Racial Pairings and Fertility: Do Interracial Couples Have Fewer Children?

Racial Differences in Fertility

The U.S. total fertility rate, although declining, remains among the highest of Western post-industrial countries. In much of Europe, especially in Southern and Eastern European countries, fertility levels are well below replacement (e.g., below 1.5 births per woman) (Adsera, 2011; Billari & Kohler, 2004). The high rates of U.S. fertility are exceptional in comparison, reflecting in large part the growth of racially-diverse immigrants of reproductive age coming from high-fertility countries (e.g., Mexico or other parts of Latin America). Birth registration data from the Centers for Disease Control and Prevention (CDC) reveal that the TFR of 1.73 in 2018 hides substantial racial and ethnic variation (Martin et al., 2019). Indeed, the TFR of the non-Hispanic White population is only 1.64, virtually identical to the rate among American Indians (1.65) but slightly higher than the TFR of Asian Americans (1.52). For Blacks, the TFR is higher at 1.79, but still well below the replacement level. Among Hispanics, the TFR plummeted over the past decade, to 1.96 in 2018. The total fertility rate among Hispanics fell by 31% from 2006 to 2017 (Alvira-Hammond, 2019).

These recent estimates of racial differentials are increasingly suspect in an era of growing interracial marriage. The CDC did not report biracial or mixed-race births until 2016, identifying the race or ethnicity of newborn infants based on the reported race and ethnicity of mothers only. Although registration data now include information about fathers' race/ethnicity, the CDC does not report fertility rates by racial pairings of both parents (Qian & Shen, 2020). However, such estimates are now available from nationally representative data on past-year fertility for marital and cohabiting unions identified in the American Community Survey. We ask: Do interracial couples, on average, have fewer children? If so, why?

Fertility in Racially Endogamous and Exogamous Unions

The usual assumption is that interracial unions have depressed fertility (Choi & Goldberg, 2018; Fu, 2008). Although societal acceptance of interracial marriages has increased over the past several decades, crossing racial barriers in marriage still often generates opposition from parents, relatives, and friends. Disapproval is expressed unevenly across racial pairings, depending on the stigma associated with each kind of marriage (Herman & Campbell, 2012). Interracial couples, for example, are unlikely to receive the same levels of social support given to endogamous couples. Children of interracial couples face adjustment difficulties as they grow up with—or without—acceptance from relatives, friends, and social networks (Childs, 2005; Root, 2001, 2003). Cheng and Powell  (2007) claim that to compensate, interracial couples make more educational investments in their biracial children. The parents from biracial families provide their children more resources, such as home computers or private school education, than their racially endogamous counterparts. Interracial couples nevertheless are less likely to develop strong social network ties and mobilize external resources than racially endogamous couples (Cheng & Powell, 2007). The financial and emotional costs of childbearing may affect the fertility decisions of interracial couples (Fu, 2008). Indeed, interracial couples are often concerned that their mixed-race children will be stigmatized, never fully accepted by family or friends on either side of the racial divide, and therefore will require greater parental commitments of time and energy (Romano, 2003; Root, 2001). Interracial couples may therefore have fewer children than they desire; childbearing will likely be depressed in comparison to racially endogamous couples. Our first hypothesis is that fertility is lower among interracial couples than among their endogamous counterparts.

Interracial cohabitation (as a share of all cohabiting unions) is more common today than interracial marriage (as a share of all marriages)(Choi & Goldberg, 2020). Yet, studies of childbearing among interracial cohabiting couples are uncommon, even though fertility among cohabiting couples is on the rise overall (Guzzo & Hayford, 2020; Lichter et al., 2016). In fact, nearly 60% of all U.S. nonmarital births now occur in cohabiting unions (Lichter et al., 2014); these births account for roughly 22% of all first births today, up from 12.4% in 2002 (Copen et al., 2013; Martinez et al., 2012). These figures may be biased if some cohabiting couples marry after becoming pregnant but before childbirth (Choi & Goldberg, 2020; Lichter et al., 2006; Sassler et al., 2018) or if cohabiting couples break up before birth (Lichter et al., 2006).

These complex underlying patterns of fertility decision-making make it difficult to forecast fertility rates among interracial cohabiting couples. As a baseline, we start with the premise that fertility among interracial cohabitating couples will be lower than fertility among their endogamous cohabiting and interracially married counterparts. Interracial couples are less likely than endogamous couples to transition from cohabitation to marriage, which suggests that interracial couples may be less committed to their relationships or that they acknowledge the challenges of mixed-race families in American society (Blackwell & Lichter, 2000; Kao et al., 2019). It is also the case that cohabitation, especially interracial cohabitation, is highly selective of more egalitarian partners with less traditional gender roles (Sassler & Miller, 2011). If so, fertility is expected to be lower among interracial cohabiting couples than among endogamous couples—their cohabiting or married counterparts.

The Racial Mix of Interracial Couples and Differential Fertility

A large demographic literature typically conceptualizes majority-minority marriages as a step in the assimilation process (Gordon, 1964), although such formulations seem increasingly anachronistic (Qian et al., 2018; Waters & Pinceau, 2016). Non-White minorities who married Whites presumably have adopted the cultural patterns (e.g., language, education, and residence) and have become more culturally integrated into mainstream American society (Gordon, 1964; Park & Burgess, 1969). Fertility among interracial couples should therefore be more similar to endogamous White couples than to their endogamous Hispanic, Black, American Indian, and Asian American counterparts.

Of course, America's immigrant populations have diverse national origins, racial and ethnic backgrounds, and economic resources. Cultural and economic assimilation is therefore highly segmented (Van Hook & Glick, 2020). Moreover, Alba and Nee (2003) have challenged classical straight-line assimilation theory and rejected the usual assumption that assimilation is a one-way process of immigrants and minorities adopting the cultural, social, and economic patterns of America's White middle-class mainstream majority. Indeed, the use of “majority” and “minority” itself risks implying a hierarchal relationship between unequal racial and ethnic groups (Buggs et al., 2020). The idea of a White American “mainstream” has now become more contentious than ever. While this assumption may have applied to White ethnic immigrants at the turn of the 20th century, it seems much less applicable today in America's racially diverse society. Still, immigrants and racial minorities invariably seek and achieve better lives—and integration—through schooling and upward mobility. Greater exposure and new opportunities for social interaction with native-born Whites may lead to intimate relationships and marriages that cross racial lines (Qian & Lichter, 2011). For Whites, intermarriage with minorities suggests an openness or acceptance to racial and cultural diversity, providing evidence of a two-way integration process between majority and minority populations (Alba & Nee, 2003). As a two-way integration process, interracial couples may or may not adopt or conform to the fertility behavior of Whites. The racial pairing presumably matters to the extent each partner brings different interpersonal resources to the marriage market, where they are located vis-à-vis Whites in America's racial hierarchy, and whether they face either racial antipathy or acceptance.

A recent study by Choi and Goldberg (2018) illustrates this general point. They used data from the 2002 annual file and 2006–2015 continuous file of the National Survey of Family Growth to compare pregnancies of interracial couples with pregnancies of racially endogamous couples. They found that Black-White couples—but only those involving White women—had a rate of pregnancy that was 77% higher than racially endogamous White couples, a result that could not be explained by socioeconomic disparities or other couple-level factors. Black-White couples involving Black women had significantly lower pregnancy rates than Black-White couples involving White women (see appendix table A11 in Choi & Goldberg, 2018). Long-lasting racial discrimination and prejudice against Blacks may indicate strong opposition to intermarriage between Blacks and Whites. Perhaps paradoxically, however, Black-White unions may have overcome serious challenges to their relationships, becoming more committed and resilient and achieving higher fertility in the process (Fu, 2008). This is speculation that requires empirical study.

For other interracial couples, Choi and Goldberg (2018) reported pregnancy rates that were more similar to those of endogamous White couples than to endogamous Black or Hispanic couples. Hispanic-White marriages may consist of Hispanic partners who themselves often identify as White and therefore are likely to have fertility patterns similar to endogamous White couples (Qian & Cobas, 2004). These results, although focused on pregnancies rather than births (which is our focus), seemingly suggest that intermarriage and fertility are markers of cultural assimilation, reflecting diverse racial realities and hierarchy (Van Hook & Glick, 2020). To the extent that racial stigma is high (e.g., in the case of Black-White marriages, which are comparatively rare vis-à-vis American Indian-White, Asian American-White, and Hispanic-White marriages), we hypothesize that fertility rates will be more closely associated with the fertility levels of the stigmatized non-White populations than with the White population. This stands in contrast with the alternative hypothesis that interracial fertility between non-Hispanic Whites and non-Whites may lay intermediate between the fertility rates of the two populations.

Gender and Fertility in Interracial Couples

Much of the research on fertility differentials focuses on women rather than taking into consideration the fertility desires of both partners (Nitsche & Hayford, 2020). Women's fertility results from her intentions and decisions to engage in sexual activity, use contraception, or bring the pregnancy to term (Morgan & Taylor, 2006). Women's partners also matter although it is often the case that partners typically share similar childbearing goals or aspirations (Thomson et al., 1990). Among partners who disagree, fertility rates usually fall midway between couples who agreed on having larger families and couples who agreed on having smaller families (Thomson et al., 1990). Fertility decision-making is a matter of compromise, reflecting the desires of both partners (Ray et al., 2020).

For interracial unions, observed racial differences in fertility suggest that partners may have different fertility desires from those in endogamous marriages. It is unclear, however, whether the fertility behavior of interracial couples should fall midway between the average fertility levels of each race represented in the union. For example, if intermarriage with Whites is selective of racial or cultural groups that have assimilated or who are similar to Whites on other characteristics associated with fertility, such as education, then interracial couples are likely to conform to the fertility levels of Whites. It also is possible that interracial unions are selective of economically-advantaged male and/or female partners. If so, fertility may well fall below average along the usual socioeconomic fertility gradient (Dribe et al., 2017).

Childbearing patterns may also reflect the fertility preferences or the racial backgrounds of one partner over the other. Indeed, marriage is a highly gendered institution that reflects cultural attitudes and norms supporting traditional gender relations or, in some cases, patriarchy (Bittman et al., 2003; Sayer et al., 2011). Traditional gender roles mean that women may have difficulty realizing their own fertility aspirations if they are different from their male partner's desires. For interracial couples with White partners, a gender perspective also suggests that fertility levels may vary by race of the male partner. Non-White women may adopt the fertility patterns of Whites while White women may follow the patterns of their Black, American Indian, Asian American, or Hispanic spouses. Selection into intermarriage also may play a role. For example, Asian American and Hispanic women in endogamous unions may be subject to traditional gender ideologies, including heightened pressure to have more children (Espiritu, 1997; Landale & Oropesa, 2007). One reason for out-marriage may in fact be to break away from traditional cultural or patriarchal norms (Mishra, 2018; Vasquez-Tokos, 2017). Among Black women, historically high labor force participation rates also suggest less adherence to traditional gender roles, even when married to a White man. Their gender and motherhood ideology may instead emphasize economic self-sufficiency and financial independence from men—any man, regardless of race (Collins, 2009; Florian, 2017). Fertility rates among interracially married American Indians are also lower than among endogamous American Indians (Eschbach, 1995). Our baseline hypothesis is that fertility rates among White male/non-White female couples will be more similar to those of endogamous White couples than their endogamous Black, Hispanic, American Indian, or Asian American counterparts.

Increasingly, men and women with more education or earnings potential are more likely today than in the past to marry rather than cohabit or to transition from cohabitation to marriage before childbearing (Ishizuka, 2018; Lichter et al., 2016), even as women are increasingly likely to “marry down” (Qian, 2018). This suggests less patriarchy and greater decision-making power among women in today's egalitarian unions. However, a recent study of fertility intentions among couples with at least one child revealed that male partners more strongly influenced subsequent fertility than did female partners (Ray et al., 2020). Traditional gender roles still matter and, in fact, may be stronger among racially disadvantaged families, especially if women occupy a subservient economic role. For interracial couples, this also suggests that non-White males may exert more influence on fertility than their White female partners. We hypothesize that fertility rates among non-White male/White female couples will thus be more similar to those of the endogamous non-White couples than to endogamous White couples.

Other Factors Affecting Fertility

Our empirical approach also accounts for other variables that matter in fertility decision-making among endogamous and exogamous couples. Here, women's reproductive ages range from ages 15 to 50, which capture cohabitation and marital timing and other life course events, including schooling and labor force participation (Seltzer, 2019). Another important consideration is marital order. Men and women in their first marriage are much more likely to have children than those in remarriage. Similarly, the number of older children living in the household is likely to shape recent fertility. Nativity status also plays a role in two ways: foreign-born couples are more likely to have higher fertility rates than native-born couples and in cases of mixed-nativity, native partners may have more fertility decision-making power than their foreign-born partners (Parrado & Morgan, 2008). Educational pairings of partners are also expected to affect fertility (Yang & Morgan, 2003). Highly-educated couples typically have lower fertility than their less educated counterparts (Yang & Morgan, 2003). We expect that more educated partners will have more fertility decision power than less educated partners.

In addition to the aforementioned individual and couple attributes, national and neighborhood context also influences fertility behavior (Browning & Burrington, 2006; South & Crowder, 2010), just as it shapes patterns of union formation, including whether couples are racially endogamous or exogamous (Campbell & Martin, 2016; Qian et al., 2018). For example, following national trends over the past decade, interracial fertility rates are likely to have declined even as interracial coupling has increased. A large international literature of the fertility transition also reveals that reproductive values and behavioral norms (e.g., contraceptive use) often diffuse from high-SES populations (e.g., high income or education) to other populations, even in cohabiting unions (Casterline, 2001; Vitali et al., 2015). Normative constraints on fertility, however, are less likely to be observed in racially- and economically-diverse neighborhoods and communities—those with greater shares of mixed-income, minority, multiracial, and foreign-born populations. Non-White or interracial couples in diverse rather than predominately White neighborhoods are less likely to conform to White fertility patterns.

Current Study and Hypotheses

Data and Measurement

We use data from the American Community Survey (ACS), which samples about 3 million households annually. Because interracial fertility and interracial relationships are relatively rare, we pool data from the 2008–2017 annual files of the ACS to increase sample size. The ACS includes census-like information on marital status and year of marriage, and asks whether women aged 15–50 had a birth in the past 12 months, which we use as a measure of recent fertility. We limit the sample to the heterosexual couples who married in the past 5 years. By considering only recently married couples, we reduce selection biases associated with differences across couples in marital duration and instability, which is observed disproportionately among mixed-race couples (Zhang & Van Hook, 2009). Because only women aged 15 to 50 were asked the question about the past-year fertility, we select only couples that include female partners in this age range.

Our analyses also consider currently cohabiting couples (defined by whether the household includes an “unmarried partner” of the householder). The ACS provides no information about when cohabiting unions started, which means that we are unable to restrict cohabiting couples formed in the past 5 years. This suggests the need for a cautious interpretation when making fertility comparisons between married and cohabiting couples. Fortunately, the overwhelming share of cohabiting unions lasts less than 5 years (Bumpass & Lu, 2000), with most of them ending in dissolution rather than marriage (Lichter et al., 2006).

For our purposes, we link individual records of each co-residential married or cohabiting partner into a couple record. The racial and ethnic identity of each partner is defined in Directive 15 by the Office of Management and Budget (OMB). Following these guidelines, we first distinguish Hispanics (of any race) from non-Hispanics. Non-Hispanics are classified as White, Black, Asian American, American Indian, and Native Hawaiian or Other Pacific Islanders (NHPI). The number of recent interracial marriages is insufficient in the ACS to conduct separate analyses of the NHPI population, which mostly includes the original peoples of Hawaii, Guam, and Samoa. Because Whites represent a statistical majority of the total U.S. and adult populations, we sometimes refer to the other racial and ethnic groups, when combined, as either the (statistical) minority or as non-Whites (i.e., as racial groups other than non-Hispanic White). Couples are then classified into the following categories: (a) both White, (b) both Black, (c) both American Indian, (d) both Asian American, (e) both Hispanic, (f) White male/Black female, (g) Black male/White female, (h) White male/American Indian female, (i) American Indian male/White female, (j) White male/Asian American female, (k) Asian American male/White female, (l) White male/Hispanic female, (m) Hispanic male/White female, and (n) minority male/minority female (i.e., partners identify with different non-White groups). We distinguish whether the non-White spouse or partner is male or female among non-White-White intermarried couples to explore whether gender of the non-White spouse or partner is linked to variation in recent fertility.

Our empirical strategy yields 910,001 couples that satisfy our selection criteria. Of these couples, 652,321 married in the past 5 years and 257,680 are cohabiting at the time of survey. Because our couple-level data are at the household level, we apply household weights to adjust percentages and means in the descriptive statistics. Tables A1 and A2 provide descriptive statistics on the sample. Nearly 15% of all couples are in interracial marriages, of which the overwhelming majority—nearly 90%—involve Whites. Gender asymmetries in interracial unions are especially large among Blacks and Asian Americans. Most Black-White unions involve Black men and White women whereas Asian American-White couples are selective of Asian American women and White men. These patterns contrast with Hispanic-White and American Indian-White pairings, which exhibit little gender asymmetry. As expected (Choi & Goldberg, 2020), the percent of interracial couples was lower among married than among cohabiting unions (14.2% vs. 17.2%).

Multivariate Analyses

We apply logistic regression models to predict odds ratios of whether married and cohabiting couples had a child in the past year. Unobserved heterogeneity may confound odds ratios (Mood, 2009), but does not affect average marginal effects (AME) when comparing coefficients across logistic regression models. We therefore also present AME—the average effects of variables on the probability of past-year childbearing across all observations (Mood, 2009). Our main independent variable is the racial pairing or mix of the couple. We introduce robust standard errors to correct for potential dependence of those living in the same Public Use Microdata Areas (PUMAs).

We control for a number of variables which may influence the relationship between racial pairing and past-year fertility. (Descriptive statistics are provided in Tables A1 and A2.) For example, compared with other couples, Hispanic women have children at comparatively young ages. We consider the following age groups: 15–19, 20–24, 25–29, 30–34, 35–39, and 40–50. Immigrants are more likely to have children than their native-born counterparts (Lichter et al., 2012; Parrado & Morgan, 2008), so we classify couples' nativity combination into both native-born; native-born male and foreign-born female; foreign-born male and native-born female; and both foreign-born. Educational attainment of both men and women influences interracial pairing as well as fertility (Musick et al., 2009; Qian & Lichter, 2011). Educational pairings of partners include: Neither with completed college; both with completed college; the female with some college or less and male with completed college; and the male with some college or less and female with completed college. Because fertility varies by union type (marriage or cohabitation) and marital order (first- or re-marriage) (Guzzo & Hayford, 2020), we classify couples into five distinct groups: (a) cohabiting, (b) both first married, (c) first married male and remarried female, (d) remarried male and first married female, and (e) both remarried. We also break down the 10 years of ACS data into two periods, 2008–2011 and 2012–2017, to capture the period effects of the Great Recession on fertility (Schneider, 2015).

Where racial and ethnic minorities live affects opportunities for interracial relationships and socioeconomic mobility. To explore how local conditions may influence couples' fertility behavior, we use several measures of racial composition and socioeconomic status. PUMAs contain at least 100,000 residents. A geographical area greater than 200,000 residents is usually divided into as many PUMAs of 100,000 or more residents as possible. PUMAs reflect local area residential conditions, which may be relevant to respondents' fertility behavior (Guzzo & Hayford, 2020; Su, 2019). We derive PUMA-level measures based on combined samples from several years, specifically pooled samples of 2008–2011 and 2012–2017. U.S. Census Bureau redraws PUMA boundaries every 10 years; PUMA boundaries for 2008–2011 were based on the 2000 census whereas boundaries used for 2012–2017 came from the 2010 census. For each PUMA, we calculate percent of racial and ethnic minorities (Blacks, American Indians, Asian Americans, and Hispanics combined), percent of multiracial individuals, percent of foreign born, percent of those with completed college education, and median household income. These measures are mean-centered and logged because the distributions across neighborhoods are highly skewed (an empirical fact reflective of racial and economic segregation across neighborhoods).

Fertility among Endogamous and Exogamous Couples

Our first objective is to ascertain whether exogamous couples have lower past-year fertility than endogamous couples. As a baseline, data in Table 1 show that racially endogamous married couples accounted for the overwhelming share—67.4%—of all births. The percentage rose to 85.4% when cohabiting couples were added. Nearly 9 out of 10 births were born to racially endogamous couples.

Overall, 13.8% of couples reported having a birth in the past 12 months (Table 1, bottom row). Consistent with our hypothesis, the percentages were substantially lower among cohabiting than married couples (9.8% vs. 15.6%). Moreover, past-year fertility was slightly lower, as expected, in exogamous unions than in endogamous unions, both for cohabiting and married couples. The highest fertility was experienced among same-race married couples (15.6%) and the lowest among different-race cohabiting couples (9.7%). Still, the difference between endogamous and exogamous couples was similar by union status (marriage or cohabitation). For married couples, fertility among exogamous couples was 3.2% lower than among endogamous couples. The differences are relatively small but confirm our hypothesis of lower fertility among exogamous couples.

The Racial and Gender Mix: Fertility Differentials

Fertility differentials between racially endogamous and exogamous couples are likely to obfuscate substantial heterogeneity across different types of interracial couples. In Table 2, we report past-year fertility among married and cohabiting couples, disaggregated by ethnoracial background of each male and female partner. We consider only interracial unions that involve partners who are White, Black, Asian American, Hispanic (of any race), and American Indian. The 5 × 5 cross-tabulation reported in Table 2 provides the percentages for each racial pairing—25 different combinations overall. The cells on the main diagonal highlight the percentage of racially endogamous couples with children born in the previous year. The off-diagonal cells show the percentage of interracial unions with children born in the past year. To illustrate, these data show that 3,894 Black-White married couples involved a Black female and White male, compared with 8,288 involving a Black male and White female. Almost 70% of Black-White marriages in the sample involved a Black male and White female. White male/American Indian female couples (3,840) were about equal to American Indian male/White female couples (3,845). As expected, these contrast with White-Asian American marriages, which are over-representative of White men and Asian American women.

The diagonal data in Table 2 reveal racial differentials in fertility among endogamous couples, both married (upper panel) and cohabiting (lower panel). The percentage of married couples having a birth over the past 12 months ranged from a low of 14.0% among Black couples to a high of 16.8% among Asian American and Hispanic couples. For each racially endogamous pairing, cohabiting couples were less likely than married couples to report a birth in the past year. Racial differences, however, were large. Among Whites, fertility among cohabiting couples was only 7.8%, only about half the percentage observed among married couples. Cohabitation-marriage fertility differences were much smaller for Blacks (12.6 vs. 14.0) and Hispanics (14.0 vs. 16.8).

These data also indicate substantial variation in fertility across different racial-gender pairings (Table 2, off-diagonal cells). Of these racial pairings, fertility was lowest among White husband/Asian American wife couples (12.7%), a figure lower than among White wife/Asian American husband couples (14.9%). Both pairings have lower fertility than endogamous Asian American couples (16.8%). In each case, non-White husband/White wife couples had a higher percentage of past-year fertility compared to non-White wife/White husband couples. In other words, White/non-White couples had a higher percentage of past-year fertility if the non-White spouse was male rather than female. This descriptive finding is consistent with our hypotheses.

The results among minority-minority couples are mixed. There is no evidence that minority-minority couples had lower past-year fertility. In fact, a larger percentage of Black-Other minority and American Indian-Other minority married couples had a child in the past year than did their racially endogamous counterparts. Most Asian American-Other minority and Hispanic-Other minority couples had lower percentages than their respective endogamous couples, but mostly higher than their peers who married Whites. Fertility rates were lower for each type of racial pairing involving cohabiting couples, but much lower among couples involving Whites or Asian Americans. For example, past-year fertility rates were 12.7% and 14.9%, respectively, among White husband/Asian American wife couples and Asian American husband/White wife couples. They were only 3.7% and 4.4%, respectively, among their cohabiting counterparts. In summary, these descriptive results suggest that fertility is suppressed in White male/non-White female marriages whereas intermarried Blacks have higher levels of fertility than their endogamous counterparts.

Multivariate Results of Fertility

Fertility among all couples

Racially endogamous and exogamous unions are likely to be very different in their sociodemographic makeup across racial pairings. Specifically, the bivariate findings reported in Table 2 may be the result of many different confounding factors rather than of factors (e.g., stigma or gender power relations) inherent to each racial pairing. To address this issue, Table 3 includes results from a series of logistic regression models predicting the odds of having a child in the past year. We start by asking whether exogamous couples are less likely than endogamous couples to have had children in the past year. Model 1 confirms that the odds of having a child in the past year are 7% lower compared to endogamous couples, consistent with estimates in Table 1.

Table 3. Results from Logistic Regression Predicting Odds of Having a Child in the Past Year among Married and Cohabiting Women, 2008–2017

	Model 1	Model 2	Model 3	Model 4
Racially endogamous
Endogamous (Reference)
Exogamous	0.93***
Racial/ethnic pairing
White couples (Reference)
Black couples		0.97*	1.21***	1.32***
American Indian couples		1.17***	1.56***	1.57***
Asian American couples		1.20***	0.98	1.02
Hispanic couples		1.22***	1.22***	1.28***
White man Black woman couples		0.87***	0.93	0.98
Black man White woman couples		1.06*	1.21***	1.27***
White man American Indian woman couples		0.97	1.04	1.06
American Indian man White woman couples		1.00	1.05	1.07
White man Asian American woman couples		0.75***	0.81***	0.84***
Asian American man White woman couples		0.89***	0.89****	0.93*
White man Hispanic woman couples		0.91***	0.93***	0.97
Hispanic man White woman couples		1.10***	1.10***	1.15***
Minority-minority couples		1.11***	1.21***	1.29***
Female partners' age group
15–19 (Reference)
20–24			0.75***	0.75***
25–29			0.63***	0.63***
30–34			0.68***	0.70***
35–39			0.50***	0.52***
40–50			0.10***	0.10***
Couples' nativity
Both native-born (Reference)
Native-born man and foreign-born woman			0.94***	0.97*
Foreign-born man and native-born woman			0.99	1.02
Both foreign-born			1.15***	1.20***
Couples' educational attainment
Both no completed college (Reference)
Woman with some college or less and man with completed college			0.85***	0.88***
Man with some college or less and woman with completed college			0.81***	0.83***
Both completed college			0.74***	0.81***
Cohabitation or marital order
Cohabiting (Reference)
Both first married			2.30***	2.26***
First married man and remarried woman			2.44***	2.39***
Remarried man and first married woman			2.21***	2.16***
Both remarried			1.43***	1.39***
Number of older children living in the household			0.87***	0.86***
Time period
2008–2011 (Reference)
2012–2017			0.94***	0.93***
PUMA level characteristics
ln (percent minorities) mean centered				0.96***
ln (percent multiracials) mean centered				0.99
ln (percent foreign born) mean centered				0.98**
ln (median household income) mean centered				1.47***
ln (percent completed college) mean centered				0.70***
Constant	0.16***	0.16***	0.20***	0.19***
Log-Likelihood Ratio	64	949	30,645	31,930
N	910,001

• Note. *p < .05, **p < .01, and ***p < .001.

As a next step, Model 2 distinguishes patterns of fertility among five types of endogamous couples by race/ethnicity and nine types of exogamous couples (eight White-non-White pairings plus one minority-minority pairing). Minority-minority couples are treated as a single category, recognizing that sample sizes for some minority combinations are sparse and that the large majority of exogamous unions involve White partners. These analyses indicate that the odds of having a birth in the past year were 17%, 20%, and 22% higher among endogamous American Indian, Asian American, and Hispanic couples, respectively, than the reference group (i.e., White-White couples). The odds of past-year fertility are significantly lower—but only 3% lower—among endogamous Black couples than among endogamous White couples. The low fertility rates reported earlier (Table 2) among married endogamous Black couples, along with higher rates among their cohabiting counterparts, seem to bring overall fertility among all Black and all White co-residential unions into general alignment.

Patterns of fertility by racial pairing are considerably more idiosyncratic, but support our hypotheses that the gender and race of non-White partners matter. On the one hand, non-White women who formed unions with White men, for example, generally had lower odds of past-year fertility than did endogamous White couples. Among non-White women with White partners (married or cohabiting), the odds of past-year fertility were 13% lower for Blacks, 25% lower for Asian Americans, and 9% lower for Hispanics than for their White-White counterparts. On the other hand, among non-White men with White partners, the odds of past-year fertility were 6% higher for Blacks, 11% lower for Asian Americans, and 10% higher for Hispanics. The odds of past-year fertility among American Indian-White couples, regardless of race/gender mix of partners, were not statistically different from endogamous White couples.

In Model 3 (Table 3), we control for social and demographic characteristics, including female age group, couples' pairing by nativity, education, union status and marital order, and time period. These results show that endogamous minority couples had greater odds of past-year fertility than endogamous White couples. The odds of past-year fertility also increased with controls among Black endogamous couples (from .97 to 1.21) and American Indian couples (from 1.17 to 1.56). The disadvantaged demographic and social profiles of Blacks and American Indians clearly have the effect of suppressing observed fertility differences from White-White couples. For Asian American couples, the odds of past-year fertility were not significantly different from the reference group (White-White couples), once other variables were included in the models.

The inclusion of PUMA characteristics in Model 4 (Table 3) generally reveals little overall effect on our substantive conclusions regarding fertility. Perhaps the most notable changes were found among Black endogamous couples and Black-White couples. With PUMA-level controls, the odds of past-year fertility among Black-Black couples increased from 1.21 to 1.32 between Models 3 and 4. Moreover, the odds of fertility among Black male/White female couples were 21% greater than among White-White couples (Model 3), rising to 27% greater when PUMA controls are introduced in Model 4. In contrast, the odds of fertility among White male/Black female couples were statistically similar to those of endogamous White couples, both in Models 3 and 4. The results confirm our central findings: Endogamous minority couples or interracially married couples are as likely—and sometimes more likely—to have births in the past year than endogamous White couples.

Changes in odds ratio across models may be confounded by changes in unobserved heterogeneity unrelated to the variables in logistic regression models whereas changes in average marginal effects (AMEs), the average of predicted probabilities of all individuals with observed values of the variables in these models, would be unaffected (Mood, 2009). Table A3 replicates Table 3 with results of AMEs. The substantive results remain the same. Figure 1 presents average marginal effects of racial/ethnic pairing on fertility in the past year (including confidence intervals) based on Model 4 in Table A3. These results clearly reveal the larger marginal effects among Black (.034), American Indian (.058), and Hispanic (.029) endogamous couples vis-à-vis White endogamous couples. They also highlight significantly higher past-year fertility among Black-White (.028) and Hispanic-White (.017) couples, closer to the AMEs of their Black and Hispanic endogamous counterparts, but only when Black or Hispanic partners are male. In contrast, AMEs of American Indian-White couples and White endogamous couples were not statistically different. For Asian American-White couples, AMEs were one or two percentage points lower than those of either Asian American or White endogamous couples. These results confirm the bivariate patterns observed in Table 2 and the results in Table 3.