Class Differences in Women’s Cohabitation in Early Adulthood
It is common for couples of every social class to cohabit sometime in early adulthood in the U.S., and most marry later. However, women from different class backgrounds have different patterns of when (i.e. at what age) they cohabit and how much of their early adult years they spend cohabiting, married, and living outside a co-residential union.
In this post, for each age from 15 to 33, we show the percent of the time the average woman was cohabiting with a man, comparing women whose mothers had different levels of education. We present these results separately for white and black women.
While the figures show the percent of the year the average women cohabited, for brevity below we will refer to this as the percent of women who were cohabiting at this age. We also present parallel analyses for the percent of time at each age that the average woman was married to a man, and not living in a co-residential union with a man. We use the woman’s mother’s education as a measure of class background. We sometimes refer to this as “privilege,” or “advantage.”
More details of what we did are in the Technical Appendix at the end of this post.
Class Background Affects White Women’s Age Patterns of Cohabitation
While teen cohabitation is unusual, it is the most disadvantaged young women that are the mostly likely to cohabit that early, as our first graph below shows. Among white women whose mothers did not finish high school, 16% are cohabiting at age 18, and 20% at 19. Between 18% and 20% of these most disadvantaged of white women cohabit at each age until age 23, after which levels decline. By contrast, white women whose mothers have at least a bachelor’s degree cohabit less at every age than the group with the least educated moms, moving from 2% at age 17 to 10% at 20, reaching a high of 17% by age 26 and then falling. Those whose mothers had an intermediate level of education (finished high school) have levels of cohabitation in between the other two groups in their teens, but have the highest levels at some ages. It is women from the most privileged class backgrounds that stand out as distinctive: they cohabit the least at almost every age. Their peak age of cohabitation is also much later than for other groups.
Women who are not cohabiting with a man are either married to one, or they don’t live in a union with a man. (They may cohabit with a woman, or live with roommates, children, other kin, or alone.) In all groups, cohabitation declines starting at some age in the 20s. As the graph below shows, the proportion of white women who are married goes up during much of the age range. Is the reason that privileged women have lower cohabitation levels at most ages because they are more likely to be married? One might expect this because they readily meet more privileged men, and their combined resources are likely to be what contemporary couples believe is necessary for marriage. But the data do not support the idea that more advantaged class-background groups cohabit less because they marry more—at least for women under 30. We saw above that the most privileged white women are least likely to be cohabiting at all ages, and the graph below shows that they are also the least likely to be married at all ages before 30. At age 20, 26% of the most disadvantaged but only 9% of daughters of college graduates are married. Through the 20s marriage levels rise for all three groups of whites, but they rise at a faster rate for the advantaged. The most advantaged catch up by age 29 when slightly over 60% of each group is married. Until this age, the most advantaged white women are the least likely to marry and the least likely to cohabit. It is only after age 30 that the privileged have lower cohabitation levels which may be explained by their higher marriage levels.
Since, as we just saw, the most advantaged white women are the least likely to cohabit and least likely to be married in their teens and 20s, they must be more likely to live outside a co-residential union with a man. This is what the next graph shows.
Class Background Affects Black Women’s Age Patterns of Cohabitation
As is true for whites, the percent of black women cohabiting goes up with age, then descends, and the women with college educated mothers have the lowest percent cohabiting at most ages. Among black women, the class gradient on cohabitation is largely monotonic; those in the middle group of mother’s education are usually in between the other two groups (whereas the pattern was less consistent for the middle education group among whites). The patterns of cohabitation differing by class background are shown in this graph:
What about class differences among black women in marriage? Like whites, they too show the most disadvantaged women being most likely to marry early, with some crossover where more privileged women pull ahead, as the graph below shows. However, black women have smaller differences by class-background in their marriage levels than white women.
Are the class differences in cohabitation among blacks explained by class differences in marriage, or vice versa? As with whites, the answer seems to be no. Even more consistently than was true for whites, we saw a negative class gradient in black women’s cohabitation, such that those with more educated mothers are less likely to cohabit at most ages. But among blacks, marriage differs only modestly by social class, and through the mid-20s the small differences that exist cannot explain the cohabitation patterns because black women with the most educated mothers are much less likely to cohabit and slightly less likely to marry.
For blacks, the class gradient in time spent outside union with a man shows a similar class gradient as for whites; as the next graph shows, it is the most privileged women least likely to be in a union at most ages.
Race Differences in Levels and Patterns
We’ve focused mostly on class differences within race. Let’s shift the focus now to comparing black and white women at comparable levels of mother’s education. Black women generally have higher levels of cohabitation at most ages than do whites. Black women, as a whole, have lower levels of marriage; among white women, by their early 30s, over 60% are married, whereas no class-background group of blacks exceeds 44% at any age. Finally, at most ages, a higher percent of black than white women are not in unions with men. Here we don’t explore the reasons for these race differences. But we note that while mother’s education is a frequently used rough indicator for class background, and we advocate this for within race comparisons, comparing white and black women whose mothers had the same education is less likely to be indicative of similar socioeconomic circumstances. In a between-race comparison, it is likely that a black woman grew up in poorer circumstances than a white woman whose mother is equally educated, because the black mother is likely to have confronted race discrimination and have lower earnings. The same is likely to be true for the woman’s father if he is black. Moreover, and the black woman is more likely to have been raised by a single mother and perhaps largely had access to income from only one parent.
Summary and Discussion
We have focused on differences in cohabitation patterns by class background within race groups.
For whites, until age 30, women from the most advantaged backgrounds have the lowest cohabitation and marriage levels; they are the most likely to be living outside a co-residential union with a man. Thus, it is not higher marriage levels that explain the lower cohabitation levels of the most privileged women in these years before 30. It is more likely their later start in relationships, their focus on education, and their greater access to the resources necessary to rent their own apartment. Women whose mothers are the most educated marry later than other groups, but after age 30, they are the most likely to be married; thus after age 30 their higher opportunities for marriage may explain their lower levels of cohabitation.
Black women show some similar patterns by class background. Blacks show an even clearer class gradient in cohabitation, wherein the least advantaged have the highest levels of cohabitation, the most advantaged cohabit least, and those from middle class backgrounds are intermediate. (For whites, this describes the pattern for the top and bottom class background groups, but the middle group is less often in between for them.) Like whites, black women show a pattern wherein the least advantaged are more likely to be married very early, but there is a crossover in the mid-to-late 20s after which more privileged women are more likely to be married. But these class differences in marriage are smaller among black women than white. At most ages, the most privileged women are more likely to live outside a union with a man, and the most disadvantaged women the least likely to do so.
Shifting to patterns of race differences while holding mother’s education constant, black women spend less of their time in early adulthood married, more time cohabiting, and more time living outside a co-residential union with a man.
TECHNICAL APPENDIX: DETAILS OF OUR ANALYSIS
Data Set used
We used 2006-2010 data from the National Survey of Family Growth (NSFG). NSFGs have been conducted for many decades. This latest NSFG did interviews of women age 15-45 between 2006 and 2010.
While the dataset contained all races, we show results for nonHispanic blacks and nonHispanic whites only, focusing on the class background differences within these two groups. For brevity, we refer to these two groups as simply “blacks” and “whites.” We eliminated from the sample and do not show results for Hispanics, because they differ substantially between immigrants and non-immigrants and numbers of each are quite small for these age-specific analyses. We also eliminated those in other race/ethnic categories (e.g. Asians and Native Americans) from the analysis because the numbers were too small for these age-specific analyses.
Measuring Average Percent of Time Cohabiting, Married, and Not in a Union With a Man at Each Age
Women were interviewed about the dates of any cohabitation(s) and any marriage(s) in their past. They were asked to give the month in which each began and ended. The survey did not use the word “cohabitation,” but asked women about when they were “not married but living together with a partner of the opposite sex.” Our graphs show the average percent of time at each age that women (in various groups) were cohabiting, married, or neither, at each age, focusing only on unions with men, not same-sex unions. (Among women not living in a co-residential union with a man, we do not distinguish between those who are in no romantic/sexual relationship, in a relationship with a man but not living together, or in a relationship with a woman. Such distinctions would not have been possible because the NSFG did not collect histories from each respondent on the ages at which they cohabited with women, or were in relationships with men or women that were not co-residential.) Our focus is on cohabitation, but at some points we look at the percent of women in any group in each of the three mutually exclusive and exhaustive categories—cohabitation with a man, marriage to a man, and not in a union with a man.
More precisely, we compute for each woman, for each year of retrospective age since 15, what percent of the time (measured as % of the months) she was cohabiting (or married, or not in a union). While for any given woman the measure is what percent of the months of the year she was cohabiting, when a mean is taken for any group, it is the percent of the time the average woman in the group spent cohabiting (or married, or not in a union) in the year of that age. For brevity, we sometimes refer to this mean percent of time as “percent cohabiting” (or married, or not in a union) at a given age. Similarly, as short hand, we will refer to whether women are cohabiting, married, or not in a union, but the reader should recall that by union we refer to co-residential unions, and we are only tracking marriages or cohabitations with men, not with women.
OLS Regression Analyses
We estimated separate regression models for each year of age, with women (at that age) as the units of analysis. (Thus a woman who was 30 when surveyed, and thus reported on her cohabitation and marital history from whenever it began through her 29th year would be an observation in the dataset used for each year between 15 and 29. Women over 33 at survey are in the dataset used for each age.) Each of these 19 age-specific regressions (for age 15 to 33) were run separately for blacks and whites, for a total of 38 regressions.
The dependent variable of each of these regression models is the percent of the year the woman spent cohabiting. Independent variables include an indicator variable for whether the woman is an immigrant, and indicator variables for education: (Less than High School, High School Graduate or Some college, BA or more). Finally, we included variables representing birth cohort, measured as year of birth entered in linear, squared, and cubed form.
Then, from each regression we estimated a predicted value for the percent of time women of each class background (measured by mother’s education) were cohabiting, married, and not in a co-residential union with a man. We used the margins command in Stata to estimate, based on our regression results, the average predicted values percent of time cohabiting for the three fixed values of mother’s education, while letting all other controls vary observed across the whole sample across their full distribution for each estimate.
The NSFG survey conducted in 2006-2010 interviewed women from 15 to 45 years of age, and thus covers women born between 1961 and 1995. The estimates in this post can be generalized to these cohorts, and, to the extent that behaviors have been changing (e.g. the retreat from marriage—especially young marriage—has continued), our estimates should approximate an average from these cohorts.