<<

Black-White Unions: West Indians and African Compared Author(s): Suzanne Model and Gene Fisher Source: Demography, Vol. 38, No. 2 (May, 2001), pp. 177-185 Published by: Population Association of America Stable URL: http://www.jstor.org/stable/3088299 Accessed: 21/12/2009 20:33

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=paa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Population Association of America is collaborating with JSTOR to digitize, preserve and extend access to Demography.

http://www.jstor.org BLACK-WHITEUNIONS: WEST INDIANSAND AFRICANAMERICANS COMPARED*

SUZANNEMODEL AND GENEFISHER

In this research we use 1990 PUMS data to compare the pro- trols are introduced, unions between West Indians and na- pensity for unions between and native whites tive whites generally occur at higher rates than unions be- with the propensity for unions between British WestIndians and na- tween African Americans and native whites. After the in- tive whites. In we women and men. addition, distinguish Descriptive troduction of controls, West Indian men of any generation statistics indicate that WestIndians, with the men who exception of have lower exogamy rates than African Americans, while arrived as adults, are more likely than African Americans to have white the introduction controls several corre- exogamy rates are higher among West Indian women who partners. After of for arrived as children or who were born in the lates of intermarriage, however, WestIndian men of any generation have lower exogamy rates than African American men, while ex- than among African American women. We find no differ- ogamy rates are higher among WestIndian women who arrived as ence, however, in the exogamy rates of African American children or who were born in the United States than among African women and West Indian women who arrived as adults or American women. Thus we find no consistent evidence of greater teens. In addition, for both West Indian men and women, exogamyfor British WestIndians thanfor African Americans. exogamy rates tend to increase with generation. We inter- pret these findings as consonant with "straight line" as- similation theory rather than with the idea that relations T he arrival of record numbers of Asians, Latinos, and Afri- between whites and West Indians are more cordial than be- cans during the second half of the twentieth century has in- tween whites and African Americans. fused new life into scholarship on intermarriage.Research- ers already have discovered high rates of intermarriagebe- THEORETICAL EXPECTATIONS tween whites and several native-born Asian American and Of the factors known to influence interracial relationships, groups (Farley 1996; Gilbertson, Fitzpatrick, and three have received special attention: opportunity, Yang 1996; Qian 1997), but intermarriage between whites hypergamy, and attitudes (Blau, Blum, and Schwartz 1982; and persons of African heritage appears to be rare. In 1994 Lieberson and Waters 1988; Merton 1941). By opportunity, only 7% of all couples containing a black partnerincluded a scholars mean the likelihood that women and men of differ- white spouse (U.S. Bureau of the Census 1995). Yet, because ent backgroundswill become acquainted. Opportunityis pri- most U.S. blacks are African Americans, aggregate figures marily a structural phenomenon, governed by conditions say little about the marital patterns of black immigrants and such as sex ratios, group size, and residential segregation. their offspring. According to some observers, relations be- Hypergamy refers to the practice whereby women marrymen tween whites and West Indians, especially British West Indi- of higher social status than themselves. A disproportionate ans, are more cordial than relations between whites and Af- number of black-white marriages involve pairings between rican Americans (Arnold 1996; Foner 1985; Waters 1999). low-status white women and high-status black men (Schoen This possibility motivates us to ask, what proportion of Brit- 1995). This may contribute to another trend: black men ish West Indians have white partners?How does that propor- marry whites more often than do (Kalmijn tion compare with the proportion of African Americans with 1993, 1998). Finally, the attitudes relevant to interracialpair- white partners?And what happens to these proportions when ings include how strongly members of a particular racial or controls are introduced for group size, education, and other oppose exogamy and how they rank members determinantsof interracial unions? of different out-groups as mates. Attitudes toward intermar- With these questions in mind, we examine partnerships riage are influenced by demographic variables such as nativ- among whites, British West Indians, and African Ameri- ity, age, and education, but attitudes also vary among ethnic cans. The data come from the 1990 U.S. Census and in- and racial groups for cultural and historical reasons. clude both married and cohabiting individuals. Before con- In developing our hypotheses, we emphasize two deter- minants of attitudes: group membership and generation. This is not to deny the effects, on attitudes, of demographic 'Suzanne Model and Gene Fisher, Department of Sociology, Thomp- variables such as age and education. Yet, because the effects son Hall, 200 Hicks Way, University of , Amherst, MA 01003- of these variables are understood more clearly, we treat 9277; E-mail: [email protected]. We thank Douglas Anderton and them as covariates We focus here on the A version of this (Kalmijn 1998). Greta Gilbertson for comments and suggestions. previous choices of members of three native-born paper was presented at the 1998 annual meetings of the American Socio- partnership groups: logical Association, held in . non-Hispanic whites, African Americans, and British West

Demography, Volume 38-Number 2, May 2001: 177-185 177 178 DEMOGRAPHY,VOLUME 38-NUMBER 2, MAY2001

Indians.1 For the British West Indians, we construct four One way to reconcile these contradictory findings is to generational categories: the foreign-born who arrived as suppose that the whites who ranked "Negroes" more favor- adults, the foreign-born who arrived as teens, the foreign- ably than West Indian blacks in the GSS had little if any con- born who arrived in childhood, and the native-born tact with West Indians. West Indians are a relatively small (Oropesa and Landale 1997). We introduce these genera- and geographically segregated group; the majority of Ameri- tional categories for two reasons. First, we introduce them can whites probably have never encountered any. Moreover, because the U.S. Census provides no information on date of some West Indians have admitted to ethnographersthat they marriage. As a result of this omission, we cannot distinguish take steps, when meeting whites, to communicate their heri- between foreign-born persons in unions initiated before ar- tage and the meaning of that heritage. Another respondent rival from those in unions initiated after arrival. Also, the told Faye Arnold, "I always go to school the first week. I longer single immigrants live in the United States, the more volunteer. I join the PTA, the Parents Council, you name it. I likely they are to find a partner of another background let those teachers know...that we come from Trinidad and (Hwang and Saenz 1990); one objective of our generational Tobago so our culture is different. We expect our kids to categories is to take this fact into account. Second, we ex- be...students and will do what is needed to see that they do" pect that the attitudes associated with West Indian partner- (1996:25). If this scenario is common, then whites who have ships are a nonlinear function of generation. contact with West Indians may view the Islanders more fa- Before justifying the expectation of nonlinearity, we de- vorably than they perceive African Americans. velop our key hypothesis: native white-West Indian pairs will With respect to the black side of the equation, no study be more common, all else being equal, than native white- has systematically addressed whether West Indians and Afri- African American pairs. To support this expectation, we take can Americans differ in their attitudes toward whites. In into account the attitudes of whites and both groups of blacks. Hildebrandt's (1999) interviews with intermarried blacks, In this regard, two data sources are helpful: the responses to a however, African American respondents reported hearing special module of the General Social Survey (GSS) and the stronger objections from friends and family than did Carib- responses to ethnographers' interviews with West Indian bean respondents. Mary Waters also speculates that West In- Americans. In 1990 the GSS inquired about feelings toward dians view whites with less rancor.A Trinidadianteacher ex- several new immigrant groups, including West Indian blacks plains, "I'm not American, and I do not see myself as having (Smith 1991). Respondents were asked to rank groups in been deprived by the whites of America. To the contrary, I terms of their social standing, on a scale ranging from 1 (low) came here, I was accepted, I was acknowledged for what I to 9 (high). West Indian blacks received an average standing knew and I am in a position now where I am earning a good of 3.56, a score lower than for "Negroes" (4.17) and nearly salary" (Waters 1999:144). In sum, there is no reason to ex- identical to that for African blacks (3.58). The term Negroes pect that West Indians will view the taking of a white partner was used in order to maximize comparability with older ver- less favorably than do African Americans, and some reason sions of the survey. The terms African American and/or Brit- to expect that they will view it more favorably. ish WestIndian probably would have yielded higher scores. With regard to the effect of generation, "straight line" Still, if these figures mirrorperceptions about intermarriage, assimilation theory predicts that the foreign-born are less ex- then whites feel more positively about unions with African ogamous than the native-born, and that the younger the age Americans than with West Indians. at which an immigrant arrives, the more predisposed to in- Yet, a large ethnographic argues otherwise termarriage he or she will be (Alba and Nee 1997). In the (Arnold 1996; Foner 1985; Gordon 1979; Waters 1996, case of West Indians and whites, however, we suspect that 1999). Most British West Indians maintain that they receive the "straight line" model will not hold. Rather, we believe better treatment from whites than do African Americans. A that the intermarriagerate will be highest among West Indi- Belizean teacher in told Faye Arnold, "West In- ans who arrived as teens, followed (in order) by those who dians are treated the best....[T]o be truthful-I have spoken arrived in childhood, the native-born, and West Indians who to a few people and they told me that they had dealings with arrived as adults. Whites-you know people working with -and Our reasons for expecting this pattern are based on both they feel that foreign Blacks are treated a little more consid- whites' and West Indians' attitudes. With respect to whites, erate than American" (1996:16). Caribbean blacks have a even though we hypothesize that they rank West Indians variety of explanations for this more cordial reception. Their above African Americans, they cannot always distinguish diligence and dependability are cited frequently; among re- between the two groups of blacks. The usual way to identify lated considerations, West Indians are said to be more com- a West Indian heritage is by "hearing that rich English ac- fortable with white authority than are African Americans, to cent coming from a black face" (Arnold 1996:18). For this be less angry toward whites, and not to feel that American reason, some accentless West Indians have special mecha- society "owes them something" (Foner 1985; Waters 1999). nisms for conveying their heritage. Mary Waters describes the of adolescents: "One carried a 1. For we refer to native-born whites as "na- strategies girl simplicity, non-Hispanic as of her chain so that when tive whites." We eliminate foreign-born non-Hispanic whites because in- Guyanese map part key people cluding them would require additional theory and would make the model looked at her keys they would ask her about it and she could more complex. tell them that her parents were from Guyana. One young BLACK-WHITEUNIONS 179 woman described having her mother teach her an accent so non-Hispanic; they were either born in one of 39 British West that she can use it when she applied for a job or a place to Indian countries or native-born and reportedat least one Brit- live" (1996:182-83). With respect to retaining a distinctive ish West Indian ancestor. The remaining individuals were accent, research indicates that puberty is the most common classified as "other." Following the development of these turning point (Pinker 1995). Perhaps if the accent is socially definitions, we sampled one in seven endogamous African desirable, younger arrivals can retain it; contrary to popular American pairs, one in 100 endogamous non-Hispanic white belief, however, even children's speech patterns are not en- pairs, and one in 10 endogamous "other" pairs. All other tirely voluntary (Lippi-Green 1997). Putting these consider- combinations were included in their entirety. Finally, using ations together, we expect that whites view relationships with information on year of arrival to estimate the age of arrival West Indian teen arrivals most favorably, followed (in order) for immigrants, we created seven race/ethnic/generation cat- by childhood arrivals (some of whom will retain their ac- egories: native-born West Indians, foreign-born West Indi- cent), the native-born, and unmarriedadult arrivals. ans who arrived as children (before age 13), those who ar- Finally, from the West Indian perspective, all observers rived as teens (after age 12 for all; before 18 for women and agree that those who arrive as adults will be least interested before 20 for men), those who arrived as adults (after age 17 in a white partner. Some will be involved in relationships for women, after 19 for men), African Americans, native they initiated before emigrating; others will marry later, but whites, and all others. their relatively slight familiarity with American life will re- Before moving on, we must point out that research has duce the attractiveness of a white mate. If level of assimila- cast doubt on the reliability of the ancestry question for iden- tion were the only relevant factor, we would predict that na- tifying native-born West Indians. Qualitative studies suggest tive-born West Indians would be the most likely to choose a that native-born West Indians of lower socioeconomic status native white partner. Research indicates, however, that the may be more likely to identify as African American (Waters less obvious their accent and the more assimilated they be- 1996, 1999). Because West Indians of higher socioeconomic come, the less confident are West Indians that racism can be status are probably more exogamous than their lower-status surmounted (Vickerman 1999; Waters 1999). As a result, counterparts,this bias in reporting may lead to inflated esti- later generations are more antagonistic toward whites and mates of exogamy among native-born West Indians. At the feel a greater bond with African Americans. Therefore we same time, our decision to disaggregate foreign-born West hypothesize that teen arrivals will be most open to white Indians by age of arrival allows us to measure quite confi- mates, followed by childhood arrivals, the native-born, and, dently the effects of generation. finally, adult arrivals. This is the same order of preference Drawing on prior research, our analysis also includes we associate with whites. five covariates: age, education, hypergamy, cohabitation, and place of residence. All but the last of these are dichotomized DATA AND METHODS because our sample does not contain enough interracial The data come from the 5% Public Use Microdata Sample unions to support division into a very large number of cat- (PUMS) of the 1990 U.S. Census. Persons under 16 and egories. For the same reason, we employ only the man's age same-sex pairs were excluded. We consider both marriedand and education ratherthan the age and education of each part- cohabiting couples because "much of the decline in marriage ner. Here we briefly justify the choice of variables. has been offset by increased living together without being To begin with age, younger people are expected to ex- married" (Bumpass and Sweet 1989:615). Partners were hibit higher rates because the passage of time is associated matched irrespective of their position in the household. To with increasing tolerance for exogamy (Patterson 1997). Age identify cohabiting pairs, the U.S. census provides the code is divided into two categories: under 33 and over 32. "unmarriedpartner of household head." Calculations based Educational attainment is also an influence, partly be- on this code underestimatethe cohabiting population, but the cause educational institutions provide opportunities for data offer no alternative. Another problem is that the PUMS otherwise segregated groups to intermingle and partly be- does not take dissolved unions into account. Because inter- cause higher education promotes universalistic attitudes racial partnerships are likely to be more fragile than (Kalmijn 1998). Education is divided at the median: those intraracial partnerships, we underestimate the total number with more than a high school diploma and those with the of interracial unions over the life course (Jones 1996). Yet, diploma or less. we see no reason to expect a significant difference in the rate Next is the variable that identifies hypergamous couples; of dissolved unions between West Indians and whites and that is, couples in which the man's education is greater than between African Americans and whites. Still, the 1990 cen- the woman's. Merton (1941) suggested that black-white sus is less than ideal for the study of intermarriage. unions were disproportionately hypergamous because this The large size of the PUMS disposed us to sample, but combination upgraded both partners: it brought economic defining the groups of interest was a precondition for sam- improvement to lower-status white women while providing pling. Following Kalmijn (1996), we defined as African black males a partnerfrom a more prestigious race. We used American persons black on race, born in the United States, the entire range of educational values in the PUMS to create non-Hispanic, and reporting no ancestry other than black, this measure; all couples in which the male's code was higher African, or American. West Indians were black on race and than the female's were defined as hypergamous. 180 DEMOGRAPHY,VOLUME 38-NUMBER 2, MAY2001

TABLE 1. UNIONS BY RACE, ETHNICITY,GENERATION, AND GENDER

Men Women

FBWI FBWI FBWI African Race/Ethnicity/Generation NB Whites NBWI Children Teens Adults Americans Others Total Native-BornWhites 2,177,800 47 21 21 61 1,753 108,425 2,288,128 Native-BornWest Indians 80 111 11 8 46 263 151 670 Foreign-BornWest Indian Children 29 15 40 19 66 191 71 431 Foreign-BornWest Indian Teens 45 19 57 124 237 227 120 829 Foreign-BorWest Indian Adults 109 51 98 188 2,837 699 351 4,333 AfricanAmericans 6,174 195 162 122 420 146,608 5,921 159,602 Others 90,012 110 95 77 341 3,977 277,231 371,843 Total 2,274,249 548 484 559 4,008 153,718 392,270 2,825,836

Another variable distinguishes cohabiting from married do so, we multiply the antilog of the exogamy preference couples. Research shows that "inclusion of both cohabitation coefficient by 1,000.3 and marriage as union types reduces by nearly one-half the We estimate five models. The first controls only for the race difference in the proportion of women forming a union subgroupmarginals; the second adds the covariates; the third by age 25" (Manning and Smock 1995:509-10). In other adds the bivariate associations among the covariates and the words, cohabitation should be more common in interracial bivariate associations between the covariates and the men's than in intraracialpartnerships. and women's subgroups; the fourth adds interactions of the Finally, we incorporate region, our only trichotomy. covariates with the exogamy preference coefficients; and the Here we distinguish residents of the New York CMSA, the fifth adds asymmetry parameters to accommodate the gen- South, and those living elsewhere. New York is crucial be- der imbalance in interracial pairings (Smits, Ultee, and cause it contains the largest number of Caribbean blacks. Lammers 1998). We obtain an acceptable fit with Model 4, Similarly, the South is a region of African American concen- but Model 5 improves upon it slightly. Additional efforts to tration. Residence in these two areas should foster West In- refine the estimate were not productive. dian and African American endogamy.2 We use these five covariates in two ways. First, we con- RESULTS trol for differences in the distribution of the subgroups on The basic cross-classification of women and men by sub- the covariates. For instance, African Americans' mean group membership appears in Table 1. The diagonal cells in- schooling is lower than that of native-born West Indians. Our dicate endogamy. Percentaging the diagonals (not shown) re- estimates must adjust for this difference. Second, we con- veals that native whites have the highest endogamy rate (96% struct interactions between each covariate and each type of for women, 95% for men), followed closely by African interracial union. This step is needed, not only because it Americans (95% for women, 92% for men). Among West In- makes African Americans and native-born West Indians sta- dians, as expected, the foreign-born who arrived as adults tistically the same on schooling, but also because we expect are the most endogamous; for instance, 71% of the women that persons with more education will be more receptive to in this subgroup are endogamous. interracial partnershipthan persons with less education. A useful preliminary calculation appears in Table 2, We estimate a series of hierarchical log-linear quasi- which gives the percentages of partneredblack individuals, symmetry models (Bishop, Feinberg, and Holland 1975). by ethnicity, generation, and gender, who report a native The parameter of theoretical interest in these models is the white partner.These descriptive statistics show that West In- exogamy preference coefficient, which is the logarithm of dians, with only one exception, are more likely than African the odds that two individuals choose an exogamous union to Americans to have native white partners. (The exception is the odds that these individuals choose an endogamous West Indian-bornmen who arrived as adults.) This is not sur- union. For ease of interpretation, we translate these log- prising, however, because adult arrivals may be previously odds into exogamy rates per 1,000 endogamous unions; to partnered. Yet, West Indian-born women arriving as adults are slightly more likely (1.52%) to have native white part- ners than are African American women (1.14%). This find- ing is a surprise. 2. Because the couples we study are already together, any contempora- neous measure of "opportunity"is flawed. Ideally, residential location be- fore marriage should be used. Better still would be measures of residential segregation, but these are not available in the 5% PUMS. 3. A methodological appendix is available from the authors on request. BLACK-WHITEUNIONS 181

TABLE2. PERCENTAGESOF PARTNEREDBRITISH WEST exogamy preference coefficients and the gender asymmetry INDIANAND AFRICAN AMERICAN INDIVIDUALS coefficients; the second combines these two parameters into WITHNATIVE WHITE PARTNERS, BY GENERA- one. This approachproduces different coefficients for unions TIONAND NATIVITY between black men and white women and unions between white men and black women. In the results for Model as Generation/Nativity Men (%) Women (%) 1, expected, exogamy preference is lowest for West Indian adult FB Adult ArrivedBWI 2.52 1.52 arrivals. In a sample containing partneredWest Indian adult (109) (61) arrivals and partnered native whites, we find only 1.53 FB Teen ArrivedBWI 5.43 3.76 unions between West Indian adult arrivals and native whites (45) (21) for every 1,000 endogamous unions formed by sample mem- FB Child ArrivedBWI 6.73 4.34 bers. The coefficients then increase monotonically with gen- (29) (21) eration: from teen arrivals (3.44) to child arrivals (4.98) to All Foreign-BornBWI 3.27 2.04 the native-born (5.90). No peak is observed for teen arrivals. (183) (103) Even more unexpected, the exogamy preference coefficient Native-Born BWI 11.94 8.58 for African Americans is highest of all (6.74). These results (80) (47) mean that group size, the only factor for which we have con- AfricanAmericans 3.87 1.14 trolled so far, is an important reason why the percentage of (6,174) (1,753) West Indians in interracialpartnerships is higher than the per- centage of African Americans. In other words, the West In- Notes:N is shownin FB = parentheses. foreign-born. dian advantage is reduced considerably by controlling for the difference in group size. Interestingly,we find little difference in the size and rank In addition, Table 2 shows that West Indian-nativewhite order of the exogamy preference coefficients for Models 1, 2, and 3. This indicates that differences in the distri- unions generally display less gender asymmetry than do West subgroups' Indian-African American unions. To be sure, for all sub- bution on age, education, hypergamy, residence, and cohabi- tation are not determinants of interracial unions. groups, black men are more likely to have white partnersthan important the addition of interactions between the covariates are white men to have black partners,but the ratio of the per- However, and causes in centage of interracially paired black males to the percentage exogamy preference (Model 4) slight changes the rank orderof the in their of interraciallypaired black females is about twice as high for coefficients, brings large changes and an fit. addition African Americans as for any of the West Indian subgroups. magnitude, produces acceptable Finally, of the for leaves the The effect of generation also merits emphasis. Within parameters gender asymmetry (Model 5) new rank order but makes a difference in the genders, West Indians arriving as adults show the lowest rate unchanged mag- nitude of the coefficients. the of of white partnerships, but the percentage rises steadily and Unexpectedly, asymmetry all four West Indian is a peaks among the native-born. This trend fits "straight line subgroups capturedby single param- estimates do not the fit. assimilation theory":rates of interracialunions are higher for eter; separate by subgroup improve In both the West Indian and the African those who have been resident longer. addition, (0.88) American are less than that unions Thus, in the absence of controls, our key hypothesis re- (0.49) parameters 1; is, between black men and white women are more common than ceives some support. Relations between whites and most unions between white men and black women. West Indian subgroups appear to be warmer than between whites and African Americans. We find no support, however, for our subsidiary hypothesis, which is that the presence of a West Indian accent facilitates an interracial partnership. Later-generation, "unaccented" West Indians are the most TABLE3. FIT STATISTICS,INTERRACIAL PARTNERSHIP likely to report a white partner. MODELS These results, however, do not include controls for the Log- Likelihood-Ratio many factors that might affect interracial pairings. Table 3 Model Likelihood Chi-Square BIC Pseudo-R2 the results of the exercise. The fit presents model-fitting 1 -192,096.3 378,221.7 360,330.4 0.721 statistics show as we add progressive improvement (df= 1,461) associations inter- marginals, covariates, among covariates, 2 91,963.3 74,145.5 0.929 of covariates with and -48,967.0 actions exogamy preference, finally (df= 1,455) for Model 4 the parameters gender asymmetry. provides 3 -12,525.4 19,080.2 2,315.5 0.982 first fit: the likelihood-ratio is acceptable chi-square (df= 1,369) times its of and the BIC sta- roughly eight degrees freedom, 4 -8,442.0 10,193.3 -4,320.7 0.988 Model 5 offers a modest tistic is negative. improvement. (df= 1,244) Table 4 the coefficients for presents exogamy preference 5 10,114.0 -5,083.2 0.998 and model. The results of -8,042.2 unions between whites blacks, by (df= 1,241) Model 5 are offered in two ways: the first conveys both the 182 DEMOGRAPHY,VOLUME 38-NUMBER 2, MAY2001

TABLE 4. EXOGAMY PREFERENCE AND ASYMMETRYCOEFFICIENTS FOR FIVE LOG-LINEAR MODELS OF BLACK-WHITE PARTNERSHIP

Model 5 5 White Men/ Black Men/ 1 2 3 4 5 Black Women White Women Unions of Native Whites With AfricanAmericans 6.74 6.71 7.03 13.69 12.07 5.91 24.60 (1.72) (1.15) (0.73) (0.70) (0.69) (0.53) (1.93) Native-bornWest Indians 5.90 4.65 6.13 9.79 10.49 9.23 11.90 (1.45) (1.00) (0.83) (1.79) (1.86) (1.78) (2.29) Foreign-bornWest Indians, 4.98 3.75 6.05 10.21 10.94 9.63 12.40 arrivedas children (1.02) (0.96) (1.18) (1.62) (1.67) (1.66) (2.08) Foreign-bornWest Indians, 3.44 2.76 3.32 6.85 7.13 6.27 8.10 arrivedas teens (0.68) (0.55) (0.38) (1.04) (1.06) (1.07) (1.30) Foreign-bornWest Indians, 1.53 1.14 1.47 4.02 4.24 3.73 4.82 arrivedas adults (0.40) (0.25) (0.20) (0.48) (0.49) (0.52) (0.65) Asymmetry Native whites and African 0.49 Americans (0.03) Native whites and West Indians 0.88 (0.06) Notes:Standard errors are shown in parentheses.All coefficientssignificant at .05 level orbetter. For exogamy preference, this is a two-tailedtest; for asymmetry,a one-tailed test.

The last two columns of Table 4 separate the exogamy These patterns contradict our key hypothesis, that rela- preference coefficients of native white male-black female tions between native white Americans and West Indians are pairs from those of native white female-black male pairs. To more cordial than relations between native white Americans illustrate how these numbers are obtained, in the case of na- and African Americans. As explained earlier, the trend most tive white males, we multiply the exogamy preference coef- consonant with that scenario would be that the exogamy pref- ficient by the asymmetry coefficient; in the case of native erence coefficient of West Indian teen arrivals is larger than white females, we multiply the exogamy preference coeffi- that of African Americans. This is not the case, however. Fur- cient by the reciprocal of the asymmetry coefficient.4 thermore, none of the explanations for the hypothesized cor- The penultimate column contains exogamy preferences diality between West Indians and whites suggest that gender for unions between white men and black women. Compari- operates differently in West Indian-white unions than in son of West Indian with African American rates (significance African American-white unions. To be sure, the gender dif- tests not shown) yields the following results. First, West In- ferences displayed in Table 2 motivated us to test a model dian female teen arrivals are as likely to have white partners that included gender asymmetry. Our key hypothesis, how- as are African American women (6.27 vs. 5.91). Second, both ever, offers no basis for expecting that relationships between West Indian female child arrivals (9.63) and native-born West Indians and whites would manifest less gender asym- West Indian females (9.23) are more likely to have white metry than those between African Americans and whites. partners than are African American women (5.91). Table 5 contains the interaction terms generated by The last column contains exogamy preferences for Model 5. Multiplying a coefficient by an interaction (or its unions between black men and white women. The findings reciprocal) conveys the magnitude of the interactive effect, can be conveyed in a single generalization: all West Indian just as multiplying a coefficient by an asymmetry parameter men are significantly less likely to have white partners than (or its reciprocal) conveys the magnitude of the gender ef- are African American men. The closest West Indian figure, fect. To illustrate, for an African American female with more for child arrivals, is about half the African American figure than a high school education, the exogamy preference coef- (12.4 vs. 24.6). ficient for native white male-African American female pairs (5.91) is multiplied by 1.355; this step raises its value to 8.01 unions unions formed 4. For example, the exogamy preference coefficient for native white exogamous per 1,000 endogamous by male-African American female pairs is 12.07 x 0.49 = 5.91; for native white individuals with these characteristics. On the other hand, for female-African American males, it is 12.07 x 1 / 0.49 = 24.6. an African American female with a high school education or BLACK-WHITEUNIONS 183

TABLE5. INTERACTIONSOF EXOGAMYPREFERENCE WITH COVARIATES

CovariateAdjustment Factor Unions of Native Whites With Mean Preference Education Age < 32 New York South Cohabitation AfricanAmericans 12.07 1.355 1.394 1.016a 0.744 2.105 (0.69) (0.049) (0.052) (0.066) (0.046) (0.079) NB West Indians 10.49 1.737 1.456 1.126a 0.611 2.088 (1.86) (0.152) (0.135) (0.154) (0.086) (0.312) FB West Indians,Arrived as Children 10.94 1.186a 1.185a 2.193 0.773a 2.362 (1.67) (0.161) (0.140) (0.492) (0.147) (0.274) FB West Indians,Arrived as Teens 7.13 1.097a 1.346 1.350 0.722 2.299 (1.06) (0.135) (0.127) (0.158) (0.082) (0.269) FB West Indians,Arrived as Adults 4.24 1.570 1.719 0.981a 0.890a 2.641 (0.49) (0.108) (0.134) (0.103) (0.095) (0.232) Notes:Standard errors are shown in parentheses.All coefficientsare significant at .05 (two-tailedtest) unless indicated by a.NB = native-born;FB = foreign- bomn. less, the coefficient for native white male-African American American-white union. Except for hypergamy, the analysis female pairs must be decreased by a factor of 1.355: that is, supports the theory. African Americans who are young, are 5.91 x 1 / 1.355 = 4.36 exogamous unions per 1,000 endoga- well educated, live outside the South, and/or cohabit are sig- mous unions formed by persons with these attributes.A value nificantly more likely to have a white partner.In most cases, of 1.000 implies no effect; a value larger than 1.000 implies these variables operate similarly for West Indian-white that the exogamy preference coefficient increases as the co- unions, but the association does not attain significance in a variate increases; and a value smaller than 1.000 implies few of the subgroups. In sum, most expectations regarding that the coefficient decreases as the covariate increases. the covariates are supported. Interactions with hypergamy are absent from the table because they did not prove significant in any case. This is DISCUSSION not because black-white unions are not hypergamous; for in- Before drawing together our findings, it is appropriateagain stance, African American-white unions are slightly more to acknowledge that 1990 Census data have several short- hypergamous than African American-African American comings. They are cross-sectional; they provide no informa- unions (33.8% vs. 26.2%). Yet, additional analysis (not tion about previous unions; they identify native-born West shown) reveals that compositional differences on the Indians inadequately; and they underestimate cohabitation. covariates, particularly education, account wholly for this Yet, data that surmount these obstacles are neither currently effect. When all the covariates are introduced, the interac- available nor likely to be available soon. West Indian blacks tion between hypergamy and the African American exogamy are simply too small a group to be represented adequately in preference rate declines to 0.98. This number differs insig- a source other than U.S. Census microdata samples. Thus nificantly from 1.000. quantitative researchers must keep the shortcomings in mind The remaining interactions operate in the expected di- when formulating conclusions. rection and at least occasionally are statistically significant. A large ethnographic literature indicates that West Indi- Cohabitation is the strongest covariate; all the interactions ans have more cordial relations with whites than do African are greater than 2 and all are significant. The others are in- Americans. Of course, it is difficult to know how the public significant for one or more of the subgroups. For instance, defines West Indians, but research suggests that a Caribbean education does not significantly increase interracial partner- accent is the main means of conveying group membership. ships among West Indians who arrived as teens or children. If this assumption is correct, then observers are most likely Residence in New York was included for its compositional to identify as West Indian those immigrants who arrived as significance. Although we did not anticipate that it would adults or teenagers. Descriptive statistics indicate that West exert an interactive effect, West Indians arriving as teens and Indian women who arrived as adults or teenagers are more children are significantly more likely to have a white partner likely to have native white partners than African American if they are New Yorkers. As for residence in the South, we women, and that West Indian men who arrived as teenagers included it both to control for regional composition and be- (but not as adults) are more likely to have native white part- cause southern residence should reduce acceptance of inter- ners than African American men. These trends mesh fairly racial unions. This expectation is supported, but not for West well with public perceptions. Indian child and adult arrivals. Yet, when we introduce controls for many of the covari- Put differently, Table 5 contains most of the variables ates of intermarriage,West Indians are more likely than Af- that theory predicts will affect the likelihood of an African rican Americans to have native white partnersonly when the 184 DEMOGRAPHY,VOLUME 38-NUMBER 2, MAY2001

West Indians are later-generation females. Moreover, for REFERENCES both genders, the likelihood that a West Indian has a native white partner is higher for those who arrived as children or Alba,R. andV. Nee. 1997."Rethinking Assimilation for a New Era were born in America than for those who arrived as teens or of Immigration."International Migration Review 31:826-74. adults. These patterns accord more closely with "straight Arnold,F.W. 1996. "LosAngeles WestIndian Immigrant Women: line" assimilation theory, which anticipates that the foreign- "Claimin'De Not Black,De Jus'Tillin' De BitterHarvest." Pre- bor/native-born boundary inhibits attraction, than with the sentedat the annualmeetings of the AmericanSociological As- "cordiality" hypothesis, which anticipates that the foreign- sociation,August, New York. born/native-born boundary enhances attraction (Pagnini and Bishop,Y.M.M., S.E. Fienberg,and P.W.Holland. 1975. Discrete Morgan 1990; Waters 1999). Multivariate Analysis. Cambridge, MA: MIT Press. Yet, our results diverge from "straightline" assimilation Blau, P.M., T.C. Blum, and J.E. Schwartz.1982. "Heterogeneity theory in that they do not reveal an additional increase in and Intermarriage."American Sociological Review 47:45-62. exogamy preference among the native-born. We do not inter- Bumpass,L.L. and J.A. Sweet. 1989. "NationalEstimates of Co- pret this plateau as evidence that the "straightline" assimila- habitation."Demography 26:615-25. tion interpretation is incorrect, however, because our esti- Davis, F.J. 1991. WhoIs Black? UniversityPark: The Pennsylva- mate of native-born West Indians may be flawed. The reason nia StateUniversity Press. for the flaw, as Waters (1996, 1999) proposed, is that less Farley, R. 1996. The New American Reality. New York: Russell highly educated native-born West Indians may be more likely Sage Foundation. to identify in the census as African American than as West Foner,N. 1985."Race and Color: Jamaican Migrants in Londonand Indian. It is difficult to predict just how such a decision ." International Migration Review 19:708-27. might affect the native-born exogamy preference rate. Our Gilbertson,G., J. Fitzpatrick,and L. Yang. 1996. "HispanicInter- findings imply that an African American identity, relative to marriagein New YorkCity: New EvidenceFrom 1991." Inter- a West Indian identity, increases the exogamy rate for men national Migration Review 30:445-59. but decreases it for women. At the same time, education in- Gordon,M. 1979. "Identificationand Adaptation: A Studyof Two creases the exogamy rate, regardless of gender or ethnicity. Groupsof JamaicanImmigrants in New YorkCity." PhD dis- Unfortunately our data do not allow us to quantify the effect sertation,Department of Sociology, City Universityof New of these countervailing forces. Therefore, following York. Hirschman (1994), we place greater emphasis on the find- Hildebrandt,M. 1999. "TheRole of Black Ethnicityin Interracial ings obtained for immigrants who arrived as children or teen- Marriages."Presented at the annualmeetings of the EasternSo- agers; these findings accord with "straight line assimilation ciological Society,March 5, Boston. theory." Hirschman,C. 1994. "Problemsand Prospectsof StudyingImmi- Even so, most Americans, black and white, honor the grantAdaptation From the 1990 PopulationCensus: From Gen- "one drop" rule (Waters 1991). According to this conven- erationalComparisons to the Processof 'BecomingAmerican.'" tion, "a black is any person with any known African black International Migration Review 28:690-713. ancestry" (Davis 1991:5). Hence, for persons of West Indian Hwang,S.-S. and R. Saenz. 1990. "TheProblem Posed by Immi- descent, "straight line" assimilation means assimilation to grantsMarried Abroad on IntermarriageResearch: The Case of an African American rather than a European American ." International Migration Review 24:563-76. lifestyle. For some, this implies an "underclass"outlook. In Jones,F.L. 1996. "Convergenceand Divergence in EthnicDivorce the inner city, as Portes and Zhou observe, some children of Patterns:A Research Note." Journal of Marriage and the Fam- black immigrants are already taking on "the adversarial sub- ily 58:213-18. culture developed by native youths to cope with their own Kalmijn,M. 1993. "Trendsin Black/WhiteIntermarriage." Social difficult situation" (1993:83). For many more, assimilation Forces 72:119-46. to an African American lifestyle implies a middle-class out- . 1996. "The SocioeconomicAssimilation of Caribbean look (Neckerman, Carter, and Lee 1999). Regardless of so- AmericanBlacks." Social Forces 74:911-30. cial class, however, being an African American means being . 1998. "Intermarriageand Homogamy:Causes, Patterns treated differently: it means discrimination in the housing and Trends." Pp. 395-421 in Annual Review of Sociology 24, market, in the labor market, in obtaining a taxi. For this rea- editedby K.S. Cook and J. Hagan.Palo Alto: AnnualReviews son, Waters (1999) observes that black immigrants pursue Inc. assimilation more ambivalently than do white immigrants. Lieberson,S. and M.C. Waters.1988. FromMany Strands. New Moreover, because of the "one drop" rule, this ambivalence York:Russell Sage Foundation. is just as strong for interracially paired West Indians as for Lippi-Green,R. 1997.English With an Accent.London: Routledge. intraracially paired West Indians. Thus, although a compari- Manning,W.D. and P.J.Smock. 1995. "WhyMarry? Race andthe son of West Indian-white unions and African American- Transition to Marriage Among Cohabitors."Demography white unions is theoretically interesting, until the "one 32:509-19. drop" rule is abandoned, interracial partnerships will not Merton,R. 1941. "Intermarriageand Social Structure:Fact and carry the same implications for assimilation that interethnic Theory."Psychiatry 4:361-74. partnerships do. Neckerman,K.M., P. Carter,and J. Lee. 1999. "SegmentedAssimi- BLACK-WHITEUNIONS 185

lation and Minority Cultures of Mobility." Ethnic and Racial Tucker and C. Mitchell-Kernan. New York: Russell Sage Foun- Studies 22:945-65. dation. Oropesa, R.S. and N.S. Landale. 1997. "In Search of the New Sec- Smith, T. 1991. WhatDo Americans ThinkAbout ? New York: ond Generation: Alternative Strategies for Identifying Second American Jewish Committee. Generation Children and Understanding Their Acquisition of Smits, J., W. Ultee, and J. Lammers. 1998. "Educational Ho- English." Sociological Perspectives 40:429-55. mogamy in 65 Countries: An Explanation of Differences in Pagnini, D.L. and S.P. Morgan. 1990. "Intermarriageand Social Openness Using Country-Level Explanatory Variables."Ameri- Distance Among U.S. Immigrants at the Turn of the Century." can Sociological Review 63:264-85. American Journal of Sociology 96:405-32. U.S. Bureau of the Census. 1995. Statistical Abstract of the United Patterson, 0. 1997. The Ordeal of Integration. , DC: States: 1995. 115th ed. Washington, DC: U.S. Bureau of the Civitas. Census. Pinker, S. 1995. The Instinct. New York: Harper. Vickerman, M. 1999. Crosscurrents: WestIndian Immigrants and Portes, A. and M. Zhou. 1993. "The New Second Generation: Seg- Race. New York: Oxford University Press. mented Assimilation and Its Variants."Annals of the American Waters, M.C. 1991. "The Role of Lineage in Identity Formation Academy of Political and Social 530:74-96. Among Black Americans." Qualitative Sociology 14:57-56. Qian, Z. 1997. "Breaking the Racial Barriers:Variations in Interra- . 1996. "Ethnic and Racial Identities of Second Generation cial Marriage Between 1980 and 1990." Demography 34:263- Black Immigrants in New York City." Pp. 171-96 in The New 76. Second Generation, edited by A. Portes. New York: Russell Schoen, R. 1995. "The Widening Gap Between Black and White Sage Foundation. Marriage Rates: Context and Implications." Pp. 103-16 in The .1999. Black Identities: WestIndian ImmigrantDreams and Decline in Marriage Among African Americans, edited by M.B. American Realities. New York: Russell Sage Foundation.