GENERATIONAL DIFFERENCES IN COHABITATION AND MARRIAGE IN THE U.S.

Abstract

We use data from pooled 2000–2004 Current Population Surveys to examine generational differences in cohabitation and marriage among men and women ages 20–34 in the U.S. Consistent with our expectation and in line with assimilation theory, levels of cohabitation rise across succeeding generations. In contrast, generational differences in marriage follow a curvilinear pattern such that those in the second generation are least likely to be married, which supports some contemporary extensions of assimilation theory. These patterns persist across education groups, and tend to hold across racial and ethnic groups, too, although among women, the predicted percentages cohabiting across generations vary widely by race-ethnicity.

As immigration increases in the United States, attention has focused on how new arrivals and their offspring fare across multiple social outcomes. One of the key questions surrounding the adaptation or assimilation of immigrants is the extent to which their family formation patterns will change in accordance to the dominant patterns in the receiving society. Generational differences in marriage timing and prevalence and levels of childbearing are often attributed to this process of incorporation (e.g., Arias 2001; Glick et al. 2006). Yet the incorporation process is complex and it is not always clear to what extent we should expect these family behaviors to vary from one generation to the next or whether such differences by generation status should be attributed solely to incorporation itself.

Although several studies (Glick, et al. 2006; Qian, Blair, and Ruf 2001; Qian and Lichter 2001; Sassler and Qian 2003; Bean, Berg, and Van Hook 1997) have explored the marriage and childbearing patterns of the immigrant generation and their descendents, less attention has been given to the family behaviors that represent newly emerging family options available within the receiving society (although see Landale and Fennelly 1992; Landale and Forste 1991). Union formation in the United States has undergone significant change in the last few decades as cohabitation has increasingly become the modal form of first union formation. Several recent studies have helped document the increase in the likelihood that marriages will be preceded by cohabitation (Bumpass and Lu 2000; Smock 2000) and the increase in the likelihood that childbearing occurs within cohabiting unions as well. Yet, most examinations of immigrant and second generation family formation have considered the receiving context to be fixed rather than the reality of shifting definitions of family and union choices. The primary example of this is the significant increase in cohabitation in the United States over the past few decades. It is not clear what the generational assimilation into family formation should look like in a context of such significant change within the native population.

In this paper, we compare the generational patterns of union formation, expanding beyond prior research on marriage to include cohabitation as well. To formulate our expectations regarding the relationship of generational status with cohabitation, we draw on research on racial and ethnic variation in cohabitation and marriage, highlighting the role of socioeconomic status. Our hypotheses are also informed by theories of assimilation, which guide research on immigrant nuptiality patterns (e.g., Arias, 2001; Gordon, 1964; Portes and Zhou 1992; Qian and Lichter, 2001). We ask whether cohabitation increases across generation in a manner consistent with expectations of assimilation theory, as groups are likely to become more comfortable with the nontraditional but increasingly normative pattern of union formation in the United States. Further, we speculate that the low marriage prevalence for second generation individuals may be offset by increases in cohabitation for marriage among these individuals. In other words, it is possible that prior research suggesting second generation individuals delay union formation for greater investment in human capital has missed a substitution of cohabitation for marriage by these children of immigrants.

Although our hypotheses are guided by previous research and theoretical work, ours is among the first analyses to include both types of union formation across multiple racial and ethnic groups and by generation status. Therefore, our primary goal is a thorough description of nuptiality and cohabitation patterns across these diverse groups.

BACKGROUND

Nuptiality

Most of the research on union formation among the immigrant and second generation has largely focused on marital unions and noted particularly high prevalence of marriage for immigrants with a decrease across generations. Immigrants who enter the US as adults may have particularly high marriage prevalence due to the selection of immigrants based on family reunification policy that favors marriage (Bean and Stevens 2003) and those constraints after migration that favor staying married (Bean, Berg and Van Hook, 1996). However, both the classic and segmented assimilation theories predict that the proportion married would decline across generations because the age of first marriage and the rate of marital disruption tend to be lower in many sending countries than in the U.S. (although segmented assimilation theory would predict greater declines for some racial-ethnic groups than others). Arias (2002), in her application of Gordon’s (1964) theoretical work to nuptiality patterns, pushes these ideas further by suggesting that if structural incorporation precedes cultural assimilation, the relationship between generation and marriage is likely to be curvilinear, with higher percentages married in the first and third generation and lower percentages in the second generation.

First generation immigrants, particularly those with low education and women, may also be more likely to be married because they encounter additional challenges beyond those faced by non-immigrants of similar life course stage and socioeconomic status (i.e., they are neither structurally nor culturally assimilated). For example, newly-arrived immigrants tend to have lower returns to education and experience than natives (Duleep and Regets 1999; Duleep and Regets 1997; Duleep and Dowhan 2002), fewer social and legal rights than citizens (Jasso and Rosenzweig 1990; DeVoretz and Pivnenko 2004, 2005; Zimmermann and Tumlin 1998; Huber and Espenshade 1997), and may struggle with the uncertainties and risks associated with unauthorized migration status (Bean et al. 2006; Kamo 2000). Women who followed a husband or partner to the U.S. (Cerrutti and Massey, 2001; Donato, 1993; Donato and Patterson, 2004; Hondagneu-Sotelo, 1992) may perceive less extra-familial social support than their husbands and experience feelings of insecurity and isolation (Parrado and Flippen 2006). With these challenges faced by the first generation, culturally-enforced norms favoring marriage may become accentuated. Prior research on marriage and gender roles suggests that “traditional” familial norms are reinforced among newly-arrived immigrants, particularly in times of hardship (Bean, Berg, and Van Hook, 1996; Parrado and Flippen, 2005; Stanton-Salazar and Dornbush 1995; Valenzuela and Dornbush 1994; Van Hook and Glick, forthcoming).

Based on the marital selectivity of immigrants and the higher age at first marriage in the US than many sending countries, the percent married should decline from the first to the second generation. The percentage married is also particularly lower in the second generation relative to the first among those who are more structurally assimilated (i.e., with higher educational attainments). Among the children of immigrants who arrived as young children or who were born in the U. S., education opens up pathways to economic mobility and security. Second generation women and women who arrived in the U. S. as young children may therefore delay marriage in lieu of pursuing higher education and, once married, may be more likely to divorce as the means for supporting themselves through employment increase. This is similar to patterns of low fertility observed among second generation women who pursue higher education (Espenshade and Ye, 1994). Marriage market constraints might also contribute to delayed marriage among the second generation. The children of immigrants (second generation and those arriving as young children) may be less likely to marry than either the first or third generations due to difficulty finding partners that are considered acceptable by both themselves and their extended family (Merton, 1957). Children of immigrants may prefer to marry co-ethnics because they retain many of the nuptiality norms of their parents or, perhaps, feel pressure from their parents to marry within the group (Zhou and Bankston, 1998). Yet as an upwardly mobile group, the children of immigrants are less likely than the first generation to come into contact with co-ethnics and more likely to develop intimate relationships with those outside their ethnic group (Merton, 1941; Qian and Lichter 2007). This would place the children of immigrants, especially those experiencing upward mobility (i.e., with higher levels of education), in relatively poor marriage markets and lead to lower rates of marriage. However, because intermarriage becomes more normative and the pool of acceptable marriage partners increases with generation, marriage rates are likely to be higher in the third than the second generation.

Cohabitation

It is not sufficient to examine marriage as the only type of union formation available to immigrants and their offspring (Landale and Fennelly, 1992). Cohabitation has increased dramatically over the past three decades, climbing from 500,000 couples in 1970 to nearly 5 million couples in 2000 (U.S. Bureau of the Census, 2001a) and serves as the modal path of entry into marriage (Bumpass, Sweet, and Cherlin, 1991). Increasingly, cohabitation is also a setting for child bearing and rearing, as 40 percent of American children are expected to spend some time in a cohabiting family prior to age 16 (Bumpass and Lu, 2000). Further, the significant scholarship on cohabitation in the United States suggests that the forces shaping probabilities of forming cohabiting unions are different than those for marriage.

Our understanding of the complexity of variation in cohabitation is largely informed by comparisons of Blacks and Whites and, more recently, for some Hispanics as well (Manning and Landale, 1996; Manning and Smock, 1995; Raley, 1996; Smock, 2000) with much more limited focus on generation status. Unfortunately, to our knowledge, there are no studies available on cohabitation among other racial and ethnic groups in the U.S., particularly those with large proportions of immigrants and second generation adults (e.g., Asians). Nevertheless, the prevalence of cohabitation varies considerably across racial and ethnic groups such that cohabitation is especially common among disadvantaged minority groups, including Blacks and Hispanics (Bumpass and Lu, 2000). Additionally, the purposes and motivations for cohabitation also vary by race-ethnicity. Cohabiting unions appear more similar to marriage among Blacks and Hispanics than Whites. Blacks are as likely as Whites to form unions, but are substantially less likely to form marital unions. Instead, Blacks often form cohabiting unions (Raley, 1996). In turn, these cohabiting unions are unlikely to be formalized through marriage among Blacks, even among those who report plans to marry their partner. The most likely outcome is actually remaining together in a cohabiting relationship (Brown, 2000). Unlike their White counterparts, Black and Hispanic cohabitors are less likely to marry in response to a pregnancy (Manning, 2004). And, more Black and Hispanic children will spend greater proportions of their childhoods in cohabiting families than White children (Bumpass and Lu, 2000). Taken together, these findings suggest that cohabitation is less often a stepping stone to marriage than a substitute for it among Blacks and Hispanics when compared to Whites.

Similarly, cohabitation appears to operate as a substitute for marriage among Puerto Ricans. Landale and Forste (1991) found that unions are as likely to begin through cohabitation as marriage among young mainland Puerto Rican women. Importantly, unions that begin informally are unlikely to be formalized through marriage. Moreover, such unions typically involve childbearing. Cohabiting Puerto Rican women tend to be more similar to their married than single counterparts in terms of education, employment, and childbearing (Landale and Fennelly, 1992). Indeed, when asked, most Puerto Rican women characterized their cohabiting relationships as a form of marriage.

Cohabitation is most common among those at the lowest socioeconomic strata in the United States and other countries as well. For instance, consensual unions have been quite common in many areas of Latin America and are not only a setting for child bearing and rearing but are also recognized by the state as a form of marital union (see Landale and Fennelly, 1992 for a summary). These consensual unions are usually associated with lower levels of education. A recent analysis of consensual unions in nine Latin American countries suggests these relationships are relatively stable and likely to occur across the life course (Castro Martin, 2002).

Although cohabitation has become increasingly prevalent across other groups, it continues to be a trend driven by those with fewer economic resources. According to Landale and Forste (1991), cohabitation serves as an adaptive family formation strategy for the disadvantaged by allowing union formation despite economic uncertainty. This notion is consistent with recent ethnographic work by Smock, Manning, and Porter (2005) that shows many cohabitors describe marriage as unattainable primarily because they lack the economic stability they consider a prerequisite for marriage.

Immigrants are much more likely to be Hispanic or Asian than are natives. Among immigrants ages 20 to 39, the majority (56.1%) is Hispanic, 21.1% are Asian, and only 15% are non-Hispanic White. Among natives, only 9.6% and 2.0% are Hispanic and Asian, respectively, and the majority (72.4%) is non-Hispanic White (authors’ analysis of the 2005 March CPS). Because racial and ethnic minorities have been found to have higher levels of cohabitation, immigrants may also be more likely to cohabit. Additionally, because different immigrants are likely to cohabit at different rates depending on the level of cohabitation in their country of origin, racial and ethnic differences in cohabitation in the United States may in part be attributable to compositional differences in nativity and generational status.

Predicting cohabitation by generation status

Based on the prior research on marriage, one could envision a similar pattern of cohabitation by generation status if cohabitation were subject to all of the same constraints as marriage. However, we know that US immigration policy does not favor cohabiting couples who are not considered family members. This would probably forecast lower levels of cohabitation among immigrants themselves coupled with the higher prevalence of marriage among immigrants when comparing to the second and third or higher generations.

We have less guidance on what to expect when predicting cohabitation among those in the second generation (i.e. children of immigrants). Previous research suggesting the second generation is likely to delay family formation in pursuit of higher human capital and upward socioeconomic mobility leads us to expect lower marriage prevalence among the second generation when compared to the first generation. However, it is conceivable that this generation, born and socialized in the United States, would substitute cohabitation for marriage much the way cohabitation has increased among those with college educations in the U.S. generally. In other words, the second generation may substitute cohabitation for marriage. In this case, we would expect to see lower levels of marriage but similar levels of union formation (i.e. marriage plus cohabitation) among the second generation when compared to the first generation. Then the curvilinear pattern of marriage by generation status we expect to observe will also be complimented by an inverted curvilinear pattern of cohabitation by generation status.

Alternatively, the second generation could have cohabitation patterns distinctive from those in the third generation if we consider the pace of assimilation. In this case, the fact that the second generation was raised by immigrant parents with greater adherence to “traditional” norms of union formation will suppress cohabitation among their children thereby generating a linear pattern of increased cohabitation with increasing generation status.

We also expect there may be some differences in the generational patterns of nuptiality and cohabitation across groups. For example, generational increases may be greatest among Asians because they begin in the first generation at very low levels of cohabitation, but less pronounced among Hispanics given the tradition of consensual unions in Latin America. Similarly, because the modern form of cohabitation is relatively common among Latin American women with higher education (Parrado and Tienda, 1997), we expect these immigrants with higher levels of education to have similar levels of cohabitation as their native-born counterparts. To examine these possibilities, we analyze the generational patterns of marriage and cohabitation across the largest panethnic and racial groups.

DATA

Using data from the combined 2000–2004 March Current Population Surveys (CPSs), we document the prevalence of cohabitation, marriage, and unions for men and women by generational status. The March CPS follows housing units over time for up to 16 months; approximately half of the sample in the March CPS for one year is eligible to be followed up in the following years’ March CPS. To remove duplicate cases, we restrict our sample to those in their first March interview. Because there were too few to analyze in a meaningful way, we excluded from our sample those identifying as American Indian or “other” race-ethnicity. Finally, we restrict our focus to adults of prime union formation ages, 20–34, yielding an analytic sample size of 168,024.

Measures

Union status is captured using three dummy variables to distinguish among respondents who are cohabiting, married, or unpartnered. Cohabitation status is based on a direct question about the relationship of individuals to the householder whereby a cohabiting partner is referred to as an “unmarried partner.” Those identified as an “unmarried partner” and householders living with unmarried partners are coded as cohabitors.

Generational status refers to the number of generations a person’s family has been in the United States. First generation individuals are defined as foreign-born persons of foreign-born parents. Family formation patterns of adult immigrants are clearly related to their selectivity into the United States based on immigration policy and preferences. However, there is a group of immigrants who are not selected based on their own marital status to whom we can look for a generational comparison. Immigrants who arrived as children will engage in family formation in the United States but will be closer to the migration process and have experience from the country of origin in a way not shared by children of immigrants born in the United States. For our purposes, the 1.5 generation captures first generation persons who arrived prior to age 12 since immigrants who enter as children or young adolescents presumably will be socialized in dating and courtship patterns in the United States. In contrast to those who spend the majority of their formative years in their countries of origin and are thus more likely influenced by the marriage markets and union formation traditions of the country of origin, we may expect those who arrived in the United States when they were younger than 12 to have patterns more similar to or just in between immigrants arriving as adults and the second generation. The second generation are U.S. born persons of foreign-born parents, and the remaining are third-or-higher generation individuals. Generational status is dummy coded: 1^st generation, 1.5 generation, 2^nd generation, and 3^rd generation (reference).

Our analyses control for several sociodemographic characteristics related to union status. Gender is coded 1 for men and 0 for women. Age is measured in years. Race-ethnicity distinguishes among the following groups: Black, Asian/Pacific Islander, Mexican, Puerto Rican, Other Hispanic, and Non-Hispanic White (reference). Unfortunately, the sample sizes are too small to divide Asians by country of origin, but we acknowledge the distinct linguistic, economic, and cultural backgrounds of immigrants in this group. Puerto Ricans, though not technically immigrants but subject to many of the same conditions (i.e. speaking a non-English language in the place of origin, moving to limited marriage markets), are examined separately from other Hispanics. Children measures the number of minor resident children in the household and is dummy coded into the following categories: one child, two children, three or more children, and no children (reference). Education is coded into dummy variables: less than high school (reference), high school, some college, and college degree or more. Employment is derived from the respondent’s report of his/her current labor force activity, and is categorized as full-time (reference), part-time, unemployed, and not in the labor force. The CPS provides income-to-poverty ratios, which we code as <100% poverty, 100–124% poverty, 125–149%, and 150+% (reference).

Analytic Strategy

We begin by documenting the prevalence of unions (i.e., cohabitation or marriage), marriage, and cohabitation separately for men and women as well as race-ethnic groups by generational status. We estimate multinomial logistic regressions predicting cohabitation versus marriage as a function of generation, while controlling for gender, age, age-squared, race-ethnicity, children, education, employment, and income-to-poverty ratios. We test whether generation and gender interact in their effects on union type, since union processes are likely to differ for men and women. Then we investigate whether there is a significant interaction between generation and education such that the negative effect of education attenuates across generations. Finally, we consider whether generational differences in cohabitation and marriage are modified by race-ethnicity. Results from the multinomial logistic regressions are presented graphically as predicted percentages.

RESULTS

Descriptive Results

As shown in Table 1, there is a curvilinear pattern of the proportion in a union across generational status such that those in the second generation are least likely to be in a coresidential partnership or marriage. Among men, 48% and 46% of the first and third generations, respectively, are in a union, compared to just 35% of the second generation. The generational differences for women are similar: 67%, 45%, and 52% of the first, second, and third generations, respectively, are in a union. Notably, the 1.5 generation men and women are less likely to be in a union than those who arrived as adults but appear similar to those in the second generation.

Table 1.

Percentage in a Union, Married, or Cohabiting by Sex and Generational Status

	Union	Married	Cohabiting	Cohabiting Among Those in Union
Men
1st Generation	47.5	42.7	4.8	10.1
1.5 Generation	34.8	30.2	4.5	13.0
2nd Generation	35.0	28.0	7.0	20.1
3rd+ Generation	45.7	37.5	8.3	18.0
Women
1st Generation	67.2	62.0	5.2	7.7
1.5 Generation	47.8	41.0	6.8	14.1
2nd Generation	44.6	37.0	7.6	16.9
3rd+ Generation	51.9	43.5	8.4	16.1

Source: 2000–2004 March Current Population Surveys (months-in-sample 1–4 only), ages 20–34.

The percentage married or cohabiting follows distinct patterns. For marriage, the pattern is analogous to that observed for unions; second generation members are least likely to be married. In contrast, cohabitation exhibits a linear increase, supporting the expectations of assimilation whereby family formation patterns become more similar to the receiving society as generational status progresses. Whereas about 5% of first generation men and women are cohabiting, roughly 7% of the second generation and over 8% of the third generation are in cohabiting unions. Restricting our focus only to those currently in a union reveals a similar pattern; cohabitation levels are higher for each successive generation, although roughly equal proportions of those in the second and third generations are cohabiting.

Table 2 shows the percentages married, cohabiting, and cohabiting among those in a union for each racial and ethnic group by generational status, separately for men and women. The overall percentages, shown in the first row for each racial and ethnic group, belie considerable variability by generational status. Supporting our hypotheses, the relationship between generational status and marriage is curvilinear such that the proportion married is lowest among the second generation. This curvilinear relationship is evident for all racial and ethnic groups among both men and women, except for Black men, White women, Puerto Ricans, and Other Hispanics for whom essentially the same proportions of those in the 1.5 generation and the second generation are married.

Table 2.

Percentage Married or Cohabiting by Sex, Race/ethnicity and Generational Status

	Men			Women
	Married	Cohabiting	Cohabiting Among Partnered	Married	Cohabiting	Cohabiting Among Partnered
NH-White	39.6	7.8	16.5	57.2	48.6	8.6
1st Generation	45.6	4.3	8.6	65.0	5.2	7.4
1.5 Generation	26.2	4.3	14.0	41.2	5.9	12.4
2nd Generation	31.7	7.2	18.5	42.7	9.0	17.3
3rd+ Generation	40.0	8.0	16.7	48.5	8.7	15.2
NH-Black	23.2	8.5	26.7	28.9	22.7	6.2
1st Generation	22.2	4.7	17.5	39.4	4.8	10.9
1.5 Generation	14.9	2.8	15.7	18.0	6.9	27.6
2nd Generation	14.2	7.4	34.3	15.2	4.2	21.6
3rd+ Generation	24.0	9.0	27.3	22.0	6.3	22.4
Asian/PI	32.7	3.1	8.7	54.1	49.1	5.0
1st Generation	41.4	2.4	5.4	62.0	3.1	4.7
1.5 Generation	26.1	2.9	10.0	42.0	5.2	11.0
2nd Generation	20.6	3.9	15.8	26.2	7.2	21.6
3rd+ Generation	27.2	6.3	18.7	37.0	12.5	25.2
Mexican	41.4	6.6	13.8	61.2	54.1	7.1
1st Generation	46.7	5.3	10.2	67.9	5.8	7.9
1.5 Generation	41.3	4.9	10.6	50.7	7.0	12.2
2nd Generation	30.7	7.4	19.5	42.0	6.6	13.6
3rd+ Generation	38.6	9.7	20.1	43.0	9.6	18.2
Puerto Rican	32.1	12.2	27.5	46.0	36.3	9.7
1st Generation	43.8	12.2	21.9	47.6	9.1	16.0
1.5 Generation	32.4	10.2	23.9	31.6	14.3	31.2
2nd Generation	29.7	11.7	28.2	33.4	9.0	21.3
3rd+ Generation	27.1	14.2	34.4	36.8	8.8	19.3
Other Hisp.	33.2	6.6	16.6	52.7	45.7	6.9
1st Generation	38.7	5.9	13.2	56.6	6.6	10.5
1.5 Generation	24.9	5.8	18.9	34.4	7.4	17.6
2nd Generation	23.3	6.2	20.9	33.1	6.3	16.0
3rd+ Generation	35.6	11.5	24.4	39.4	8.5	17.7

Source: 2000–2004 March Current Population Surveys (months-in-sample 1–4 only), ages 20–34.

However, for both men and women, generational status is positively related to cohabitation. Across most racial and ethnic groups, we observe patterns consistent with assimilation, although there are some important exceptions. Distinguishing between the two types of first generation immigrants does not yield the expected results among men for several racial and ethnic groups. Except for Asians, slightly more men in the first generation cohabit than in the 1.5 generation. Among Mexicans, Puerto Ricans, and Other Hispanics, the proportions of women cohabiting are actually a bit higher among the 1.5 generation than the second generation. Restricting our focus to only those who are in unions (shown in the third column), the proportions cohabiting (versus married) follow the expected assimilation pattern for both men and women in most racial and ethnic groups. A notable exception concerns the small declines in the proportions cohabiting between second and third generations among Non-Hispanic Whites, Black men, and Puerto Rican women.

Perhaps the most striking finding though is that third generation Asians exhibit the highest levels of cohabitation among women. Whereas less than 9% of Non-Hispanic White and Black third generation women are cohabiting, over 12% of third generation Asian women report living with an unmarried partner. We do not have a compelling explanation for this finding, but note that prior studies on race-ethnic differences in cohabitation have been confined to comparisons among Whites, Blacks, and Hispanics. We know essentially nothing about cohabitation among Asians, let alone generational differences in cohabitation among Asians. The percentages cohabiting among Asian men are roughly only half as large as those for Asian women, and are considerably lower than those for either White or Black men. It is possible that these patterns we find for cohabitation reflect the distinct gender patterns of intermarriage among Asians. Third generation Asians are more likely to intermarry or interpartner with Whites than their foreign born counterparts (Qian, Blair, and Ruf, 2001). And, although intermarriage varies considerably by country of origin, several groups with high intermarriage rates contribute to the “Asian” subgroup (i.e., Filipinas have considerably higher levels of intermarriage than their male counterparts). Perhaps second generation Asian women also cohabit rather than marry their partners from outside their own ethnic group. Although these are speculative conclusions at this point, the results clearly suggest a very different process at work among men and women.

Multivariate Results

Initial analyses (not shown, available from authors by request) reveal that gender and generation interact in their effects on union type. Thus, all multivariate models are estimated separately for men and women. Figure 1 demonstrates the importance of distinguishing between marriage and cohabitation and is derived from the Appendix Table. As shown in the first panel of Figure 1, which graphs the predicted percentage married, there is a curvilinear pattern such that the second generation is least likely to be married, and the generational differences in marriage are larger among women than men. For cohabitation (panel 2), there is strong evidence of assimilation among men such that their cohabitation levels increase linearly with generation. Among women, levels of cohabitation follow the same pattern as marriage (a u-shaped pattern with generational status). This result makes sense given prior research showing that Mexican-origin women are more likely than men to arrive in the U.S. as secondary or “tied” migrants (following or accompanying their husband, partner, or father) rather than primary migrants who come to the U.S. on their own (Donato, 1993; Donato and Patterson, 2004; Hondagneu-Sotelo, 1992; Cerrutti and Massey, 2001). Indeed, the pattern observed in Figure 1 for all groups together is likely to be heavily weighted toward the Mexican pattern since Mexicans comprise by far the largest immigrant group.

Figure 1.

Predicted Proportion Married and Cohabiting by Generational Status and Sex

The predicted proportions among those in a union that are expected to be cohabiting reveal the assimilation pattern for women, as shown in panel 3. This pattern was not initially evident because relatively few second generation women are either married or cohabiting. But among those in unions, cohabitation levels rise with generation for women just as they do for men.

We consider whether generation and education interact in their effects on the likelihood of being married or cohabiting and depict our findings in Figure 2. The generation and education interaction terms were statistically significant (results not shown). Regardless of generation, there is a modest positive association between education and the predicted proportion married among men. Indeed, the predicted proportions married among men vary little by generational status, particularly at lower levels of education, although there is some evidence that the relationship between generational status and marriage is curvilinear (U-shaped) for those with some college or a college degree or more. Among women, the first generation is most likely to be married, regardless of education. Those in the 1.5 generation are also more likely to be married than other groups, but only at lower levels of education as expected. The second generation is least likely to be married at all levels of education, except less than high school. This pattern of findings is consistent with our expectation that the second generation—especially those with higher levels of education—may delay or forgo marriage to invest in human capital.

Figure 2.

Predicted Proportion Married and Cohabiting by Generational Status, Education, and Sex

We hypothesized that the effects of education and generation on cohabitation would interact such that the negative effects of education would strengthen across generations and that generational differences in cohabitation would converge at higher levels of education. Our expectation is more clearly supported in the case of women than men. Among men, the effects of education on cohabitation appear to be similar for all generational groups and cohabitation levels increase from the first to the second and third generations regardless of education level. Among women, the generational differences in cohabitation within education groups are modest compared to men. Nonetheless, consistent with our expectation, the relationship between education and cohabitation is essentially flat for the first generation (and this pattern is more pronounced for men). Additional evidence of assimilation emerges when we consider the predicted proportion cohabiting among those in a union. Here, generational status tends to be positively related to cohabitation across education levels. Moreover, cohabitation increases more rapidly across generations for those with lower levels of education.

Finally, we examined the predicted proportions married and cohabiting across racial and ethnic groups, which are depicted in Figure 3. As expected, race-ethnicity and generation interact in their effects on union type (results not shown). The first panel depicts race-ethnic variation in marriage, showing a slight curvilinear association with generation for men. For all groups, a larger proportion of first generation men (i.e., who arrived age 12 and older) are expected to be married than those in the 1.5 generation. Among women, the association between generational status and marriage is also curvilinear such that those in the second generation are least likely to be married. Race-ethnic variation in marriage does not diminish across generations, suggesting a persistence of racial and ethnic distinctiveness consistent with a segmented assimilation perspective.

Figure 3.

Predicted Proportion Married and Cohabiting by Generational Status, Race/ethnicity, and Sex

The second panel shows the predicted proportions cohabiting. Among men, there is positive association between generational status and cohabitation that supports the assimilation hypothesis. In contrast, among women, there is a U-shaped pattern of cohabitation across generations, that is, similar to the pattern observed for marriage in the first panel (except among Asians whose predicted proportions cohabiting exhibit the classic assimilation pattern).

To isolate the patterns in cohabitation from those of unions, we focus on the predicted proportion cohabiting among those in a union, shown in the third panel. Among men and women, the evidence is consistent with the assimilation hypothesis for most groups in that cohabitation levels among those in unions increase across generations. This pattern is most clearly seen for Asians, among whom predicted cohabitation levels are particularly low in the first generation but increase in a step-wise fashion with increasing time and generations in the U.S. Among Asian women, cohabitation is exceptionally high among the third generation, a result that warrants further investigation. Nevertheless, the exceptionally low levels of cohabitation among first generation Asians are not unexpected as this group originates from regions of the world where cohabitation and consensual unions have been uncommon. This classic assimilation pattern is less apparent in the case of Hispanics, particularly Puerto Rican women, among whom cohabitation levels in the first generation tend to be quite similar to the second and (sometimes) the third generations. Again, the relatively high levels of cohabitation for Hispanics are not unexpected because Hispanics tend to originate from regions of the world where consensual unions are more common.

DISCUSSION

The past few decades have witnessed a substantial increase in the immigrant population as well as a growing prevalence of cohabitation. Additionally, cohabitation varies considerably by racial and ethnic group and is especially common among those with fewer socioeconomic resources, including Blacks and Hispanics. Despite these trends and in spite of the evidence of race-ethnic variation, the cohabitation experiences of immigrants have not been extensively investigated (although Landale and colleagues have studied cohabitation among Puerto Ricans).

Using data from the combined 2000–2004 CPS data files, we document unique associations between generational status, cohabitation, and marriage among men and women in the U.S. Consistent with contemporary assimilation theory, levels of cohabitation increase across generations for all racial and ethnic groups. Marriage levels, in contrast, exhibit a curvilinear U-shape such that marriage is least common among the second generation. These patterns are evident among both men and women although they are more pronounced among the latter. We offer speculative interpretations of these patterns in lieu of more detailed information and data but consistent with previous research on union formation in the United States.

Second generation adults may delay or avoid marriage but they are not necessarily opting out of residential unions, particularly among men. Rather, these adult children of immigrants are more likely to cohabit than their foreign born peers. Second generation adults may choose partners who are of different ethnic or cultural origins and choose to cohabit rather than enter a mixed marriage. Or, the cohabiting second generation adults may be on the road to marriage with their partners as they accrue greater stores of human capital via higher education or employment. This is consistent with a delay in other family formation behaviors among the offspring of the immigrant generation (Glick et al., 2006). Studies of family formation patterns as evidence of assimilation must consider the increasing importance of cohabitation among the second generation—a family pattern that was not socially acceptable to second generation children of earlier historical periods of U.S. immigration.

The generational patterns in marriage and cohabitation described above obtain across education groups, but are less consistent across racial and ethnic groups. The classic assimilation pattern of increasing levels of cohabitation with time and generations in the U. S. is most clear for Asians and least clear in the case of Hispanics. This is consistent with the classic assimilation perspective. Because cohabitation and consensual unions are relatively uncommon in Asian countries, it is possible that immigrants originating from Asia carry these norms with them when they come to the U.S. and are unlikely to cohabit. But cohabitation levels increase over time and generations in the United States (albeit to exceptionally high levels for Asian women as we discuss further below). In contrast, consensual unions are more common in Latin America. As a result, Hispanic immigrants may cohabit nearly as much as the second generation and (for Puerto Rican women) the third-or-higher generation, so the classic assimilation pattern does not appear. We caution that the relatively flat generational pattern for Hispanics does not necessarily mean that Hispanics do not acculturate. The meaning of cohabitation for Hispanics may change across generations even if the levels of cohabitation do not. Moreover, we do see evidence that by the third generation, race-ethnic differences in cohabitation converge toward the level of the dominant group, White natives. That is, Hispanics and Asians appear more similar to Whites in terms of the proportions cohabiting. Blacks retain some distinctiveness as their absolute levels of cohabitation as well as their relative increases across generations are comparatively modest.

In addition to documenting generational differences in marriage and cohabitation this study makes an important contribution by investigating cohabitation among Asians. We find exceptionally high levels of cohabitation among third-or-higher generation Asian women in both our actual weighted percentages and in our predicted percentages based on multivariate models that control for sociodemographic factors associated with union status and type. We are not aware of any cohabitation research that includes Asians and thus we cannot be too quick to dismiss our seemingly counterintuitive finding as there is no comparison benchmark. Nonetheless, this result needs to be replicated with other data because it is unexpected in light of the limited history of consensual unions in Asia. Based on assimilation theory, we would have expected Asians (even in the third generation) to have lower levels of cohabitation than Whites and Hispanics, for instance, but that is not what we found.

Our study has some limitations. As noted at the outset, the analyses we performed are descriptive, primarily because adequate data are not available to examine the processes involved in union formation and maintenance among various immigrant groups. Moreover, to the extent that formalizations of cohabitation through marriage differ by generation or race-ethnicity, we may be underestimating cohabitation for certain groups as we rely on cross-sectional data. Data constraints also necessitate the use of broad race-ethnic categories such as lumping together all Asian and Pacific Islanders, even though we expect there is valuable information to be gained from the sub-groups comprising this category.

This study extends prior research on generational differences in family behaviors by considering shifting union choices in the receiving contexts. We find that cohabitation and marriage are two distinct paths to family formation that are uniquely related to generational status. First, our results are consistent with prior analyses of nuptiality suggesting higher marital prevalence for the first generation than others. Our results are also consistent with prior work showing low union formation among the second generation relative to the first or third and higher generation. But, our analyses suggest this difference in union formation is magnified if cohabitation is not also considered. The second generation is more likely to form these informal unions than their immigrant counterparts although not yet to the level of those in the third and higher generation. Moreover, cohabitation levels vary by generational status within racial and ethnic groups suggesting some support for a segmented pattern but also indicating that studies of racial and ethnic variation in cohabitation should attend to nativity as well.