COHABITATION: PARENTS FOLLOWING IN THEIR CHILDREN’S FOOTSTEPS?

Abstract

As cohabitation has risen dramatically in the past few decades among adults of all ages, it is possible that middle and older-aged parents are "learning" cohabitation from their young adult children. The present study uses this theory as a guiding framework to determine if parents are more likely to cohabit themselves following the start of a young adult child's cohabitation. Using three waves of the National Survey of Families and Households (N = 275), results show that union formation patterns are influenced by young adult children among parents who are single at their child’s 18^th birthday. Parents are less likely to marry than remain single and are much more likely to cohabit than marry if they have a young adult child who cohabits. These results show support for the hypotheses.

Cohabitation has risen rapidly in the United States over the past 30 years. Today, there are over 5 million cohabiting couples in the United States, which is over 9 times the number of couples cohabiting in 1970 (U.S. Bureau of the Census, 2005). Cohabitation is occurring among people of all ages although less attention has been on those who cohabit in middle and later life. Only a handful of studies have attempted to describe older cohabitors and determine what factors are considered in their decision to cohabit rather than marry (Brown, Bulanda, & Lee, 2005; Brown, Lee, & Bulanda, 2006; Chevan, 1996; King & Scott, 2005). These studies have not considered the impact of socialization by young adult children on the decision to cohabit in middle and later life.

Socialization has in the past been considered a top down process; parents socialize their children, teachers socialize their students, and so on. However, theorists have come to realize that socialization continues throughout the life course. Reciprocal socialization is simply the notion that individuals are both agents and subjects of socialization. Socialization from child to parent has been discussed in the theoretical world; however, empirical studies of these processes have been rather limited (Crouter & Booth, 2003).

To date, there are no studies that have examined the effect of young adult children’s cohabitation experience on subsequent parental cohabitation behavior. One study did examine the influence of young adult children’s cohabiting behavior on parental attitudes toward cohabitation and found that young adults’ behavior positively influenced parental attitudes (Axinn & Thornton, 1993). Taking this a step further, I examine the influence of young adult children’s cohabiting behavior on parental cohabiting behavior. While it is noteworthy that cohabitation socialization from children to parents can only occur among a select group of people, examining this type of relationship is relevant nonetheless. These issues will be discussed in detail below.

Cohabitation is the modal pathway to marriage among young adults (Bumpass & Lu, 2000). Cohabitation is also more likely to occur among the previously married population than the never married (Bumpass & Lu, 2000). Furthermore, cohabitation among those 60 and older has increased from just 10,000 in 1960 to more than 400,000 in 1990 (Chevan, 1996). Moreover, in 2000, there were approximately 1.2 million cohabiting persons over age 50 in the United States (Brown et al., 2005; Brown et al., 2006). Demographic trends such as cohort replacement, a high divorce rate, and incidence of widowhood result in a greater number of older adults “at risk” to cohabit and this group will continue to grow as the baby boomers reach older ages. It is likely that middle- and older-aged adults learned cohabitation from their young adult children.

This study contributes to the literature on both reciprocal socialization in adulthood and cohabitation in later life. The study of cohabitation among older adults is distinct from the study of marriage as cohabitation is not institutionalized and, although becoming more prevalent, is still non-normative, particularly among older adults. The rapidly increasing phenomenon of cohabitation may be something older parents need to witness among those closest to them, namely their young adult children, before they would consider cohabiting themselves.

THEORETICAL FRAMEWORK

Two main theories have addressed the social psychological processes within parent-child dyads over the life course. Both the developmental perspective and the socialization perspective posit that during adolescence, children diverge from their parents to establish their own identity and they are influenced more by their peer groups and the media. While the developmental perspective posits that young adults’ and their parents’ attitudes converge and the socialization perspective posits that they continue to diverge, the undertone of both is that it is the young adults’ attitudes that are changing while their middle aged parents’ remain stable over time.

Miller & Glass (1989), using longitudinal data (a 14-year separation between two interviews; 1971 and 1985) to compare middle-aged child/elderly parent dyads (G2/G1) with young adult child/middle-aged parent dyads (G3/G2) tested these competing hypotheses. Only the middle generation (G2) changed their attitudes in all three arenas in the direction of their young adult children. Therefore, the similarity in attitudes among the middle-aged parents (G2) and their young adult children (G3) is due primarily to changes by the middle-aged parents and is an artifact of changes being made in the same direction by both the G2 and G3 generations. Neither hypothesis was fully supported.

The relatively new aspect of socialization theory that was not included in the study by Miller and Glass (1989) is the notion that socialization is reciprocal. Not only do parents socialize their children but children also socialize their parents (Putney & Bengtson, 2002). While it is generally accepted that individuals continue to be socialized throughout their lives, minimal attention has been given in the theoretical and empirical literature to how children continue to socialize their parents in young adulthood and later life. This pattern is consistent with reciprocal socialization between these two generations in which the young adult child is in fact changing somewhat from outside influences and bringing those messages back to their parents. Their parents then adjust their attitudes to match their children. This possibility was neglected in Miller and Glass’ conclusions. This is not to say that children are no longer learning from their parents or that tensions may not arise from new perspectives or attitudes, however, the introduction of new ideas being brought to the table by the young adult child can then be open for discussion and have the potential to be agreed upon by both parties.

Why does this matter for cohabitation behavior? Axinn and Thornton (1993), focusing on the intergenerational effects of attitudes and behavior, found that parents who did not view cohabitation favorably, whose young adult children subsequently cohabited, were likely to then favor cohabitation. After young adult children began to cohabit, parents’ attitudes toward cohabitation were more favorable, particularly when the cohabitation ended in marriage (Axinn & Thornton). This pattern is evidence of the reciprocal socialization process. The present study takes this a step further to determine if parental attitudes are then transferred into their own behavior thus making them more likely to form a cohabiting union themselves following the start of their young adult child’s cohabiting union. Given that attitudes are likely to predict behavior and young adult’s cohabitation behavior positively influences parental attitudes toward cohabitation, I hypothesize that parents whose young adult children cohabit will be more likely to cohabit themselves than their counterparts whose young adult children do not cohabit.

DATA AND METHODS

The main respondent data of the National Survey of Families and Households (NSFH1)¹, as well as the union history files² for the main respondents and the focal children are employed to conduct the analyses for the current study.

The analytic sample is first limited to those respondents who have a focal child and have union history data for both parent and young adult child (n=1579). The observation period cannot begin before the child’s 18^th birthday because the child, as well as the parent, needs to be at risk to cohabit. There were no children who began a cohabiting union before they turned 18 years old. Main respondents who cohabited at any time before the child was 18 years old are not included to avoid endogeneity (n=1039). In addition, respondents who are married at the child’s 18^th birthday are not included. Therefore, the final analytic sample includes main respondent/focal child pairs for which the main respondent was not cohabiting or married at the child’s 18^th birthday and had never cohabited before the child’s 18^th birthday (n = 275). The child’s 18^th birthday is the starting point so that, along with the constraints of the sample, both the parent and young adult child are at risk to form a cohabiting or marital union. I use a multinomial logistic regression model with cohabitation and marriage modeled as competing risks.

MEASURES

Dependent Variable

The outcome variable of interest in this analysis is whether or not the parent formed a cohabiting union. Using event history analysis, there are three outcomes that remove the main respondent (hereafter referred to as ‘parent’) from the risk set: (1) he/she begins cohabiting, (2) he/she gets married, or (3) he/she is not in a union by the NSFH3 interview date. The union history file is used to determine the dates of these events.

Independent Variables

The focal variable is whether the young adult child (hereafter referred to as the child) cohabited. This is established using a time-varying cohabitation variable which records a 0 for every month the child does not cohabit and 1 for each month after the child begins a cohabiting union. Note that the variable does not simply record the months they are in a cohabiting relationship because the focus here is on whether the parent cohabits any time after the child cohabits.

Parental attitudes toward cohabitation measured at NSFH1 are controlled in the analyses. At NSFH1, parents are asked their level of agreement with the following two statements: “It is all right for an unmarried couple to live together even if they have no interest in considering marriage;” “It is all right for an unmarried couple to live together as long as they have plans to marry.” The responses range from strongly agree (5) to strongly disagree (1). These responses are summed to create a cohabitation attitudes scale (Cronbach’s alpha = .62), with higher scores indicating more favorable attitudes toward cohabitation.

The gender of the parent and the gender of the young adult child are accounted for in the analysis. Research has shown that the mother-daughter relationship is more influential than the mother-son (Axinn & Thornton, 1993), father-daughter, or father-son relationship in socialization processes (Corsaro & Eder, 1995). In addition, cohabitation trends among the middle-aged and elderly population vary considerably by gender (Chevan, 1996), with men being more likely to form marital and cohabiting unions than women. Age of the parent at the beginning of the observation period is included and ranges from 34 to 70. Cohabitation varies considerably by race and therefore race of the parent is included as a dummy variable, non-White (1). Due to small ns in the non-White categories (Black, Hispanic) they had to be collapsed. A prior union history dummy variable is included to indicate whether the parent was ever divorced or separated (0) or never-married or widowed (1) before the beginning of the observation period. Having a child in the household may prevent a parent from having a partner move in, perhaps more so for a cohabiting partner than a spouse; therefore, a dummy is included to indicate whether there is a child under the age of 18 in the household (1). Note that this variable was originally divided into dummies for children under age five and children aged five and older (school-aged), however, there were small numbers of children under age five which resulted in zero-cell problems therefore the single dummy variable is used. Education and income of the parent at NSFH1 are also measured. Education is measured in years completed. Income is measured in dollars in the last year (logged for analyses to correct for skewness). Typically, cohabitation is more likely to occur among those of lower socioeconomic status (Manning & Smock, 1995).

RESULTS

Sample Characteristics

The unweighted sample means are shown in Table 1 for the full sample (N = 275). Not many parents form unions after their young adult child turns 18 years old. Only 16 percent (n = 45) form a cohabiting union and only 12 percent (n = 33) form a marital union. As expected, half (51%) of the young adult children form at least one cohabiting union during the observation period.

Table 1.

Sample Means (N = 275)

Variables	Mean	SD	Min.	Max.
Outcome:
Cohabit	0.16	0.37	0	1
Marry	0.12	0.33	0	1
No Union	0.75	0.45	0	1
Predictors:
Young Adult Child (YAC) Cohabits	0.51	0.50	0	1
Cohabitation Attitudes	5.02	1.90	2	10
Female	0.90	0.31	0	1
YAC Female	0.51	0.50	0	1
Black or Hispanic	0.28	0.45	0	1
Divorced/Separated	0.75	0.44	0	1
Widowed	0.16	0.37	0	1
Never Married	0.10	0.29	0	1
Age at YAC 18th Birthday	45.58	5.86	34	61
Child Under 18 in HH	0.69	0.46	0	1
Education	12.63	1.60	8	16
Income	16901	12558	0	74340

Parents’ attitudes toward cohabitation at NSFH1 are slightly disagreeable, with an average score of five. A score of six would indicate that they ‘neither agree nor disagree’ with both statements.

Ninety percent of the sample is female, which is expected given that these are middle-aged parents with children and men are more likely to be partnered in middle-age. The gender distribution of young adult children is evenly split. About 28 percent of the sample is non-White. Ten percent of parents were never married, 16 percent were widowed and 75 percent were divorced or separated. Average age of the parents is 45.6 years and ranges from 34 to 61. Seventy percent of the sample had at least one child in the household less than 18 years of age at the start of the observation. At NSFH1, average education is just over a high school degree and average yearly income is $16,901, with a range from $0 to $74,340.

Multinomial Logistic Regression Results

Table 2 shows the maximum likelihood estimates of the predictor effects on the relative risk of cohabiting versus remaining single, marrying versus remaining single, and cohabiting versus marrying. Model 1 shows the zero-order effects of the young adult child cohabiting. Having a young adult child who cohabits does not significantly change the odds of cohabiting versus remaining single. It is, however, significantly associated with a decrease in the likelihood of a parent getting married (about 85% lower odds) versus remaining single. As hypothesized, there is a 706 percent increase in the odds of forming a cohabiting union versus a marital union if a young adult child cohabits. As expected, if parents form a union, there is a greater likelihood for it to be a cohabiting union than a marital union.

Table 2.

Event History Analysis Estimates of the Relative Risk of Transitioning VS. Remaining Single & Cohabiting VS. Marrying (N = 275; Person-Months = 21,252)

	Model 1						Model 2
	Cohabit v. Remain Single		Marry v. Remain Single		Cohabit v. Marry		Cohabit v. Remain Single		Marry v. Remain Single		Cohabit v. Marry
Independent Variables	Odds Ratio		Odds Ratio		Odds Ratio		Odds Ratio		Odds Ratio		Odds Ratio
Young Adult Child (YAC) Cohabits^	1.22		0.15	**	8.06	**	0.94		0.16	**	5.73	*
Cohabitation Attitudes							1.17		0.86		1.36
Female							0.22	*	0.54		0.41
YAC Female							1.28		0.71		1.80
Black or Hispanic							0.63		0.20	*	3.10
Never Marrieda							0.24		0.26		0.96
Widoweda							0.52		0.86		0.61
Age at YAC 18th Birthday							0.93		0.86	**	1.08
Child Under 18 in HH							0.37	*	1.81		0.20	*
Education							0.92		0.75		1.22
Income (Logged)							1.07		1.23		0.86

Note: All outcome contrasts shown

^{^}Time-varying covariate;

^aDivorced/separated reference group

^†p<.10

^*p<.05

^**p<.01

^***p<.001 for two-tailed test

Model 2 includes all of the covariates. As the table shows, these covariates do not have much bearing on the outcomes. The inclusion of these variables slightly decreases the magnitude of the effect of a young adult child’s cohabitation on the odds of marrying versus remaining single (.151 to .165) and there is a decrease in the odds of cohabiting versus marrying although the odds ratio remains high (5.73). The controls that do have a significant effect on the outcomes are in the expected directions. Being female and having at least one other child under 18 years old in the household at the beginning of the observation period each reduce the likelihood of cohabiting versus remaining single (78% and 63%, respectively). Black and Hispanic parents have 80 percent lower odds of marrying versus remaining single than Whites. Additionally, being older at the beginning of the observation period is associated with lower odds of marrying versus remaining single (14% lower odds). Having a child under the age of 18 present in the household at the beginning of the observation period is associated with lower odds of cohabiting versus marrying.

In sum, having a young adult child who cohabits does not contribute to forming a cohabiting union instead of remaining single although it does reduce the odds of getting married versus remaining single and greatly increases the odds of forming a cohabiting union versus a marital union.

DISCUSSION

This study used a reciprocal socialization framework to examine the young adult child/middle-aged parent dyad in the context of cohabitation behavior. Research on reciprocal socialization processes has been rather limited; however, the available studies do support the presence of child-to-parent socialization. Cohabitation among middle-aged and older adults is on the rise. Perhaps middle-aged and older adults are learning cohabitation from their young adult children. Using longitudinal data from the National Survey of Families and Households, this study tested the influence of having a young adult child who cohabits on a parent’s own union formation behavior.

There are two main findings that contribute to our understanding of the reciprocal socialization process between parents and their young adult children throughout the life course and our understanding of cohabitation in middle- and later-life. First, while having a child who cohabits is not associated with the likelihood of forming a cohabiting union versus remaining single, it does decrease the likelihood of forming a marital union. Perhaps parents with young adult children who cohabit do not feel the need to be in a formal union. It would be interesting to include dating patterns of parents as well to test this theory, however, that is not possible with these data.

Second, and more importantly for the current investigation, among those who form unions, having a young adult child who cohabits dramatically increases the likelihood of forming a cohabiting union versus a marital union for parents. These findings show support for the hypothesis and may authenticate the notion of reciprocal socialization. This finding, coupled with the fact that parents’ attitudes toward cohabitation at NSFH1 are slightly unfavorable shows that, as Axinn and Thornton (1993) found, having a child who cohabits may change a parent’s attitude cohabitation and subsequently influence the parent’s union formation behavior at a later date. There appears to be a possibility that some parents are learning cohabitation behavior from their children, although the mechanisms through which this may happen cannot be tested here.

However, it is also possible that there are other operations at work which must be addressed. Parents, most of whom have been divorced, could be modeling negative attitudes toward long-term commitment and marriage through their own cohabitation behavior. These attitudes could have been expressed to young adult children and led to their cohabiting behavior in the first place. In other words, the possibility exists that parents’ negative attitudes toward marriage influenced their young adult child to cohabit and then influenced their own cohabitation behavior (Axinn & Thornton, 1996; Cunningham & Thornton, 2005).

The issue of selection is one that merits attention. The research question addressed here is about a select group of people. First and foremost, the question is about parents. Secondly it is about parents who cohabit for the first time after their child cohabits. This means parents could not have cohabited before they had children or while their children were young. It also means that they had to be unmarried at some point during the observation period. These factors admittedly make for a select group of people to include in the study. Limitations imposed to create the analytic sample include (1) parents had to have a child who was at least 18 years of age during the observation period, (2) parents could not have cohabited before the child’s 18^th birthday, (3) parents had to be unmarried and not cohabiting at their child’s 18^th birthday. These limitations result in a select group which only accounts for 12 percent of the total NSFH sample for which parent and child data are available. They also make defining a larger population to generalize to difficult. Nonetheless, the sample under observation for the current investigation has been defined as best as possible given the nature of the research question and the available data. Furthermore, the study presented here is one of the few empirical studies to address the notion of reciprocal socialization and the first to examine the effect of young adults’ cohabitation behavior on their middle aged parents’ likelihood of cohabitation. This study presents new ideas that may contribute to our understanding of (1) why people cohabit, particularly later in life, and (2) the socialization of parents by young adult children. Thus, it is a starting point for future, prospective research in the area.

There are a few other limitations to the current study aside from selection. First, the sample is rather small. There are relatively small numbers of parents who form either a cohabiting (44; 16%) or marital (33; 12%) union. These low numbers in the categories of interest suggest that the findings should be interpreted with caution. However, obtaining significant results with a small sample size means that it is likely these same effects would be detected in a larger sample.

Second, variables such as the relationship quality between the parent and the young adult child were not included. Measuring these types of variables at NSFH1 is not appropriate since the children are very young. However, since the data are available to empirically test this relationship, I included a global measure of relationship quality measured from the parent at NSFH1 and found no effect (results not shown).

Third, while there are advantages of using the NSFH, most of the young adult children turned 18 in the mid- to late-90s, when the observation period began. Therefore, some of the unions observed may have occurred up to 15 years ago. Updating this study with more recent data is required to see if these patterns would still exist today.

Fourth, there is no information on the union histories of other young adult children parents may have. There are a few problems with this lack of information. First, if the focal child had not cohabited during the observation period but another child had, I would be underestimating the effect of children’s behavior on parent’s behavior. Second, there is the possibility of an additive effect in that if two or more children cohabit, parents may be even more likely to cohabit themselves than when only one child cohabits. This possibility cannot be tested. Limiting this sample to parents who only have one child or who only have one child over the age of 18 would result in an even smaller and more select sample and thus would not improve the validity of the results.

Last, the influence of other people in parent’s lives, such as friends and siblings, is unknown in the current study. The young adult child’s influence is of interest here, however, it is possible that having friends or siblings who cohabit may influence the parent’s decision to cohabit as well. The observation period of this study occurred during a time of rapid increase in cohabitation, therefore, the possibility that parents “learned” cohabitation from other outside influences cannot be ignored.

This study has informed our understanding of the process of reciprocal socialization by showcasing that, while having a young adult child who cohabits does not increase the odds of a parent cohabiting versus remaining single, parents are more likely to cohabit than marry if they form a union. Although there are other possible explanations for future research to address, this study supports the possibility that some parents are likely to follow in their children’s footsteps.