Jacobson, Cardell K


Inter-marriage between individuals of different racial and ethnic groups long has been used as an indicator of assimilation and acceptance of different racial and ethnic groups in society. Persistent low rates of out-marriage, or marriages across racial lines, suggest that assimilation and acceptance are far from complete, however. In this paper we examine inter-group marriage in one arena, the United States military services, where many of the traditional barriers to heterogamous marriages are significantly reduced. The weakened barriers we discuss include demographic patterns (segregation and age differences) socioeconomic factors (education), societal norms, and homogeneity of local marriage markets. We compare patterns of interracial marriage in the military to national rates, and we examine the effects of military rank, type of service, and periods of military service on inter-group marriage.

A basic proposition on group size, incorporated by Blau and others, provides a general theoretical perspective for inter-group marriage. Blau and others (see Blau, 1977; Blau, Blum, & Schwartz, 1982; Blau, Beeker, & Fitzpatrick, 1984; Blum, 1984; Sampson, 1984; Alba & Golden, 1986; Hwang et al., 1994) argue that relative group size is critical in determining the amount of outgroup contact that individuals are likely to experience. This increased contact can lead to romantic interest and increased rates of out-marriage. This is particularly true for small groups. Since group size is inversely related to the chances that individuals will marry out of their own group, a setting where traditional barriers are low is likely to have higher than normal rates of intermarriage.

Despite its multiplicity of races and ethnic groups, the United States exhibits high levels of segregation (Massey and Denton; 1993). The segregation combined with normative sanctions reduces the chances of outgroup marriage. But these factors are reduced significantly in one institution in American society, the United States military services. White non-Hispanic males, for example, constitute 85 percent of married men in the 1990 census. But they constitute only 76 percent of married men in the military. Prior research has shown that ethnically heterogeneous regions have high rates of interracial marriage (Hwang, et al., 1994; Jacobson & Heaton, 1996). Thus the increased heterogeneity in the military means that individuals are more likely to meet people from and marry outside their own group.

Cultural norms and prejudice are also part of the nexus of segregation. While group norms usually emphasize marriage within one’s own group, these forces are mitigated in the military because of its emphasis on non-prejudicial, non-discriminatory treatment of other groups. Because the military services promote an attitudinal environment that is often at variance with the traditional group and cultural norms, acceptance of other groups is higher in the military services than in society as a whole. For example Moskos and Butler (1996;2) report:

A visitor to an Army dining facility…is likely to see…blacks and whites commingling and socializing by choice. [They] also patronize equally such nonduty facilities as barber shops, post exchanges, libraries, movie theaters, and snack bars. And in the course of their military duties, blacks and whites work together with little display of racial animosity.

In sum, military personnel have contact with other groups and cultures both in the United States and abroad, under favorable conditions, that may reduce resistance to outgroup marriage.

Further, socioeconomic factors often divide minority groups from the majority population. The role of socioeconomic status in inter-group marriage long has been recognized by exchange theorists who focus on the attributes potential spouses bring to marriages. Socioeconomic statuses, along with race or ethnicity, and physical attractiveness, are commodities that are potential exchanges. Members of minority groups are said to exchange high socioeconomic advantage to overcome the socially perceived disadvantage of minority status (see Merton, 1941; Schoen & Wooldredge, 1989; Kalmijn, 1993; Heaton & Albrecht, 1996; Qian, 1997). While substantial research demonstrates that members of disadvantaged minorities who marry into majority groups have above average socioeconomic status (e.g., Fu & Heaton, 1997; Heaton & Jacobson, 2000), marriage is most likely to occur if overall attributes are perceived to be similar. Thus, where are socioeconomic status of groups is equal, inter-group marriage is more likely to occur than where the groups are unequal.

The military tends to de-emphasize socioeconomic differences because of the bureaucratized procedures for recruitment and promotion. For example, 1990 census data show less educational difference among groups who have served in the military than is the case for the population as a whole. In the general population, the average education of married white non-Hispanic males exceeds that of African-Americans males by 18 percent, and that of Hispanics by 34 percent. In contrast, white military men have only a 6 percent educational advantage over African-American males and an 8 percent advantage over Hispanics. Among married white women in the general population, the educational advantage over African-American women is 6 percent and 26 percent over Hispanics. The comparable percentages for women in the military are 1 percent and 2 percent, respectively. This substantial reduction in socioeconomic differences in the military likely increases the probability of marriage outside of one’s own group. Thus, even though the United States military services are highly structured and hierarchical, they also provide an environment that is characterized by heterogeneity and egalitarianism within ranks, a substantially different environment than exists in the larger society.

Therefore, our main hypothesis is that (1) individuals who have participated in military service will be more likely than others to marry outside of their own racial or ethnic group. This is true whether the marriage partner is also a member of the military services or, as is much more common, from outside the military. We further hypothesize that (2) type of military service will make a difference. Those who have served in the regular military, compared to those who served only in the national guard, are more likely to have experienced the factors discussed above. They have served full time and have served in more integrated settings with normative support for equal treatment than have those in the national guard. Furthermore, the longer individuals have served, the more likely they are to have been exposed to and to have adopted the egalitarian norms. We thus expect, all other things being equal, that (3) those with longer military service or higher military rank will be more likely to have married out of their group than will those with lesser service. Finally, since the stigma of marrying interracially has decreased in recent decades, we expect that (4) the time period will be related to the tendency to marry out of one’s own group. We expect higher rates of out marriage to appear in more recent cohorts. We examine veterans from four different war service periods and compare them to non-veterans. The four periods are Word War I and II combined, the Korean period, the Vietnam period, and others who served most recently.

We expect these differences to persist even when a variety of other variables have been controlled statistically. Potentially confounding variables are education, age, and year of the survey. Justification for these controls is provided below.

Education has been linked to higher rates of heterogamy (Kalmijn, 1991, 1993). Like the military, colleges provide norms that are more accepting and tolerant of inter-racial dating than in the larger society. Colleges also provide increased opportunity for inter-group contact, dating, and marriage but education can have counter effects as well. Students generally tend to marry homogeneously with regard to social class (though they sometimes date hetcrogeneously). Furthermore, as noted earlier, marriages often involve exchange relationships. Since education influences these processes, we include it as a control variable.

We also control for age. Older people generally married when interracial dating was tolerated less than it is today. Since we examine trend data, we use age as a control variable. We also control for year of survey. We examine data for four different groups: African-Americans, non-Hispanic whites, Hispanics, and others (who are primarily Asians).

We acknowledge that race and ethnicity are social constructs, and intermarriage within the broad groups we have defined here, especially Asians and Hispanics, could well constitute heterogamy as much as marriage between whites and the groups. Nevertheless, these are well-recognized census categories, and we define the groups broadly so that we have sufficient data for the analysis.


Data for this analysis are taken from two sources. The first is the Public Use Microdata Sample, 1990 (PUMS) which has good data on type of military service and rank. All married persons with spouses present were selected from the one percent sample, and husbands and wives were matched. The resulting file is over one million married couples, so we sampled 10 percent of the most common type of marriage – those between white non-Hispanic husbands and wives.

The second data source, the Current Population Surveys – March (CPS), has good data on trends and periods of military service. For the analysis presented in this paper, we combined the surveys for the years 1977 through 1997, the years for which Hispanic ancestry is available. Native Americans and other small ethnic groups are excluded from the analysis because of insufficient cases. Husbands and wives were matched in this sample as well.

Both the census and CPS data present limitations. One, as noted above, is that considerable diversity exists within the broad “groups” we examine here. The census category for Hispanics contains individuals from a variety of countries, as does “Asian.” This limitation extends to almost all studies of heterogamous marriage, however.

A second limitation is that the census data do not include couples who were married and divorced before the census was taken, and our analysis focuses only on married couples, not on those living together.

A related issue is that census data do not have date of marriage. Some studies of interracial marriage have relied on marriage certificate data (see Kalmijn, 1993) because the certificate data include year of marriage. The census data are more useful for our purposes, however, because they are available from all states, and they have the critical military and socioeconomic status variables.

Another limitation is that interracial marriages may have a higher or lower divorce rate than same race marriages. If this is the case, then the census data may not accurately estimate the amount of interracial marriage. But the census and CPS data have information on prior and present military service, have complete geographic coverage, and include a broad set of other variables. And the CPS data allow us to examine trends in inter-group marriages.

The CPS data actually underestimate the frequency of inter-group marriage in the military because they include military couples only when one member is in the military and the other is not. The survey does not include those individuals living in group quarters or in the military services.

Finally, the data in this paper do not include marital status at the time of entrance into the military services. Most military personnel, however, are unmarried when they enter the services. Segal and Harris (1993) report that less than 3 percent of those under the age of 20 are married. This is also consistent with data from the Military Family Resource Center that shows 39.3 percent of those on active duty to be single. Even those who marry after they leave the service have exposure to the normative egalitarian prescriptions of the military, and these are likely to influence subsequent mate selection.

We use rank as a rough equivalent of time in the military. One reviewer noted that this is generally true within the officer corps and within the enlisted personnel. It is not true, however, for comparisons between ranks.

In the analysis which follows we first examine the frequency and trends for inter-group marriage. We then examine the frequency of these marriages within the military, and their frequency by military rank and period. Finally, we present the results from a multinomial logistic regression that examines the effects of period, age, education, on inter-group marriage while controlling for year.


The frequencies and percentages of homogamous and heterogamous marriage from the 1990 census PUMS are reported in Table 1. While each group has a high rate of homogamous marriage, white and African-Americans are more likely than Asians and Hispanics to marry within their own groups. These results are consistent with results found by Blau and others (Blau 1977; Blau et al., 1982, Blau, et al., 1984; Jacobson & Heaton, 1996; Heaton & Jacobson, 2000). The higher outmarriage rates for Asians and Hispanics, compared to blacks and whites, are particularly interesting. These groups are obviously smaller, and we thus expect higher rates of outmarriage. However, these groups have also had high rates of recent immigration, and for several reasons recent immigrants are probably less likely than others to have married outside their own group. First, they usually move to ethnic communities where they are relatively isolated and less likely than resident minorities to have contact with and marry outside their own group. Some immigrants also marry before immigrating. Some immigrants generally have relatively low educational levels and thus are less likely to have married outside their own group (see Heaton & Jacobson, 2000). Most have few English-language skills. At the same time, some immigrants move to the United States after marrying American soldiers stationed abroad.

We expect more marriages in the white husband-white wife category simply because whites are a large share of the population (as is evident in Table 1). The results presented in Table 1 show more marriages of whites to Asians and Hispanics than to African-Americans. The total outmarriages are still a relatively small percentage compared to white-white marriages. As previous census reports have shown, the number of marriages between white women and African-American men is higher than between African-American women and white men, and Asian women are much more likely than Asian men to be married to either whites or African-Americans. This imbalance likely results, at least in part, from black and white American serviceman marrying Asian women while stationed abroad. Little gender difference in marriage pattern is observed in marriages between whites and Hispanics or Asians and Hispanics. These patterns are consistent with results found by a number of other researchers

The percentage of whites, African-Americans, Asians, and Hispanics who have married out of their own group by military service record is shown in Table 2. Those who have either prior or current military service are substantially more likely to have married heterogamously than those without military service. For wives, this is true for all four groups and for type of military service. Likewise, Asian and Hispanic husbands who have had or currently have experience in the military services are more likely than others to have married out of their own group. This is also true for African-American and white husbands who have served in the past or are currently serving in the military. In general, the largest differences occur for those who are currently serving in the military; they are much more likely than those without prior military service to have married heterogamously. For the smaller Asian and Hispanic groups, the differences by military service are substantial. These differences by military experience persist when a multivariate analysis (not reported here; see Heaton & Jacobson, 2000) was performed on the PUMS data with immigration status, metropolitan residence, age, a measure of socio-economic status, and region included in the analysis.

In sum, the PUMS data support our first hypothesis. Those who have prior or current military service are more likely than others to outmarry. Figure 1 from the PUMS data also shows support for our second hypothesis (type of military service) and shows at least partial support for our third hypotheses that rank will be related to exogamy. Figure 1 shows that those who serve in the regular military are more likely to marry out of their own group than are those in the guard or those who have not had military service.

Current military service as a non-commissioned officer shows the highest outmarriage for all groups but Asians. For Asians, the highest percentage married heterogamously are the commissioned officers. White commissioned officers are much more likely to be married exogamously than those in the reserves or those who have had past military service. But the rates of outmarriage for African-American husbands in the National Guard or reserves is the same as those who have not served in the military. The rate for whites serving in the National Guard or reserves is lower than for commissioned officers, however.

The CPS data show similar trends and results. The total percentage of these four groups who have married heterogamously is shown in Figure 2. This percentage has increased from 3.5 percent in 1977 to 6.5 percent in 1997. The trends for heterogamous marriage broken out by group are shown in Figure 3. We have several observations about Figure 3. First, as noted earlier, these classifications are very broad and inter-marriage between some groups within the broader communities could well be considered exogamy. Thus, the limitations of the data set actually underestimate the number of heterogamous marriages. second, “others,” as identified in the CPS, are primarily Asians though some other individuals are also included (e.g., Aleuts, Indians).

Third, as also shown in Figure 1, the percentage of Hispanics and “others” (primarily Asians) who married heterogamously is much higher than for African-Americans and whites. These results are consistent with Blau’s basic proposition that group size is inversely related to heterogamy. The higher percentage of African-Americans than whites having married heterogamously is also consistent with Blau’s proposition. Given that Hispanics are roughly 11 percent of the population, African-Americans between 12 and 13 percent of the population, and “others” about 9 percent of the population, we would expect the heterogamy rate for African-Americans to be much closer to the rates for Hispanics and “others” than it is. Clearly, factors other than group size are operating to affect the marriage markets. These other factors include social distance, prejudice, cultural norms, and location.

Fourth, Figure 3 shows a gradual increase for heterogamous marriages for African-American and white Americans, while the rates are relatively flat (except for sampling variations) for Hispanics and “others.” As noted earlier, the relatively stable rates for the latter two groups are probably largely a function of continued immigration by these groups. Immigrants are less likely than those born in the United States to marry heterogamously. We should also note that the percentage of those married heterogamously in the CPS data is slightly below that reported in the census data.

The CPS data also show that the period of military service is related to heterogamous marriage. Figure 4 charts the proportion of heterogamous marriages by whether the individuals had military service and by period of military service. The data presented in Figure 4 are without statistical controls for age and education level; these are presented later in Table 3. Since fewer women than men have been members of the armed forces, only data for males are presented here.

Non-veterans in Figure 4 show the gradual increase in heterogamy, from 1977 to 1997, that we have noted before. The rates of heterogamy by period and the changes over time vary substantially, however. Those veterans from the World War periods show no change in heterogamy while the Korean period veterans show a slight increase over the twenty-year period. The two groups with most recent service show the highest rates and the highest increases in heterogamy over the twenty-year period. These are the Vietnam period veterans and those who have been in the service during non-war years (primarily since the Vietnam War). The percentage of the Vietnam-perod veterans who have married out of their own group increased from slightly over 4 percent in 1977 to approximately 6 percent in 1997. Likewise, the percentage of those married heterogamously in other service periods increased from 2.9 percent in 1977 to over 7 percent in 1997. These last two groups are also the youngest groups, and as might be expected, are more likely than older groups to be married heterogamously.

More detailed information about whom these respondents married is presented in Table 3. Again, only data for males are presented here because few women appear in our sample once it is separated by race and ethnicity. Overall, the patterns are similar to those reported earlier. Whites tend to marry homogamously while African-Americans are somewhat more likely than whites to marry outside their group. Both Hispanics and “others” are much more likely to marry heterogamously than are whites and African-Americans. The Vietnam period veterans tend to marry outside more than Hispanic males.2 These results are consistent with those presented in Figure 4. But, again, these results do not include a multivariate analysis to which we now turn.


Table 4 reports the results from multinomial logistic regressions using the CPS data. These data show the association between period of service and interracial marriage, controlling for age, education, and year of the survey. The control variables generally have the expected effects. The increasing trend toward interracial marriage is evident in the negative coefficients for age and the positive coefficients for year. Education facilitates interracial marriage for Hispanic men and, to a lesser degree, for African-American men. The education coefficients are quite modest for white men, however, and negative for the “other” category. These modest effects likely result from several factors. As discussed earlier, education has countervailing effects; while educational institutions provide increased opportunities for contact and marriage, students and others often practice assortative mating. Furthermore, educational institutions are selective in whom they admit so that the effects of education on inter-group marriage is limited.

The first hypothesis that military service increases the chances of outgroup marriage is supported in the multivariate analysis. Likewise, the general pattern of positive coefficients for each period of military service supports Hypothesis 2 that interracial marriage is more likely to occur among veterans than non-veterans (the implicit category) once the control variables of age, education, and year are included in the analysis. Military service does not have a dramatic impact on marriages between white men and African-American women, but it does increase the chances that an African-American man will marry a white woman.

The pattern of the coefficients indicates that particular periods of service are especially important for some groups (Hypothesis 4). As might be expected, serving in the Vietnam period had a dramatic effect on marriages between white men and the “other” (primarily Asian) women. The effects of service in non-war years and the Korean war years on marriage to the other groups are also notable, though the effects are not as dramatic for the Vietnam period. Military service also enhances the marriage rate between white men and Hispanic women, but no particular period stands out as more important than another.

With one unusual exception, African-American men who serve in the military are more likely to marry white and “other” women, and this pattern is evident in each military period. The exception occurs for marriage between African-American men and “other” women during the Korean conflict. The effects of military service on African-American/Hispanic marriages are more modest than for African-American/white marriages.

The rates of heterogamous marriages of Asian and Hispanic men are the most affected by military service. With the exception of marriage between “other” men and African-American women, each of the coefficients is large and most are statistically significant. Most notable are the marriages between “other” men and white women, and between Hispanic men and “other” women. These effects are also evident in each period of service examined.

Data about type of service (Hypothesis 2) and rank (Hypothesis 3) are available in the PUMS. These results are reported in Table 5. Results for type of service are generally consistent with our hypothesis. Those on active duty are most likely to be in an heterogamous marriage. All but one of the coefficients for active duty are positive and most are larger than coefficients for other types of service. Coefficients for prior service are generally positive, but not as large as those for active duty.

In contrast, several of the coefficients for participation in the national guard or the reserves are negative. This is especially true for white and African-American husbands (though none of the coefficients for African-American reserves are statistically significant). Asians and Hispanics who have been in the reserves, on the other hand, are more likely to marry interracially. This difference between whites and African-Americans on one hand, and Asians and Hispanics on the other hand, may reflect recent immigration trends in these communities. Recent immigrants in the Asian and Hispanic communities are unlikely to join the reserves. Members of these communities who do join the reserves are likely more mainstream in society. Thus, being in the reserves is positively associated with outgroup marriage for these two groups.

Results for rank are not consistent with the hypothesis. Commissioned officers are generally less likely to marry interracially than others, but the coefficients are generally not statistically significant. Commissioned officers may have joined and been promoted in the military services in earlier periods when exogamy was less likely and before minorities were as highly represented in the services. But the non-significant results presented here are after age and education have been controlled statistically. As noted above, the consistently negative coefficients for age reflect the increase in inter-group marriage over time. And the coefficients for education are positive for all groups except Asians.


Interracial and inter-ethnic marriages, though still unusual in American society, are becoming more common and accepted. The CPS data indicate that the rate of exogamy has been increasing steadily over the past couple decades. Our data indicate that participation in the military services is associated with increased exogamy, even when age and education are controlled statistically.

Given the complexity of the mate selection process which is compounded by long standing norms against interracial marriage, it should not be surprising that our results are not uniform. Opportunity factors interact with socioeconomic status, gendered roles, and cultural tendencies toward endogamy in their influence on marriage patterns.

Marriage markets are local, not national or even regional. It is within these local markets that other factors facilitate or impede inter-group contact and intermarriage. The military services provide a local market that is relatively isolated from the cultural norms of society, and it creates opportunities for increased contact and interaction between members of different racial and ethnic groups. They do this in several ways.

First, the military services create contact between groups in foreign countries. Military men and women serving in other countries obviously have increased opportunities to meet, fall in love, and marry individuals from the host countries. Though military personnel are a numerical minority and though they may keep to themselves on bases or local areas, they still have opportunities to meet, date, and marry members of the host nation.

A second way the military services create opportunities for exogamous marriages is that members of the military have contact with members of other racial and ethnic groups in their own units or on the base. The military services provide contact under conditions where it is acceptable for members of all groups to meet and associate.

Finally, the military services tend to create esprit de corps so that members of all racial and ethnic groups consider themselves to be one together. All of these factors increase the chances for exogamy. Unfortunately, our data do not allow us to examine separately exogamous marriages when both are from the military and compare them to marriages where only one member of the couple is from the military services.

As would be expected, substantial endogamy remains, with the highest rates among whites and African-Americans. Nevertheless, our overall results show that military service has a consistent pattern of effects; military service increases the likelihood of outgroup marriage for all groups. Generally, those with prior service and those currently serving in the military services are more likely than others to be married exogamously. Type of military service and period of service are important. Those who served during the Vietnam period in particular were more likely to have married out of their own group. Only our hypothesis that higher military rank would be associated with exogamous marriage was not supported once education and age were controlled statistically. Conventional norms seem to have prevailed for those of higher military rank.

Overall, however, the military presents an ethnically heterogenous environment with reduced socioeconomic differences and with reduced prejudice and prescriptive cultural norms. Under these conditions, exogamy increases. Nevertheless, the results continue to reflect the complexity of the interracial and interethnic marriage markets in American society today.

Copyright Dr. George Kourvetaris Summer 2003

Provided by ProQuest Information and Learning Company. All rights Reserved.