## Abstract

In many countries, the tendency for highly educated women to marry down in education has markedly increased. Research has pointed to an oversupply of highly educated women—that is, a marriage squeeze affecting women—as the core reason for this phenomenon. This study aims to provide a more comprehensive understanding of the causes of this marriage trend by analyzing over-time data drawn from IPUMS International census microdata samples for 34 countries. Several key findings are notable. First, the degree of educational hypogamy is associated with the magnitude of the deficit in college-educated men in the marriage market, which is consistent with the marriage squeeze hypothesis. Second, the degree of educational hypogamy is related to the economic empowerment of college-educated women, even after accounting for the mating squeeze effect. Third, counterfactual simulations show that while the mating squeeze is the major driver of educational hypogamy in the majority of the sample countries, the economic empowerment of college-educated women plays an equally important role in several countries.

## Introduction

In many parts of the world, marriages in which the wife has higher status than the husband have been thought to challenge normative gender-role expectations (Bertrand et al. 2015; Cooke 2006; Liu and Vikat 2007; Zelditch 1964). Yet despite the cultural discomfort regarding a female status advantage in marital relationships, recent cross-national studies have reported that the tendency for women to “marry down” in education—also referred to as educational hypogamy—has markedly increased (Esteve et al. 2012; Esteve et al. 2016; Schwartz and Han 2014). According to these studies, educational hypogamy is now more common than its opposite (i.e., men marrying down in education) in almost every postindustrial country, and a similar trend is observed in many newly industrialized countries as well (e.g., Esteve et al. 2016; Lin et al. 2020).

Despite growing evidence of trends in educational hypogamy, relatively little is known about the factors that underlie this phenomenon. An important and frequently cited reason is the education-specific partnering squeeze against highly educated women. As studies have widely reported, women began to surpass men in college attainment in the 1990s (Buchmann and DiPrete 2006; Esteve et al. 2016; Van Bavel 2012), which has created an “excess” pool of college-educated women. Many researchers have assumed that this female-skewed sex ratio among college graduates is the driving force behind the growing tendency toward educational hypogamy (e.g., De Hauw et al. 2017; Grow and Van Bavel 2015). Similarly, the media have often portrayed this phenomenon as a mating crisis for highly educated women, who are increasingly facing a shortage of equally well-educated men (e.g., Birger 2015).

Of course, the unequal sex ratio in the marriage market is an important factor constraining one's partner choice. However, because this sex ratio approach is a purely numerical and structural framework, it tends to overlook other societal factors that may influence mate selection patterns. As scholars of assortative mating have pointed out, change and variation in educational assortative mating patterns are attributable to various factors affecting individual preferences and structural constraints and are not the result of any single factor alone (Blossfeld 2009; Kalmijn 1998; Schwartz 2013). As yet, however, little empirical research has been conducted to identify these factors and to assess the magnitude of the contribution of each factor relative to another.

To fill this gap in the literature, the current study examines what factors and societal changes best explain the rise in educational hypogamy in industrialized and newly industrializing countries since the 1960s. Drawing on comparative and macro-sociological research on educational assortative mating, I suggest several factors that may have potentially contributed to the growth in educational hypogamy: (1) a mating squeeze for college-educated women, (2) the economic empowerment of college-educated women, (3) urban concentration, and (4) premarital residential independence of young adults and the decline of parental control. Using a pooled time-series data set for 34 countries between 1960 and 2015 and a country fixed-effects regression model, I examine whether and how these four factors are associated with the degree of educational hypogamy. My analysis reveals that the partnering squeeze for college-educated women—that is, the relative shortage of college-educated men in the marriage market—is the major driver of educational hypogamy in many countries. Nonetheless, I find evidence that the availability of highly educated men is not the only factor determining the degree of educational hypogamy. In particular, my findings suggest that in a number of countries the increasing economic empowerment of women has played a role in facilitating educational hypogamy.

## Hypotheses

### The Marriage Squeeze Hypothesis

The pursuit of higher education used to be the purview of men in most societies. However, mounting evidence indicates that male dominance in higher education has dramatically reversed in the last few decades. In 1990, women began to surpass men in college attainment in nearly every industrialized country, and since then, women have steadily outpaced men in college degree attainment (Becker et al. 2010; Buchmann and DiPrete 2006; Esteve et al. 2016; KC et al. 2010).

This female advantage in higher education has far-reaching implications for marriage markets and union formation (see Van Bavel et al. 2018 for an extensive review). One important consequence is that the chance of marrying an “equally” educated man is not available to all college-educated women because there are not enough college-educated men in the marriage market. Those women facing this “male deficit” may choose either of the following options: (1) they may marry a less educated man than themselves (resulting in an increase in educational hypogamy); or (2) they may prolong their search until they find an equally educated partner, or forgo marriage altogether (leading to a decline in marriage rates among college-educated women).

A growing body of evidence suggests that the former scenario is more common (De Hauw et al. 2017; Esteve et al. 2016; Grow and Van Bavel 2015). For instance, Esteve et al. (2016) showed that the reversal of the gender gap in education is strongly associated with increased rates of educational hypogamy. There is relatively little compelling evidence that the female advantage in college completion necessarily leads to a decline in marriage rates among college-educated women. On the contrary, growing evidence indicates that highly educated women are now more likely to marry compared to their less educated counterparts (e.g., Goldstein and Kenny 2001; Perelli-Harris and Lyons-Amos 2016; Pessin 2018). In light of these facts, we can expect that the growing female advantage in college completion generates an education-specific mating squeeze, leading to higher rates of educational hypogamy. This suggests that there should be a positive relationship between the female-to-male ratio of college graduates and the degree of educational hypogamy.

• Hypothesis 1: The female-to-male ratio of college graduates is positively related to the degree of educational hypogamy.

### Female Autonomy/Gender Equality Hypothesis

Broadly speaking, the marriage squeeze hypothesis (and the following urbanization hypothesis) concerns the opportunity structure of the marriage market. Besides these structural factors, people's preference for marriage partners also strongly influences assortative mating patterns. The increasing prevalence of educational hypogamy may be a consequence of shifts in people's mate preferences, along with changes in the opportunity structure.

An important factor that shapes mate preferences is women's economic self-sufficiency. As scholars have pointed out, increases in women's socioeconomic status relative to men may reduce women's preferences for a highly educated man (Schwartz 2013). This argument is based on the premise that economically self-sufficient women can be less dependent on the income of their husbands and thus more able to enjoy the luxury of choosing romantic love as opposed to financial security (Fernández et al. 2005; Smits and Park 2009; Smits et al. 1998). This is called the female autonomy hypothesis (or more broadly, the gender equality hypothesis), and it generally predicts that women's economic empowerment will lead to less matching on socioeconomic status.

Consider a hypothetical situation in which (1) college-educated women, in general, do not participate in the labor market; and (2) they have to choose their spouses from among college-educated men and men who did not attend college. Since most of these hypothetically nonemployed women cannot support themselves financially, their mate choice will be influenced by a potential mate's financial contribution to the marriage. Consequently, women may be hesitant to marry down in education insofar as education is correlated with earnings. In contrast, if college-educated women engage in paid work as much as men do, the economic returns to marriage would become less crucial in their calculation because they have leeway to choose their mates on “non-pecuniary grounds” (Schwartz 2013:456). I propose the following hypothesis:

• Hypothesis 2: Progress toward gender equality among college graduates in the labor market is positively associated with the degree of educational hypogamy.

### Modernization Hypothesis

Inspired by the theory of modernization and individualization (e.g., Giddens 1991), assortative mating scholars have argued that industrialization increases society's general openness, thereby facilitating intermarriage between socioeconomically diverse individuals (Blossfeld and Timm 2003; Smits 2003; Smits et al. 1998). Since industrialization is a comprehensive transformation entailing various subprocesses, studies have sought to provide more concrete explanations as to why advanced industrialization would facilitate intermarriage. A review of these studies suggests that two subprocesses of industrialization are likely to promote marital openness: first, the increase in the share of the urban population (urbanization hypothesis); and second, increased residential independence of young adults and the consequential decline in parental control (independence hypothesis).

#### Urbanization

The probability of intermarriage depends on the availability of opportunities for out-group contacts and interactions (Blau and Schwartz 1984; Blossfeld and Timm 2003; Kalmijn 1998). If a person has ample opportunity to meet members outside his or her own group, this may increase the person's probability of educational intermarriage. Drawing on this idea, researchers have argued that greater urbanization may lead to higher rates of educational intermarriage (e.g., Smits et al. 1998). This is due to the commingling nature of urbanization: in the course of industrialization, people move out of small homogeneous communities to urban areas, where they have greater opportunities to meet people from other socioeconomic and cultural backgrounds (Kalmijn 1991). This increased intergroup contact may facilitate marriages between members of different groups.

More importantly, greater urbanization may increase tolerance toward nontraditional marital relationships, such as those characterized by educational hypogamy. Cities tend to produce and accommodate unconventional beliefs and behaviors with regard to lifestyle (Fischer 1995). By virtue of this liberal atmosphere, marital relationships that used to be considered nontraditional may be more easily formed in urban settings. Indeed, some studies show that certain types of marital union, such as interracial and same-sex unions, are more prevalent in cities than in rural areas (e.g., Mitchell et al. 2010; Rosenfeld 2007). In light of these studies, we would expect to find that higher levels of urbanization would facilitate educational hypogamy.

• Hypothesis 3a: The share of people living in urban areas is positively related to the degree of educational hypogamy.

#### Premarital Residential Independence

Urbanization is often associated with an increase in the number of young adults who migrate to urban regions. Scholars have pointed out that this geographic mobility of young adults may provide them with the so-called “independent life stage” (Rosenfeld and Kim 2005:542). This refers to the time in young adults' lives when they have moved out of their parents' home for reasons other than marriage, such as finding a job or pursuing a college degree.

During the independent life stage, adult children are physically and socially separated from their families of origin. According to Rosenfeld (2007), this physical separation inevitably weakens parental control over children's marriage. With the decline of direct parental intervention, young adults are able to explore a variety of partnerships beyond the oversight of their parents. As several researchers have suggested, this may result in nontraditional unions such as nonmarital cohabitation, interracial unions, and same-sex unions (Goldscheider et al. 2014; Rosenfeld and Kim 2005; Zhang and Sassler 2019).

The decline of parental control might be related to the increasing prevalence of educational hypogamy as well. Historically, people have preferred to avoid romantic relationships in which the woman has higher status than the man (Zelditch 1964). The disinclination regarding women's status advantage is still found among more recent generations, who tend to believe that it challenges the normative gender roles in romantic relationships (Atkinson et al. 2005; Fisman et al. 2006; Hitsch et al. 2010).

Given this lingering disinclination to female status advantage, we may assume that parents would strive to dissuade their children from forming educationally hypogamous partnerships. But as pointed out earlier, parental control is usually mediated through coresidence, meaning that the suppressing effect of parental control on educational hypogamy may be diminished as more and more young adults enjoy the independent life stage. Given this scenario, I expect to find that increases in the share of single adults living independently from their parents facilitate educational hypogamy.

• Hypothesis 3b: The share of young adults who do not coreside with their parents before getting married is positively related to the degree of educational hypogamy.

## Data and Methods

### Data

I use the Integrated Public Use Microdata Series-International (IPUMS International)

as a main data source (Minnesota Population Center 2020). IPUMS International is an ongoing data project that harmonizes and disseminates many decades of census microdata from much of the world (Ruggles et al. 2015). Its extensive geographic and temporal coverage provides a unique opportunity for comparative cross-national analyses of educational assortative mating.

Up to the present, 94 countries have participated in the IPUMS International project. Among the participating countries, I restrict my analysis to industrialized and newly industrialized countries in order to study comparable cases with respect to patterns of marriage and partner selection (e.g., median age at first marriage and age difference between spouses at the time of marriage). This results in 138 country–year observations from 34 countries. Among these countries, 15 are OECD members and four are key partners of the OECD—Brazil, India, Indonesia, and South Africa, which are often categorized as newly industrializing countries.1 The other 15 countries are non-OECD nations whose Human Development Index (HDI) was close to or exceeded 0.7 as of 2015, representing “high development.”

Within each country–year observation, I analyze heterosexual married or cohabiting unions in which both partners are aged between 30 and 40 years old. This specific age window is chosen for three reasons. First, the lower bound allows most individuals to have finished formal education and makes it unlikely that spouses' relative education changes afterward. Second, the upper bound makes it unlikely that the unions were formed under significantly different social conditions than for 30-year-olds. Third, the age range allows me to analyze nonoverlapping cohorts because, in the majority of cases, IPUMS International is based on decennial population censuses.

Prior studies of educational assortative mating have tended to analyze slightly younger age-groups, such as 25–29 or 25–34. These groups may be less appropriate for the current study because hypogamous marriages tend to be formed at later ages. For instance, Nomes and Van Bavel's study (2017) reported that educationally hypogamous marriages in Belgium tend to be formed between spouses who are generally older than their peers in educationally homogamous or hypergamous marriages. If a similar age selectivity is also the case in other countries, analyzing younger age-groups (e.g., 25–29) may significantly underestimate the prevalence of educational hypogamy.

Utilizing this age window can also help minimize biases from selective divorce and remarriage. To address this selection bias, studies have tended to restrict their analysis to either recently contracted marriages (e.g., Mare 1991; Qian and Preston 1993; Raymo and Xie 2000) or intact marriages in which both spouses fall into certain age-groups (e.g., Schwartz and Mare 2005). Because IPUMS International does not provide information on the date of union formation, there is no way to select unions based on when they were contracted. Hence, I have chosen the latter option.

### Dependent Variable

My dependent variable is the degree of educational hypogamy in a country. As noted, I define educational hypogamy as a heterosexual marriage or nonmarital cohabitation in which a woman with at least a bachelor's degree marries (or is in a union with) a non-college-educated man. To measure the degree of educational hypogamy for each country–year observation, I use the ratio-based measure that has been utilized in previous research examining global trends in educational hypogamy (Esteve et al. 2012; Esteve et al. 2016). This measure is defined as H = A / B, where A and B are the numbers of hypogamic couples (college-educated women marrying down) and homogamic couples (college-educated women marrying their educational equals), respectively. This straightforward measure captures the degree to which college-educated women opt for educational hypogamy as opposed to educational homogamy: H = 1 when the number of hypogamic marriages is equivalent to the number of homogamic couples, H > 1 when hypogamic marriages are more prevalent than homogamic marriages, and H < 1 represents the opposite situation.

Although this ratio-based measure is easy to calculate and interpret, it is not without limitations. An important limitation is that it is unable to account for the relative size of the two education groups. The role of relative group size for educational assortative mating has been formalized in prior studies (e.g., Besanceney 1965; Blau et al. 1982; Blau and Schwartz 1984) and was more recently examined by Lewis and Oppenheimer (2000; for the case of interracial marriages, see Choi and Tienda 2017). According to these researchers, educational assortative mating is influenced not only by the imbalance between the number of men and women available to marry within a certain education group (as indicated by the sex ratio approach), but also by the size of one education group relative to the others. For instance, if the pool of college-educated men is very small relative to that of non-college-educated men, college-educated women may face a “restricted market” for homogamous marriage but an extensive market for hypogamous marriage (Schoen 1986:50), even with a perfectly balanced sex ratio among the college-educated. In other words, the impact of the skewed sex ratio on educational hypogamy, if any, might be confounded with the relative group size of the college-educated and non-college-educated. Hence, to precisely estimate the impact of the marriage squeeze (as well as other explanatory factors) on educational hypogamy, the potentially confounding influence of relative group size should be netted out.

To achieve this, I rely on two strategies. First, I include the percentage of college graduates for each country–year as a control variable to adjust for relative group size. Second, I use a group-specific marriage rate suggested by Schoen (e.g., Schoen and Thomas 1990) as an alternative dependent variable, which adjusts for group size differences. More specifically, the male marriage rate for non-college-educated (group A) men marrying college-educated (group B) women is calculated by mWAB = CAB / Am, and the analogous marriage rate from women's perspective is calculated by fWAB = CAB / Bf. WAB denotes an observed marriage rate between males in group A and females in group B, and the superscripts m and f indicate the male and female populations, respectively. CAB indicates the observed number of marriages between group A males and group B females, Am indicates the observed group A male population, and Bf indicates the observed group B female population.2 Marriage propensity between group A males and group B females can be represented by the sum of the two group-specific marriage rates (mWAB + fWAB). As a robustness test, I will check whether using this alternative hypogamy measure as the dependent variable (without including the relative group size as a control variable) would generally yield consistent results. The summary statistics for the two different hypogamy measures—H and mWAB + fWAB—are presented in Table 1.

### Explanatory Variables

The marriage squeeze hypothesis is tested with an indicator based on the female-to-male ratio of college-educated individuals. This is calculated by dividing the number of college-educated women aged 30–40 by the number of college-educated men in the same age category. Thus, values greater than one indicate that there are more women than men among the college-educated in that age-group, which would trigger a partnering squeeze for college-educated women.

To test the female financial autonomy hypothesis, I use two indicators of gender equality in the labor market that are complementary to each other. One indicator is the gap in the employment rate between college-educated men and women. This is defined as the ratio of the employment rate of college-educated women to that of college-educated men aged 25–54. This indicator captures the quantitative aspect of gender equality in the labor market for highly educated individuals (i.e., how “many” college-educated women participate in the labor market relative to college-educated men). But a limitation of this employment rate–based indicator is that it is blind to the qualitative aspects of gender (in)equality in the labor market, such as women's relative pay or the concentration of women in low-paid industries and occupations. To capture these qualitative features of the gender employment gap, I use the degree of occupational sex segregation among the college-educated as an additional indicator. I employ this variable as a proxy for college-educated women's economic empowerment relative to men, given that occupational sex segregation has been consistently identified as a key source of the gender wage gap in various national contexts (e.g., Blau and Khan 2017; Boll et al. 2017; Gauchat et al. 2012; Mandel and Semyonov 2014). The degree of occupational segregation is measured by the index of dissimilarity, D (Duncan and Duncan 1955), based on the nine major categories in the International Standard Classification of Occupations (ISCO) scheme for 1988, as is available in IPUMS International.3 The original index D ranges from 0 to 1, where 0 indicates complete integration in the occupational structure and 1 indicates complete segregation. I reverse this original scale so that higher values indicate greater gender equality in the labor market.

The modernization hypothesis includes two aspects of social modernization: urbanization and the decline in parental control/constraints (i.e., premarital residential independence of young adults). Urbanization is measured by the ratio of the urban population to the rural population, with higher values representing greater levels of urbanization. Urban (rural) population indicates people residing in urban (rural) areas (percentage of total population) as defined by national statistical offices. The source of data for this variable is the United Nations World Urbanization Prospects: 2018 Revision. Premarital residential independence is measured by the proportion of single/never-married adults who are younger than 35 and do not coreside with their parents at the time of the survey. The data for this variable are obtained from IPUMS International. In addition to these two indicators of social modernization, I include GDP per capita (logged) as an indicator of the level of economic development. The squared term of this variable is also included to capture a possible nonlinear relationship between educational assortative mating and level of economic development (for a relevant argument, see Smits et al. 1998). Table 2 presents summary statistics for these four explanatory variables.

### Regression Models

To reduce the risk of omitted variable bias, I use a country fixed-effects regression model as this removes the confounding effects of any unobserved factors that are constant over time. For instance, sociocultural and institutional factors, such as the dominant religion or the political context, which have been found to affect the level of socioeconomic intermarriage, can be eliminated in the fixed-effects estimation (for relevant arguments, see Smits et al. 1998). Second, in addition to the country fixed effects, I include a set of time-varying control variables in the estimation. As mentioned earlier, I control for the percentage of college graduates to adjust for relative group size. I also control for the GDP growth rate because the speed of economic growth may influence educational assortative mating patterns (Smits and Park 2009; Smits et al. 1998). Finally, I include a linear time trend variable (measured in years) and its squared term to pick up any unobservable curvilinear time trend.

Table 3 presents results from the regression analysis in which the simple ratio-based measure of educational hypogamy (H) is regressed on the theoretically derived explanatory variables, and Table 4 uses the group-specific marriage rate as an alternative hypogamy measure. Since the latter measure already implicitly accounts for the relative group size issue, controlling for the relative size of the college-educated population is unnecessary. In addition to unstandardized regression coefficients, the partial eta squared ($ηp2$) effect size for each explanatory variable is presented; the effect size describes the ratio of the variance in the dependent variable that is explained by a given explanatory variable after accounting for variance explained by other predictors in the model. This partial effect size provides researchers with a guide to gauge the relative explanatory power of a given variable compared to other variables in the model.

## Results

### Regression Findings

In Model 1 of Table 3, the sex ratio variable (i.e., the female-to-male sex ratio of the college-educated among those aged 30–40) is positively associated with the level of educational hypogamy (b = .580, p < .001). This is consistent with the prediction of Hypothesis 1 and reaffirms prior findings that the decline and reversal of the gender gap in college completion are followed by the increasing prevalence of educational hypogamy among college-educated women. The partial eta squared effect size indicates that 15.6% of the variance in the level of educational hypogamy is explained by the sex ratio among the college-educated. Based on Cohen's (1988) benchmarks defining small (η2 = .01), medium (η2 = .06), and large (η2 = .14) effects, the effect size of the sex ratio variable can be referred to as large.

Model 2 introduces two indicators of gender equality in the labor market for highly educated individuals. Occupational segregation is positively associated with educational hypogamy (b = .963, p <.01). This indicates that as college-educated women become more economically empowered, the prevalence of educational hypogamy increases, even after accounting for changes in the sex ratio among the college-educated. The partial eta squared estimate suggests that the gender equality variable has a relatively smaller effect than the sex ratio variable. The second gender equality indicator, the employment rate gap, shows a negligible effect size (η2 = .006), and its association with educational hypogamy is not statistically significant. This suggests that an increase in college-educated women's labor market participation does not in itself lead to an increase in educational hypogamy.

In addition, Model 2 includes a set of variables related to modernization theory. Contrary to Hypothesis 3a, the level of urbanization is negatively associated with the prevalence of educational hypogamy (b = –.076, p < .001), with a nonnegligible effect size. This negative relationship contradicts the prediction of Hypothesis 3a that the growth in the size of the urban population will increase the tendency for educational hypogamy. I further checked this unexpected result with a few alternative indicators of urbanization, including the rate of urban population growth and the share of the population living in rural areas. Results from these robustness checks also suggested that the growth of urbanization is associated with a decline in educational hypogamy. I also tested for multicollinearity with VIF (variance inflation factor) because the unexpected sign on the coefficient might be a product of multicollinearity; this test did not show that multicollinearity is a problem. I discuss the implications of this unexpected finding in the Discussion section.

Model 2 shows that premarital residential independence of young adults is positively associated with the prevalence of educational hypogamy (b = .118). While the direction of the coefficient is consistent with Hypothesis 3b, the effect size is very small and is not statistically significant. As shown in the table, GDP per capita is not strongly associated with the degree of educational hypogamy. In summary, my data do not provide convincing evidence supporting modernization theory.

Table 4 presents results using the group-specific marriage rate. In general, I find that these results are comparable to those in Table 3. The sex ratio among the college-educated, as well as the level of occupational sex segregation, is positively associated with the degree of educational hypogamy. In contrast, the level of urbanization is negatively related to the degree of educational hypogamy when using either dependent variable. The consistency of the results indicates that my general findings are not sensitive to how the relative group size issue is accounted for.

### Assessing the Relative Contribution of Each Factor via Counterfactual Simulations

Results in Tables 3 and 4 show that the sex ratio of the college-educated and the economic empowerment of college-educated women are two important factors contributing to the rise of educational hypogamy, with the former having a larger effect size. An empirical question that is worth exploring further is whether the relative importance of the two determinants differs among countries and whether the relative importance has changed over time. I seek to answer this question by estimating the relative importance of these two explanatory mechanisms for each country.

To assess the independent contribution of each factor, I conduct a set of simulations for each country. The purpose of this simulation analysis is to compare actual trends in educational hypogamy with two counterfactual scenarios in which the degree of hypogamy is determined by a single factor alone. In the first scenario, I compute the degree of educational hypogamy by allowing the sex ratio of the college-educated to evolve as it actually did while holding the level of occupational sex segregation at its initial level. In the second scenario, I reverse the procedure by allowing the level of occupational sex segregation to change as it did over time while holding the sex ratio of the college-educated constant at its initial level. I then compare the counterfactual scenarios with observed levels of educational hypogamy. This allows me to examine how much of the growth in educational hypogamy within a country is due to the partnering squeeze against college-educated women versus the economic empowerment of college-educated women.

Counterfactual simulations reaffirm that the sex ratio of the college-educated is the more important driver of educational hypogamy. Nevertheless, the relative magnitude of the contribution of each factor varies across countries. For instance, while changes in the sex ratio played an important role in the majority of countries, in several countries both factors equally contributed to the rise of educational hypogamy or the relative importance of the two factors changed over time. To effectively describe these cross-country variations, I divide countries into three categories as shown in Figure 1. (I present two selected countries per category; graphs for all countries are available upon request.)

Each panel in Figure 1 plots the actual trend in educational hypogamy along with two counterfactual scenarios.4 Scenario 1 is the hypothetical trend that would have happened if there were only changes in the sex ratio of college graduates, while Scenario 2 shows what would have happened if there were only changes in the level of occupational sex segregation. The United States and Portugal represent the first category, in which Scenario 1 more closely represents what actually happened. In these countries, the rise in educational hypogamy has been mostly due to changes in the sex ratio of college graduates. For example, the increase in hypogamy in the United States from 1970 to 2015 was due simply to the increasing number of college-educated women relative to men, while the independent contribution of gender equality in the labor market was relatively smaller. More than 60% of countries in the sample (21 out of 34 countries) fall into this category.

France and Argentina represent the second category and illustrate somewhat different patterns compared to the first category. In these countries, both factors played a role in increasing educational hypogamy in the earlier periods of analysis, but after 1990 the importance of the partnering squeeze increased, whereas the role of gender equality in the labor market declined. This period-specific pattern is observed in 15% of countries in the sample (5 out of 34 countries).

Lastly, it appears that trends in educational hypogamy in the third category are not clearly explained by either of the two factors alone, suggesting that neither mechanism is necessarily more important than the other. About 20% of countries in the sample (8 out of 34 countries) are classified into this category.

In sum, the overall findings in Figure 1 show that the growth in educational hypogamy in the majority of countries has been largely driven by the changing sex ratio among college graduates. Nonetheless, I find evidence that the economic empowerment of college-educated women also played a nonnegligible role in increasing the level of educational hypogamy in several countries.

### Comparison of Trends in Educational Hypogamy Among Women

This study aimed to analyze the trends and determinants of educational hypogamy among college-educated women, with the intention of building on the recent literature about the reversal of the gender gap in college/tertiary education and its implications for changing patterns of mate selection (e.g., De Hauw et al. 2017; Esteve et al. 2012; Esteve et al. 2016; Goldin et al. 2006; Van Bavel 2012). While this focus has generated useful insights into highly educated women's mate preferences and mate choice, one may wonder about trends for moderately educated women as well. For instance, has educational hypogamy among such women increased as much as it has among college-educated women?

Table 5 presents a comparison of the prevalence of educational hypogamy among both college-educated and moderately educated women (i.e., women who finished secondary education but did not obtain a college degree). An increasing (or decreasing) trend is indicated by a positive (or negative) average rate of change in educational hypogamy per year. The relative prevalence of hypogamy among college-educated women (as measured by the number of educationally hypogamous marriages divided by the number of educationally homogamous marriages) has increased in most countries in my analytic sample (28 out of 34 countries). In contrast, the relative prevalence of educational hypogamy among moderately educated women has decreased in the majority of the sample countries during the same period (23 out of 34 countries). These findings suggest that moderately educated women's mate preferences and mate choice have contributed relatively little to the overarching trend toward educational hypogamy.

### Robustness Tests and Additional Analyses

Countries in my sample represent various continents of the world, ranging from Europe/North America to Latin America, Asia, and the Near East, which have different cultures regarding family formation. Moreover, the countries are at fairly different stages of socioeconomic development. Pooling of these diverse countries could be problematic if different countries have different causes of educational hypogamy depending on their level of socioeconomic development. This means that I need to consider at least some type of subgroup analyses to explore potential heterogeneity among countries. Hence, I conducted a set of regression analyses that included interaction effects between GDP per capita and the four potential reasons for educational hypogamy (i.e., marriage squeeze, gender equality, urbanization, and residential independence). Results from these interaction analyses showed that none of the interaction terms are statistically significant. This indicates that the determinants of educational hypogamy are generally similar across the countries in my sample. The results from this supplementary analysis are provided in the online appendix Table A1.

I calculated the measures of educational hypogamy using individuals aged 30–40. As mentioned earlier, I chose this age window—as opposed to younger age-groups (e.g., 25–29 or 25–34) that have often been analyzed—because analyzing the younger groups may underestimate the prevalence of educational hypogamy, given that hypogamous marriages tend to be formed at later ages (e.g., Nomes and Van Bavel 2017). Having said that, it would be worthwhile checking whether analyzing slightly younger age-groups would lead to different results. To do so, I reestimated the models in Table 4 using men and women aged 25–34 as an analytic sample. The results from this sensitivity analysis are similar to those from the original models, suggesting that my preferred results are generally robust to alternative age specifications (see the online appendix Table A2).

Finally, to assess whether my major findings are sensitive to influential single data points or single countries, I conducted robustness checks as follows. First, to check if the results were sensitive to the exclusion of single countries, I reestimated Model 2 in Table 3 by alternately excluding one country at a time. This procedure did not change the main findings in any significant manner. Second, I conducted a visual inspection with a diagnostic graph provided by leverage-versus-residual-squared plot. In doing so, I found that one single data point (Israel, 1995) is identified as an outlier and could potentially distort the estimates of regression coefficients. I performed regression analyses with and without this observation and determined that excluding it did not substantially change the results reported in Tables 3 and 4. Results from these robustness checks and sensitivity analyses are available from the author upon request.

## Discussion and Conclusions

What factors best explain the rise of educational hypogamy over the past several decades? Is it primarily a consequence of the gender gap reversal in college education? If not, what other factors have contributed to this marriage trend? This study seeks to shed light on these questions by analyzing over-time census data from 34 countries. Key findings and their implications can be summarized as follows.

First, I find that the rise of educational hypogamy in most countries is strongly associated with the increase in the female-to-male ratio of college graduates, which clearly demonstrates the role of the partnering squeeze for educational hypogamy. However, I find evidence that increases in hypogamy are also attributable in part to college-educated women's economic empowerment, even after the sex ratio effect is taken into account. This indicates that women's

preference for an educationally homogamous mate may change as they become economically self-reliant. This finding provides support for the argument that increases in women's economic status lead to less matching on education by allowing women leeway to choose their mates on “non-pecuniary grounds” (Schwartz 2013:456). Of course, this should not be interpreted as meaning that women who marry hypogamously will necessarily be the sole (or primary) breadwinner in their households. That claim can be tested only by examining the earnings trajectory of hypogamous couples, which is beyond the scope of this article.

The impact of women's economic empowerment on hypogamy could also potentially be understood from the perspective of a male searching for a mate. Women's wages can create an income effect that increases the attractiveness of a highly educated woman as a mate (e.g., Buss et al. 2004; Sweeney 2002). This implies that increases in women's economic prospects may result in competition between males (i.e., college-educated vs. non-college-educated men) for highly educated women, which in turn may determine the level of educational hypogamy. The degree of competition is likely to depend on men's economic prospects: if the labor market situation for men is good, wives' financial contributions would be less of a concern and thus the competition for highly educated women may be lower. In contrast, wives' financial contributions would become a more salient issue to men if the job market conditions are bleak for themselves. Moreover, how men think about appropriate gender roles (a cultural factor) may also interact with the labor market situation for men to affect their willingness to “marry up” in education. Considering this, a worthwhile direction for future research would be to investigate whether and how the declining labor market prospects of less educated men together with men's gender-role ideologies are related to the rise of educational hypogamy.

Second, an unexpected but interesting finding is the suppressing effect of urbanization on educational hypogamy. Although it is beyond the scope of this study to examine why increases in urbanization deter educational hypogamy, I suggest a few possible explanations here. One is the spatial segregation between socioeconomic groups. As researchers have indicated, advanced urbanization is usually accompanied by residential segregation along class lines (Farley 1977; Intrator et al. 2016; Nielsen et al. 2017; Reardon and Bischoff 2011; Telles 1995). Residential segregation in highly urbanized areas may restrict opportunities for intergroup interactions through the physical separation of different social groups. Moreover, in such highly stratified urban settings, psychological barriers between socioeconomic groups could be heightened (Maguire et al. 2016), rendering intermarriage across socioeconomic boundaries less likely. These hypothetical scenarios imply that spatial segregation in highly urbanized areas could potentially lead to fewer, not more, opportunities (or lower preferences) for educational hypogamy.5

This article's findings suggest that increases in educational hypogamy over the past several decades are not attributable to any single factor. Researchers have tended to assume that the rise of educational hypogamy was solely driven by the numerical imbalance among the highly educated, which was caused by the female advantage in higher education completion. This view has often been echoed in journalistic articles dubbing this marriage phenomenon as “a mating crisis” for highly educated women, who are facing a male scarcity problem. The results of this study demonstrate that although the male deficit is the most important factor, other factors have played a nonnegligible role in determining the level of educational hypogamy.

The present study is not without limitations. Because of its exclusive attention to structural factors, it directs little attention to the role of cultural factors that might facilitate (or deter) educational hypogamy. For instance, given that cultural discomfort regarding female status advantage in marital relationships has been common, a decline in adherence to traditional gender norms and values (e.g., gender essentialism and male primacy) might have contributed to the rise of educational hypogamy. The role of these gender-specific cultural changes was in part parsed out by the measurement of occupational sex segregation in the models, given that the level of such segregation in a country reflects gender-related cultural norms (e.g., Charles and Bradley 2009; Levanon and Grusky 2016). To directly capture changes in gender norms and ideologies and their relationship with the rise of educational hypogamy, I also tried including an indicator of male primacy as measured by the degree of son preference (i.e., the sex ratio at birth) in the models. When all other independent variables are controlled, this variable is not meaningfully associated with the level of educational hypogamy. This variable, however, may lack sufficient over-time variability and thus might not thoroughly capture change in gender norms and ideologies. Future research could profitably explore the relationship between changes in gender norms and the rise of educational hypogamy.

## Acknowledgments

I would like to express my gratitude to Mary Brinton, Alexandra Killewald, and Albert Esteve for their continuous support of this work. The article also benefited from feedback by Steven Ruggles and the participants of a virtual session at the 2020 annual meeting of the Population Association of America. Finally, I would like to thank Ohjae Gowen and Yun Zhou, who read the final draft and gave me valuable comments and suggestions.

## Notes

1

Some OECD countries participating in the IPUMS International project (e.g., Canada, Germany, the Netherlands, the United Kingdom) are excluded from the analysis because the SPLOC (spouse’s location in household) variable is not available for these countries. The absence of this variable makes it impossible for researchers to identify the characteristics of spouses, such as age and educational level. Other OECD countries (e.g., Colombia, Poland) lack harmonized labor market–related variables and are also excluded from the analysis.

2

Schoen’s (1988) well-known harmonic mean–based marriage rate is a further step forward in this group-specific marriage rate, which aims to estimate the degree of in-group/out-group marriages after controlling for the influences of both relative group size and a marriage squeeze. Since one of the purposes of the current study is to estimate the impact of the latter and compare it with the influences of other factors, the use of the harmonic mean–based marriage rate would be inadequate.

3

The categories include (1) legislators, senior officials, and managers; (2) professionals; (3) technicians and associated professionals; (4) clerks; (5) service workers and shop and market sales workers; (6) skilled agricultural and fishery workers; (7) crafts and related trades workers; (8) plant and machine operators and assemblers; and (9) elementary occupations. The Duncan and Duncan (1955:211) index of dissimilarity (D) is calculated using the equation

$D=12∑i = 1N|miM−fiF|.$

$M$ is the total number of non-college-educated males in the country, and $mi$ is the number of non-college-educated males in the ith occupation. $F$ is the total number of college-educated females in the country, and $fi$ is the number of college-educated females in the ith occupation. I calculate this variable based only on married adults aged 25–54.

4

Note that the measure of hypogamy in the graphs is based on the group-specific marriage rate (mWAB + fWAB). I have checked that the simulation results are basically the same if the ratio-based measure of educational hypogamy (H) is used.

5

If this is the case, we would find that the tendency for educational hypogamy is particularly lower in highly urbanized countries, where most people reside in urban regions. In other words, the suggested negative association may be driven by country–year observations that are hyperurbanized and where people of certain education credentials are highly clustered with one another. If so, the current measure of urbanization, which is not granular enough to catch such heterogeneity, may mask the true relationship between urbanization and the degree of educational hypogamy. This possibility has been explored by rerunning the full model with a spline function for urbanization. This piecewise function reveals that the relationship between urbanization and hypogamy is negative only at very high levels of urbanization, while low to moderate levels of urbanization are not negatively related to hypogamy. Admittedly, this is at best suggestive evidence and might not necessarily mean that the observed negative relationship between urbanization and educational hypogamy is explained by residential segregation between socioeconomic groups. Future research should use finer-grained data to identify why highly urbanized countries show a lower tendency for educational hypogamy.

## References

Atkinson, M. P., Greenstein, T. N., & Lang, M. M. (
2005
).
For women, breadwinning can be dangerous: Gendered resource theory and wife abuse
.
Journal of Marriage and Family
,
67
,
1137
1148
.
Becker, G. S., Hubbard, W. H. J., & Murphy, K. M. (
2010
).
Explaining the worldwide boom in higher education of women
.
Journal of Human Capital
,
4
,
203
241
.
Bertrand, M., Kamenica, E., & Pan, J. (
2015
).
Gender identity and relative income within households
.
Quarterly Journal of Economics
,
130
,
571
614
.
Besanceney, P. H. (
1965
).
On reporting rates of intermarriage
.
American Journal of Sociology
,
70
,
717
721
.
Birger, J. (
2015
).
Date-onomics: How dating became a lopsided numbers game
.
New York, NY
:
Workman Publishing
.
Blau, F. D., & Kahn, L. M. (
2017
).
The gender wage gap: Extent, trends, and explanations
.
Journal of Economic Literature
,
55
,
789
865
.
Blau, P. M., Blum, T. C., & Schwartz, J. E. (
1982
).
Heterogeneity and intermarriage
.
American Sociological Review
,
47
,
45
62
.
Blau, P. M., & Schwartz, J. E. (
1984
).
Crosscutting social circles: Testing a macrostructural theory of intergroup relations
.
New York, NY
:
.
Blossfeld, H.-P. (
2009
).
Educational assortative marriage in comparative perspective
.
Annual Review of Sociology
,
35
,
513
530
.
Blossfeld, H.-P., & Timm, A. (Eds.). (
2003
).
European studies of population: Vol. 12. Who marries whom? Educational systems as marriage markets in modern societies
.
Dordrecht, the Netherlands
:
.
Boll, C., Rossen, A., & Wolf, A. (
2017
).
The EU gender earnings gap: Job segregation and working time as driving factors
.
Journal of Economics and Statistics
,
237
,
407
452
.
Buchmann, C., & DiPrete, T. A. (
2006
).
The growing female advantage in college completion: The role of family background and academic achievement
.
American Sociological Review
,
71
,
515
541
.
Buss, D. M., Shackelford, T. K., Kirkpatrick, L. A., & Larsen, R. J. (
2004
).
A half century of mate preferences: The cultural evolution of values
.
Journal of Marriage and Family
,
63
,
491
503
.
Charles, M., & Bradley, K. (
2009
).
Indulging our gendered selves: Sex segregation by field of study in 44 countries
.
American Journal of Sociology
,
114
,
924
976
.
Choi, K. H., & Tienda, M. (
2017
).
Marriage-market constraints and mate-selection behavior: Racial, ethnic, and gender differences in intermarriage
.
Journal of Marriage and Family
,
79
,
301
317
.
Cohen, J. (
1988
).
Statistical power analysis for the behavioral sciences
(2nd ed.).
Hillsdale, NJ
:
Lawrence Erlbaum Associates
.
Cooke, L. P. (
2006
).
Doing gender in context: Household bargaining and risk of divorce in Germany and the United States
.
American Journal of Sociology
,
112
,
442
472
.
De Hauw, Y., Grow, A., & Van Bavel, J. (
2017
).
The reversed gender gap in education and assortative mating in Europe
.
European Journal of Population
,
33
,
445
474
.
Duncan, O. D., & Duncan, B. (
1955
).
A methodological analysis of segregation indexes
.
American Sociological Review
,
20
,
210
217
.
Esteve, A., García-Román, J., & Permanyer, I. (
2012
).
The gender-gap reversal in education and its effect on union formation: The end of hypergamy?
Population and Development Review
,
38
,
535
546
.
Esteve, A., Schwartz, C. R., Van Bavel, J., Permanyer, I., Klesment, M., & García-Román, J. (
2016
).
The end of hypergamy: Global trends and implications
.
Population and Development Review
,
42
,
615
625
.
Farley, R. (
1977
).
Residential segregation in urbanized areas of the United States in 1970: An analysis of social class and racial differences
.
Demography
,
14
,
497
518
.
Fernández, R., Guner, N., & Knowles, J. (
2005
).
Love and money: A theoretical and empirical analysis of household sorting and inequality
.
Quarterly Journal of Economics
,
120
,
273
344
.
Fischer, C. S. (
1995
).
The subcultural theory of urbanism: A twentieth-year assessment
.
American Journal of Sociology
,
101
,
543
577
.
Fisman, R., Iyengar, S. S., Kamenica, E., & Simonson, I. (
2006
).
Gender differences in mate selection: Evidence from a speed dating experiment
.
Quarterly Journal of Economics
,
121
,
673
697
.
Gauchat, G., Kelly, M., & Wallace, M. (
2012
). Occupational gender segregation, globalization, and gender earnings inequality in U.S. metropolitan areas.
Gender & Society
,
26
,
718
747
.
Giddens, A. (
1991
).
Modernity and self-identity: Self and society in the late modern age
.
Cambridge, UK
:
Polity Press
.
Goldin, C., Katz, L. F., & Kuziemko, I. (
2006
).
The homecoming of American college women: The reversal of the college gender gap
.
Journal of Economic Perspectives
,
20
(
4
),
133
156
.
Goldscheider, F. K., Hofferth, S. L., & Curtin, S. C. (
2014
).
Parenthood and leaving home in young adulthood
.
Population Research and Policy Review
,
33
,
771
796
.
Goldstein, J. R., & Kenny, C. T. (
2001
).
Marriage delayed or marriage forgone? New cohort forecasts of first marriage for U.S. women
.
American Sociological Review
,
66
,
506
519
.
Grow, A., & Van Bavel, J. (
2015
).
Assortative mating and the reversal of gender inequality in education in Europe: An agent-based model
.
PLoS One
,
10
,
e0127806
. https://doi.org/10.1371/journal.pone.0127806
Hitsch, G. J., Hortaçsu, A., & Ariely, D. (
2010
).
Matching and sorting in online dating
.
American Economic Review
,
100
,
130
163
.
Intrator, J., Tannen, J., & Massey, D. S. (
2016
).
Segregation by race and income in the United States 1970–2010
.
Social Science Research
,
60
,
45
60
.
Kalmijn, M. (
1991
).
Shifting boundaries: Trends in religious and educational homogamy
.
American Sociological Review
,
56
,
786
800
.
Kalmijn, M. (
1998
).
Intermarriage and homogamy: Causes, patterns, trends
.
Annual Review of Sociology
,
24
,
395
421
.
KC, S., Skirbekk, V., Barakat, B., Sanderson, W., Goujon, A., & Lutz, W. (
2010
).
Projection of populations by level of educational attainment, age, and sex for 120 countries for 2005–2050
.
Demographic Research
,
22
,
383
472
. https://doi.org/10.4054/DemRes.2010.22.15
Levanon, A., & Grusky, D. B. (
2016
).
The persistence of extreme gender segregation in the twenty-first century
.
American Journal of Sociology
,
122
,
573
619
.
Lewis, S. K., & Oppenheimer, V. K. (
2000
).
Educational assortative mating across marriage markets: Non-Hispanic Whites in the United States
.
Demography
,
37
,
29
40
.
Lin, Z., Desai, S., & Chen, F. (
2020
).
The emergence of educational hypogamy in India
.
Demography
,
57
,
1215
1240
.
Liu, G., & Vikat, A. (
2007
).
Does divorce risk in Sweden depend on spouses' relative income? A study of marriages from 1981 to 1998
.
,
34
,
217
240
.
Maguire, A., French, D., & O'Reilly, D. (
2016
).
Residential segregation, dividing walls and mental health: A population-based record linkage study
.
Journal of Epidemiology and Community Health
,
70
,
845
854
.
Mandel, H., & Semyonov, M. (
2014
).
Gender pay gap and employment sector: Sources of earnings disparities in the United States, 1970–2010
.
Demography
,
51
,
1597
1618
.
Mare, R. D. (
1991
).
Five decades of educational assortative mating
.
American Sociological Review
,
56
,
15
32
.
Minnesota Population Center
. (
2020
).
Integrated Public Use Microdata Series, International
(Version 7.3) [Data set].
Minneapolis, MN
:
IPUMS
. https://doi.org/10.18128/D020.V7.3
Mitchell, C., Fu, X., Heaton, T. B., & Jacobson, C. K. (
2010
).
Urbanization, education and racial intermarriage in Brazil
.
International Journal of Contemporary Sociology
,
47
(
2
),
273
294
.
Nielsen, M. M., Haandrikman, K., Christiansen, H., Costa, R., Sleutjes, B., Rogne, A. F., & Stonawski, M. (
2017
).
Residential segregation in 5 European countries: Technical report
(ResSegr Working Paper 2017:2). https://doi.org/10.13140/RG.2.2.18252.31365
Nomes, E., & Van Bavel, J. (
2017
).
Education and marriage: The shift from hypergamy to hypogamy in Belgium, A 20th century cohort analysis
.
Revue Quetelet Journal
,
5
(
2
),
37
67
.
Perelli-Harris, B., & Lyons-Amos, M. (
2016
).
Partnership patterns in the United States and across Europe: The role of education and country context
.
Social Forces
,
95
,
251
282
.
Pessin, L. (
2018
).
Changing gender norms and marriage dynamics in the United States
.
Journal of Marriage and Family
,
80
,
25
41
.
Qian, Z., & Preston, S. H. (
1993
).
Changes in American marriage, 1972 to 1987: Availability and forces of attraction by age and education
.
American Sociological Review
,
58
,
482
495
.
Raymo, J. M., & Xie, Y. (
2000
).
Temporal and regional variation in the strength of educational homogamy
.
American Sociological Review
,
65
,
773
781
.
Reardon, S. F., & Bischoff, K. (
2011
).
Income inequality and income segregation
.
American Journal of Sociology
,
116
,
1092
1153
.
Rosenfeld, M. J. (
2007
).
The age of independence: Interracial unions, same-sex unions, and the changing American family
.
Cambridge, MA
:
Harvard University Press
.
Rosenfeld, M. J., & Kim, B.-S. (
2005
).
The independence of young adults and the rise of interracial and same-sex unions
.
American Sociological Review
,
70
,
541
562
.
Ruggles, S., McCaa, R., Sobek, M., & Cleveland, L. (
2015
).
The IPUMS collaboration: Integrating and disseminating the world's population microdata
.
Journal of Demographic Economics
,
81
,
203
216
.
Schoen, R. (
1986
).
A methodological analysis of intergroup marriage
.
Sociological Methodology
,
16
,
49
78
.
Schoen, R. (
1988
).
Modeling multigroup populations
.
New York, NY
:
Plenum Press
.
Schoen, R., & Thomas, B. (
1990
).
Religious intermarriage in Switzerland, 1969–72 and 1979–82
.
European Journal of Population / Revue Européenne de Démographie
,
6
,
359
376
.
Schwartz, C. R. (
2013
).
Trends and variation in assortative mating: Causes and consequences
.
Annual Review of Sociology
,
39
,
451
470
.
Schwartz, C. R., & Han, H. (
2014
).
The reversal of the gender gap in education and trends in marital dissolution
.
American Sociological Review
,
79
,
605
629
.
Schwartz, C. R., & Mare, R. D. (
2005
).
Trends in educational assortative mating from 1940 to 2003
.
Demography
,
49
,
629
650
.
Smits, J. (
2003
).
Social closure among the higher educated: Trends in educational homogamy in 55 countries
.
Social Science Research
,
32
,
251
277
.
Smits, J., & Park, H. (
2009
).
Five decades of educational assortative mating in 10 East Asian societies
.
Social Forces
,
88
,
227
255
.
Smits, J., Ultee, W., & Lammers, J. (
1998
).
Educational homogamy in 65 countries: An explanation of differences in openness using country-level explanatory variables
.
American Sociological Review
,
63
,
264
285
.
Sweeney, M. M. (
2002
).
Two decades of family change: The shifting economic foundations of marriage
.
American Sociological Review
,
67
,
132
147
.
Telles, E. E. (
1995
).
Structural sources of socioeconomic segregation in Brazilian metropolitan areas
.
American Journal of Sociology
,
100
,
1199
1223
.
Van Bavel, J. (
2012
).
The reversal of gender inequality in education, union formation, and fertility in Europe
.
Vienna Yearbook of Population Research
,
10
,
127
154
.
Van Bavel, J., Schwartz, C. R., & Esteve, A. (
2018
).
The reversal of the gender gap in education and its consequences for family life
.
Annual Review of Sociology
,
44
,
341
360
.
Zelditch, M.Jr. (
1964
).
Family, marriage, and kinship
. In Faris, R. E. L. (Ed.),
Handbook of modern sociology
(pp.
680
707
).
Chicago, IL
:
Rand McNally
.
Zhang, X., & Sassler, S. (
2019
).
The age of independence, revisited: Parents and interracial union formation across the life course
.
Sociological Forum
,
34
,
361
385
.
This is an open access article distributed under the terms of a Creative Commons license (CC BY-NC-ND 4.0).