases the risk of separation by 13%, while when conceived after, the risk of separation is reduced by 28%.
Finally, our most interesting finding concerns the effects of having stepchildren. For married individuals, the presence of a stepchild raises the risk of separation by 74%, whereas for common-law unions the presence of a stepchild reduces the risk of dissolution by 45%. This suggests that in the case of marriages, the presence of a stepchild puts a strain on the marital relationship, while for common-law unions it involves a greater commitment to the relationship.
Discussion and Conclusion
Extensive research has been conducted to understand the progression of union instability in Canada and elsewhere in the western world. It has been shown repeatedly that the unstable nature of cohabitation is associated with a greater risk of union dissolution. Building on this finding, our main goal in this study was to examine the influence of different factors on the risk of union disruption for cohabiting unions and marriages, whether they are preceded or not by cohabitation, using a dynamic approach with retrospective data from the 2006 Statistics Canada General Social Survey. In contrast to most previous studies, which analyze the risk of dissolution for all types of union together, we ran separate models for the two types of union. This allowed us to assess whether these factors influence marriages and common-law unions differently.
As found in prior research, our results support the idea that the experience of cohabitation leads to higher rates of union dissolution. In fact, cohabiting unions remain more unstable than direct and indirect marriages even in Québec where they have become the new family setting. After controlling for premarital cohabitation, we observed that several factors are associated with higher risk of dissolution among marriages, such as being younger than 25 upon entering a union, experiencing parental separation before age 18, being in school, being employed, not attending religious services, living in an urban area, and forming a union in more recent years. These factors were also found to be significant risk factors for the dissolution of common-law unions, although the effects of parental separation, religious attendance and employment seem to have a lesser influence on common-law unions than on marriages. However, when looking at region of current residence, we noticed that region has a contrary effect on the two types of union; living in the rest of Canada, as opposed to living in Québec, reduces the risk of marital dissolution while increasing the risk of dissolution among common-law unions. This illustrates the divergence in demographic behavior of people living in Québec compared to those living elsewhere in Canada, which is consistent with findings from previous Canadian studies.
Finally, we were surprised to see that the timing of conception (or adoption) of the first child, and the presence of a stepchild have different effects on the risk of first union dissolution for marriages and cohabiting unions. For married couples, the traditional family setting where children are conceived after the start of the union is a deterrent to separation, while stepchildren or children conceived before the start of the union lead to increased rates of dissolution. For common-law unions, the birth or adoption of a common child acts as a stabilizing force between partners, regardless of the timing of the conception of the child. Moreover, the presence of a stepchild appears to increase the commitment to the relationship, and in turn reduces the risk of union dissolution. This last result contradicts prior research on union dissolution that has shown that family contexts other that the traditional setting are associated with higher risks of union dissolution. One possible explanation for this discrepancy could relate to our methodological choices. Instead of running analyses on all types of union together, as in previous studies such as that of Hall and Zhao (1995), which would have masked these effects [Note 5], we performed separate analyses for each type of union.
We can try to make sense of these unexpected findings in relation to the literature on cohabitation and stepfamilies. Stepfamilies tend to fare more poorly than intact families because stepchildren play a disruptive role on union stability as they compete with the stepparent for the biological parent’s time and attention (Cherlin 2009; White and Booth 1985). As they attempt to construct a new family life, they lack the kinds of norms, regulations and cultural expectations that guide traditional families (Cherlin 1978). The type of union in which stepfamilies are formed also tends to affect family stability. Indeed, the less institutionalized and less stable nature of common-law unions makes cohabitating stepfamilies more prone to disruption than married stepfamilies (Desrosiers, Le Bourdais, and Laplante 1995). Nevertheless, from the perspective of our study, the presence of stepchildren actually appears to play a role as a deterrent to the dissolution of common-law unions. Cohabiting unions come in various forms, ranging from highly unstable to long-term relationships that are preludes or substitutes to marriage (Dumas and Bélanger 1997). The presence of stepchildren might, therefore, increase the level of commitment within common-law unions that would otherwise be more casual. Nevertheless, it is not clear why the effect of stepchildren is so large considering that the presence of biological and adopted children has been accounted for. Other factors, such as the partner’s age at union and prior conjugal history, might contribute to the greater stability of cohabiting unions with stepchildren compared to those without stepchildren. Due to limits of the 2006 GSS data, our study could not take these factors into account.
With the growth of common-law unions and more diverse family structure in Canada, more research is needed if we are to deepen our understanding of the changing patterns of union and family formation, and their implications on family life and on children’s well-being. Future research on cohabiting unions could benefit from the use of longitudinal data that would include more detailed information on current and ex-partners. It could also benefit from the input of qualitative studies that would explore the dynamics of different types of cohabiting families (intact, blended, and stepfamilies).
1. Several respondents who cohabited before their marriage, and got married the same year of starting living together gave the years of entry into these unions, but not the months of entry. They are, nonetheless, coded as having started their union by unmarried cohabitation followed by marriage. Moreover, if respondents separated within the same month (and year) of starting their first unions, the duration of their first unions is coded as half a month.
2. “An urban area has a minimum population concentration of 1,000 persons and a population density of at least 400 persons per square kilometer, based on the current census population count. All territory outside urban areas is classified as rural. Taken together, urban and rural areas cover all of Canada.” (Statistics Canada 2010)
3. It is important to keep in mind that the children variable is a time-varying covariate that measures the influence of the presence of at least one child on the first union dissolution. Therefore, it might not be representative of the number of children that respondents have had by the time of the survey.
4. The direct comparison between the different models should be done with caution since the sample sizes are different.
5. This analysis was run, but the results were not included here.
France-Pascale Ménard is a PhD candidate at McGill University in Sociology. Her major areas of interest are family transformations, intergenerational transfers, social and family networks, family policies, and work and family balance. Her doctoral research focuses on the effect of family transformations associated with the decline of family size and marital instability on family networks at retirement age. She holds a FQRSC Doctoral Scholarship. France-Pascale can be reached at france-pascale.menard [at] mail.mcgill.ca
This paper is based on research carried out in the Québec Inter-University Centre for Social Statistics (QICSS), which provides researchers with access to detailed longitudinal survey data collected by Statistics Canada. The opinions expressed here do not represent the views of Statistics Canada.
Axinn, William G. and Arland Thornton. 1992. “The Relationship between Cohabitation and Divorce: Selectivity or Causal Influence?” Demography 29 (3): 357-374.
Beaujot, Roderic P. 2000. Earning and Caring in Canadian Families. Peterborough, Canada; New York: Broadview Press.
Brines, Julie, and Kara Joyner. 1999. “The Ties that Bind: Principles of Cohesion in Cohabitation and Marriage.” American Sociological Review 64 (3): 333-355.
Bumpass, Larry L., R. Kelly Raley and James A. Sweet. 1995. “The Changing Character of Stepfamilies: Implications of Cohabitation and Nonmarital Childbearing.” Demography 32 (3): 425-436.
Bumpass, Larry L. and Hsien-Hen Lu. 2001. “Trends in Cohabitation and Implications for Children's Family Contexts in the United States.”Population Studies 54 (1): 29-41.
Cherlin, Andrew J. 1978. “Remarriage as an Incomplete Institution.”American Journal of Sociology 84 (3): 634-650.
Cherlin, Andrew J. 2004. “The Deinstitutionalization of American Marriage.”Journal of Marriage and Family 66 (4): 848-861.
Cherlin, Andrew J. 2009.The Marriage-go-round: The State of Marriage and the Family in America Today. New York: Alfred A. Knopf.
Clark, Warren and Susan Crompton. 2006. Till Death Do Us Part? The Risk of First and Second Marriage Dissolution. Statistics Canada Catalogue 11-008: 23-33. Ottawa, Canada.
Desrosiers, Hélène, Céline Le Bourdais, and Benoît Laplante. 1995. “Les dissolutions d'union dans les familles recomposées: l'expérience des femmes canadiennes.” Recherches sociographiques 36 (1): 47-64.
Dumas, Jean and Alain Bélanger. 1997. Report on the Demographic Situation in Canada:1996. Statistics Canada Catalogue no. 91-209-XIE. Ottawa, Canada.
Hall, David R. and John Z. Zhao. 1995. “Cohabitation and Divorce in Canada: Testing the Selectivity Hypothesis.” Journal of Marriage and the Family 57 (2): 421-427.
Hamplová, Dana. 2003. “Marriage and Educational Attainment: A Dynamic Approach to First Union Formation.” Czech Sociological Review 39 (6): 841-863.
Laplante, Benoît. 2006. “The Rise of Cohabitation in Québec: Power of Religion and Power over Religion.” Canadian Journal of Sociology 31 (1): 1-24.
Le Bourdais, Céline and Évelyne Lapierre-Adamcyk. 2004. “Changes in Conjugal Life in Canada: Is Cohabitation Progressively Replacing Marriage?” Journal of Marriage and Family 66 (4): 929-943.
Le Bourdais, Céline, Ghyslaine Neill and Nicole Marcil-Gratton. 2000. “L'effet du type d'union sur la stabilité des familles dites " intactes ".”Recherches sociographiques 41 (1): 53-74.
Liefbroer, Aart C. and Edith Dourleijn. 2006. “Unmarried Cohabitation and Union Stability: Testing the Role of Diffusion Using Data from 16 European Countries.” Demography 43 (2): 203-221.
Milan, Anne, Mireille Vézina and Carrie Wells. 2007. Family Portrait: Continuity and Change in Canadian Families and Households in 2006, 2006 Census. Statistics Canada Catalogue no. 97-553-XIE. Ottawa, Canada.
Mongeau, Jaël, Ghyslaine Neill, and Céline Le Bourdais. 2001. “Effet de la précarité économique sur la formation d’une première union au Canada.” Cahiers québécois de démographie 30 (1): 3-29.
Raley, R. Kelly and Larry Bumpass. 2003. “The Topography of the Divorce Plateau: Levels and Trends in Union Stability in the United States after 1980.” Demographic Research 8 (8): 245-260.
Seltzer, Judith A. 2000. “Families Formed Outside of Marriage.” Journal of Marriage and the Family 62 (4): 1247-1268.
Smock, Pamela J. and Wendy D. Manning. 1997. “Cohabiting Partners' Economic Circumstances and Marriage.”Demography 34 (3): 331-341.
Statistics Canada. 2010. 2006 Census Dictionary. Statistics Canada Catalogue no. 92-566-X. Ottawa, Canada.
White, Lynn K. and Alan Booth. 1985. “The Quality and Stability of Remarriages: The Role of Stepchildren.”American Sociological Review 50 (5): 689-698.
Wu, Zheng and T.R. Balakrishnan. 1995. “Dissolution of Premarital Cohabitation in Canada.” Demography 32 (4): 521-532.
"This published article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Copyright for this article is held by McGill Sociological Review."