American Indian intermarriage

SOCIAL

SCIENCE

RESEARCH

15,

347-371 (1986)

American Indian Intermarriage GARY D. SANDEFIJR Departments of Sociology and Social Work and Institute for Research on Poverty, University of Wisconsin-Madison AND

TRUDY MCKINNELL Department of Sociology, University of Wisconsin-Madison An analysis of intermarriage patterns with data from the Public Use Microdata Sample of the 1980 Census shows that there is considerably more intermarriage between Indians and whites than between blacks and whites. Indians who live in traditional Indian areas are more likely to be endogamous than those who live in areas where Indians have not traditionally lived, though Indians in traditional Indian areas are also more likely to be married to whites than are Indians in nontraditional Indian areas (after adjustments are made to take into account the number of men and women in the different racial groups). The results also indicate that endogamous American Indians are poorer and less educated than any other type of couple, including endogamous black Americans. 8 1986Academic PRSS, II-K.

Although there has been some research on contemporary intermarriage involving blacks and whites (see, for example, and some research on intermarriage among ethnic groups (see, Alba and Kessler, 1979), there has been almost no research porary patterns of American Indian and white intermarriage.’

patterns of Heer, 1974) for example, on contemIn addition,

We thank Ada Deer, Irv Gartinkel, Douglas Gurak, Robert Hauser, Robert Schoen, Russell Thorton, and Henry Walker for their comments on an earlier version of this paper that was presented at the 1985 American Sociological Association annual meeting. We thank Betty Evanson for editorial advice. This research was supported by Grant 4OA-83 from the U.S. Department of Health and Human Services to the Institute for Research on Poverty and by a grant from the National Institute of Child and Human Development to the Center for Demography and Ecology. Requests for reprints should be sent to Gary D. Sandefur, Institute for Research on Poverty, 1180 Observatory Drive, 3412 Social Science Building, Madison, WI 53706. ’ Throughout the paper we define an American Indian as someone who listed his or her race as American Indian, Eskimo, or Aleut in the 1980 Census. Individuals who listed 347 0049-089X/86 $3.00 Copyright 0 19%6 by Academic Press. Inc. All rights of reproduction in any form reserved.

348

SANDEFUR

AND MC KINNELL

we know almost nothing about the socioeconomic characteristics of Indian endogamous and racially intermarried couples. Existing evidence indicates that intermarriage of Indians and whites is much more prevalent than that of blacks and whites, and that the extent of Indian/white intermarriage has increased dramatically in recent decades. Data from the 1960 Census showed that over 15% of Indian men, in contrast with 1% of black men, were married to white women (Carter and Glick, 1970). The 1970 Census showed that 33% of married Indian men had white wives and 35% of married Indian women had white husbands (Passe], 1976). Data from the 1976 Survey of Income and Education showed that over 40% of Indian men between the ages of 24 and 54 were married to white women, whereas around 2% of black men in that age range were married to white women (Sandefur and Scott, 1983). Part of this apparently dramatic increase in intermarriage involving whites and American Indians probably results from the fact that a number of people changed their self-identification from white to Indian over the period from 1960 to 1980. Passe1 (1976) has shown that approximately 67,000 people changed their self-identification from white to Indian between 1960 and 1970, and Passe1 and Berman (1985) have shown that as many as 358,000 people did so between 1970 and 1980. Thus, a substantial proportion of individuals who reported that they had a white spouse in 1976 were perhaps people who identified themselves as whites in a previous census or censuses. This helps account for the change from an intermarriage rate with whites of around 15% in 1960 to 40% in 1976. Many of these individuals are probably the products of intermarriages or perhaps a succession of intermarriages. Unfortunately, it is not possible to know from census data who has changed his or her racial identification since a previous census. This makes data on intermarriage difficult to interpret. There are, however, areas of the country in which Indian self-identification has been fairly consistent since 1960. These are the areas with the largest concentrations of Indians, which Passe1 and Berman (1985) refer to as “traditional Indian areas.” One would expect patterns of Indian/white intermarriage to be somewhat different in “traditional Indian areas” than elsewhere. This paper utilizes data from the 1980 Census to examine the patterns of intermarriage involving whites and Indians and compares them with patterns of intermarriage involving whites and blacks. We examine the levels of intermarriage involving the two minority groups and whites inside and outside traditional American Indian areas, and we compare the characteristics of endogamous and exogamous couples. Indian ancestry but non-Indian race are not treated as Indians in the analysis. Lieberson and Waters (1985) have examined intermarriage patterns among persons with Indian ancestry. along with a number of other ancestry groups.

AMERICAN

FACTORS

INDIAN

ASSOCIATED

INTERMARRIAGE

349

WITH INTERMARRIAGE

Sociological studies of intermarriage have focused on two distinct levels: (1) explanations of why some groups are more likely to intermarry with the dominant group than others and (2) explanations of why some individuals are more likely to intermarry with individuals in the dominant group than others. Group Determinants of Intermarriage

Theorists and researchers who have examined intermarriage at the group level suggest that the general nature of group relationships with the dominant group will be important in explaining differences in the prevalence of intermarriage with the dominant group. One of these factors is the history of the interaction between the group and the dominant group. Park’s (1950) race-cycle theory suggested that relations between racial groups go through a number of identifiable stages: contact, competition and conflict, accommodation, and assimilation. One aspect of assimilation was assumed to involve the large-scale intermarriage of members of the focal groups. Park suggested that the character and nature of the initial stages in the race cycle had a tremendous influence on subsequent stages: if the initial stages of contact, competition and conflict, and accommodation occurred more rapidly for one group than for another, or if the nature of these stages differed, the speed of amalgamation and assimilation might well differ, even if initial contact had occurred at approximately the same time. A detailed analysis of the history of Indian/white and black/white relations is beyond the scope of this paper. There are, however, key differences in these histories that may be important in understanding current patterns of interracial relations, including intermarriage. First, practical political and economic reasons promoted marriages between whites and Indians, but not marriages between blacks and whites. Prior to the decision of the U.S. government in the early 1800s to “remove” most of the Indians who were east of the Mississippi to western areas, good relations between Indians and whites were facilitated by intermarriage. The French were reputed to have no aversion to marrying Indians (Lauber, 1913), and one celebrated intermarriage in the colony of New York involving a prominent white man and a wife from the Six Nations was said to have greatly facilitated cooperation between the New York colonial government and the government of the Six Nations (Maury, 1872). Soldiers on the frontier sometimes married Indians; trappers, traders, and agents often did so. There were, in fact, some legal attempts to promote marriages between whites and Indians. In 1784, a bill was presented to the Virginia legislature providing that “every white man who married an Indian woman should be paid ten pounds, and five for each child born of such a marriage;

350

SANDEFUR

AND MC KINNELL

and that if any white woman married an Indian she should be entitled to ten pounds with which the county court should buy them livestock” (Beveridge, 1919, pp. 239-241). In 1824, William H. Crawford advocated similar legislation before the U.S. Congress. Neither measure was passed, however (Beveridge, 1919), and the official policy of the U.S. government toward American Indians became one of forced removal and isolation. There was no such inconsistency in colonial and early U.S. attitudes toward black/white intermarriage. Several colonies had laws explicitly condemning such unions. Available data suggest that though intermarriage was strongly discouraged and subject to legal penalties, sexual relations involving whites and blacks were more common, resulting in a substantial population of mulattoes in the early colonial days (Johnston, 1970).* The second difference in the two racial histories concerns their contrasting population sizes. The threat posed by Indians to whites was largest during the late eighteenth and early nineteenth centuries; thereafter, a combination of disease and war, especially the former, drastically reduced the size of the Indian population from somewhere between 1 to 15 million at initial contact in 1492 (Dobyns, 1966; Thornton, in press) to 248,253 in 1890 (U.S. Bureau of the Census, 1975).3 The size of the black population increased during this time to approximately 7.5 million in 1890 (U.S. Bureau of the Census, 1979), largely because it was to the advantage of whites for the black population to increase during the slavery era. By the beginning of the twentieth century Indians were a very small group, and well removed from the major population centers of the United States. Indians have only recently moved in large numbers to urban areas, and in no major metropolitan area do they constitute more than 5% of the population of the area. * There is also evidence of intermarriage between Indians and blacks, especially in the Southeast and mid-Atlantic states, and among Indians from these areas and blacks who moved with them during the removal period in the early 1800s (Berry, 1963). In some of the tribes that were removed from the Southeast. blacks were granted tribal citizenship during Reconstruction (Gibson, 1971). This intermarriage resulted in the “racial hybrid” groups (to use the language of Berry) that now live in the eastern United States, and the “red-black” people of Oklahoma (Wright, 1981). However, the small size of the black and Indian populations relative to the white population meant that intermarriage involving the former two groups had a smaller impact than white/black and white/Indian unions. Further, after removal of most Indians from areas east of the Mississippi, the opportunities for intermarriage between blacks and Indians greatly diminished. Since that time the isolation of Indians in the rural Southwest, West, and North Central parts of the United States and the concentration of blacks in the South and Northeast have combined to impede the black/Indian marriages that might have otherwise taken place. ’ There is considerable disagreement over the size of the American Indian population in 1492. Dobyns’ estimate of 15 million is at the high end, the 1 million figure at the low end. Thornton (in press) suggests that it was probably somewhere in the middle. Whatever its size, it is clear that the size of the American Indian population declined considerably between 1492 and 18%.

AMERICAN

INDIAN

INTERMARRIAGE

351

Because of the isolation of Indians and their small numbers, there has been much less conflict between whites and Indians over the issues that have divided blacks and whites during this century-including residential segregation, school desegregation, and economic well-being. The current relative population sizes of blacks and Indians are also important for another reason. Blau (1977) argues that the rate of intergroup associations is directly related to group size; specifically, the proportion of group members who are intermarried is an inverse function of group size. Much of this effect is simply due to the numbers involved: the size of an ethnic or racial group sets a limit on the number of potential spouses from the group that are available. There are small groups (e.g., Jewish Americans), however, for which the rate of intermarriage is quite low. The fact that Indians have not posed a threat to white society in this century may help explain why studies of social distance as perceived by whites between themselves and other groups have consistently shown that whites are more favorably inclined to accept associations with Indians than with blacks (Bogardus, 1968). Whatever the reason, these studies show that whites say they would be more willing to marry Indians than they would be to marry blacks. Individual Determinants of Intermarriage A number of empirical findings regarding the individual determinants of intermarriage between whites and other groups are important in understanding Indian/white intermarriage. First, a great deal of research suggests that black men are much more likely to marry white women than black women are to marry white men (Davis, 1941; Heer, 1974). We do not know whether this gender effect also applies to intermarriages between whites and American Indians. Second, research suggests that minority men are able to exchange nonracial status (e.g., occupation or education) for the racial status of white women. This option is assumed not to be open to minority women because men are not as interested in the occupation or education of potential spouses as are women (Blau, 1977; Davis, 1941; Merton, 1941; Schoen and Cohen, 1980). Schoen and Wooldredge (1985) argue that we should view this explicitly within the framework of exchange theory: the status of the minority group member becomes higher through marriage to a white spouse, and the status of the majority group member improves through marriage to someone with higher status on dimensions other than race. Other research shows, however, that education is related to greater tolerance of other racial groups and perceptions of smaller social distance between an individual and members of other racial groups (Bogardus, 1968). From this research, one would expect the occurrence of intermarriage to increase with the education of individuals, regardless of their race. Others suggest thar colleges provide one of the few settings in our still

352

SANDEFURANDMC

KINNELL

highly segregated society in which individuals from different racial groups come into contact as equals, and thus we should expect the level of intermarriage to be highest among the college educated. Third, research indicates that generation is an important factor in determining the likelihood of intermarriage (Kikumara and Kitano. 1973; Schoen and Cohen, 1980). The more extensive a minority group member’s ancestry in this country, the more likely an individual is to intermarry. This is impossible to apply in the case of Indians and blacks, however, since both groups have been in this country for some time. Hypotheses

Previous research and theories of minority group interaction suggest, then, that the prevalence of intermarriage will vary across racial groups and that, within racial groups, the prevalence of intermarriage will vary with gender, status, and generation. In this paper we are most interested in marriages between Indians and whites, and we compare the prevalence of their intermarriage to that involving blacks and whites. As we mentioned above, generation would not be a meaningful variable in the case of Indians and most blacks, so we do not attempt to include it in the analysis. We also do not examine the effects of status (e.g., education and occupation) on intermarriage. The theories and research outlined above allow us to develop a number of alternative hypotheses regarding the association between race and gender on the one hand and endogamy and intermarriage on the other. We develop alternative hypotheses in order of increasing complexity, beginning with hypotheses that involve only race, and then moving to those that involve gender. The first hypothesis is that endogamy is more common than exogamy. Past research suggests that this is indeed the case, and at least one of the classic definitions of minority group (Wagley and Harris, 1964) includes a tendency toward endogamy. The second hypothesis is that the level of endogamy varies across racial groups, i.e., some racial groups may be more endogamous than others. The third hypothesis builds on the second by assuming that in addition to differing levels of endogamy, the level of intermarriage between whites and blacks differs from the level of intermarriage between whites and Indians. There are good reasons to expect more intermarriage of whites and Indians than of blacks and Indians. These reasons were discussed above. Each of the preceding hypotheses deals only with the relationship between race of husband and race of wife. Yet, as the research discussed above has shown, marriages of black women and white men seem to occur far less frequently than marriages of black men and white women. So, the fourth hypothesis is that the frequency of intermarriage varies with the gender of the majority and minority group member involved.

AMERICAN

INDIAN

INTERMARRIAGE

353

Marriages involving minority men and white women should be more frequent than marriages involving minority women and white men. This hypothesis builds on the third, since we can retain the assumption that the level of each type of intermarriage is higher for Indians than for blacks. Changes in Racial Zdentijcation Any findings regarding intermarriage between whites and Indians could be misleading, since a number of Indians in the 1980 Census had recently changed their racial identification from non-Indian to Indian. It is not possible to determine who did so, and even if it were, it is not at all clear that we should define only those who have consistently designated themselves Indians as the “real” Indians. It is, however, possible to take advantage of the work of Passe1 and Berman (1985), who demonstrate that in those areas of the country which they term “traditional Indian states,” little change in Indian self-identification has taken place over the period from 1960 through 1980.4 “Traditional Indian states” are defined as those with 3,000 or more American Indians in 1950, excluding California (Passe1 and Berman, 1985). They include New York, Michigan, Wisconsin, Minnesota, North Dakota, South Dakota, Nebraska, North Carolina, Oklahoma, Montana, Idaho, Wyoming, New Mexico, Arizona, Utah, Nevada, Washington, Oregon, and Alaska. These states contained 87% of the Indian population in 1950 and 66% of the Indian population in 1980. California is excluded because a great deal of change in self-identification has occurred there. In our analysis we look at intermarriage patterns in both “traditional Indian” and other states. We do not argue that one group consists of “real” Indians and another group does not, or that one group is “socially Indian” 4 Another strategy would be to examine separately those Indians of solely Indian ancestry and Indians of mixed ancestry. We did not do so because there are reasons to exercise caution in interpreting the ancestry information on American Indians collected in the 1980 Census. Both historical and contemporary evidence suggest that census data underestimate the extent of mixed ancestry among American Indians. In the 1910 Census, 48.5% of the Indian population under 20 years of age identified themselves as being of single ancestry (Thornton, in press). In the 1980 Census, 62.4% of Indians identified themselves as being of single ancestry (Passe1 and Berman, 1985). It is hard to believe that the proportion of Indians with unmixed ancestry has increased since 1910, given the increases in the percentage of people who identify themselves as Indians. Further, information from the Bureau of Indian Affairs (BIA) on verifiable blood quantums for individuals who have applied for and received Certificates of Degree of Indian Blood suggest a much higher degree of mixed ancestry than indicated in the 1980 Census. For example, among those who listed Chickasaw as their tribe in the census, 64.5% listed themselves as having a single ancestry. The BIA figures show that only 10% of Chickasaws have single ancestries (Chickasaw Times, 1985). Given these discrepancies, we decided to forgo use of the information on ancestry.

354

SANDEFUR

AND MC KINNELL

while another group is “socially white.” We simply suggest that intermarriage patterns will be quite different in the two sets of states.” Characteristics of Intermarried and Endogamous Couples

Another way to examine intermarriage is to compare the characteristics of endogamous and intermarried couples. There has been relatively little research on this topic. For this purpose census data have clear advantages, since they contain a much wider variety of information than is available from marriage records. Heer (1974) found that black/white intermarried couples differed in a number of ways from endogamous couples. Specifically, individuals who were involved in interracial marriages were more likely to have been married before than individuals who were endogamous, and intermarried couples had fewer children on the average than did black couples. Stevens (1985) found that individuals who were exogamous were less likely to use a non-English language than were individuals who were endogamous. With cross-sectional data it is impossible to determine the direction of causality-i.e., to know whether non-English speakers are more likely to be endogamous, or if intermarriage leads to less use of non-English languages, or if the relationship is reciprocal. It is, however, worthwhile to compare the characteristics of intermarried and endogamous couples to gain some idea of how they differ. We do this below by examining the age, education, income, household size, incidence of poverty, and language usage for endogamous and intermarried American Indian, white,. and black couples. DATA AND METHODS Data

The data for this paper come from the 1980 one-percent Public Use Microdata Sample of the 1980 Census. From this file we selected all households that contained an American Indian, 25% of the households that contained at least one person identified as black, and 3% of the households that contained at least one person identified as non-Hispanic white. This resulted in a sample of 25,571 couples. Approximately 70% resided in non-Indian states. 6 In the 1980 Census, black, white, and 5 Within traditional Indian states, there is also likely to be variation. For example, in states such as Wisconsin and Oklahoma, tribally owned land was allotted to individual Indians near the turn of this century. “Surplus” land was opened to white settlement and Indians were allowed to sell land to whites. This produced a great deal of contact between Indians and whites that did not occur in states such as Arizona and New Mexico, where most Indians continue to live on tribally owned land. We do not attempt to look at differences within Indian states in this paper. 6 This sampling strategy was used to reduce computational costs. Weights were used to adjust for this strategy in the computations for Tables 1, 5, and 6. The statistical method used in the analyses for Tables 3 and 4 is insensitive to this sampling strategy.

AMERICAN

INDIAN

INTERMARRIAGE

355

American Indian were racial categories and householders (household heads) were requested to identify the racial classification of each individual in the household. Thus, the basis for racial classification in this sample is self-identification. Further, since we were interested in black Americans, white Americans and American Indians, we did not attempt to extract a representative sample of households of “other” racial groups. We do, however, want to include marriages involving individuals in our three in the analysis. This aspect of the focal racial groups and “others” sampling design is handled by weighting the Other/Other couples to zero. One of the problems in using census data for the study of intermarriage is that it provides no information on the incidence of intermarriage, but only on the prevalence of intermarriage at a particular point in time. Thus, we are more likely to sample marriages that are stable rather than a representative sample of all marriages. If endogamous and intermarried couples differ in their marital stability, we may gain a misleading picture of the incidence of intermarriage. For this and other reasons, many analyses of intermarriage use data from marriage records in particular locales or states. On the other hand, analyses of intermarriage in local areas may not produce results that are generalizable to the nation as a whole. And since American Indians made up only 0.6% of the U.S. population in 1980, it is difficult to find communities in which this group is large enough to allow analyses of the incidence of intermarriage. We do not argue that studying the prevalence of Indian intermarriage is more appropriate than studying the incidence of intermarriage, but we do feel that both kinds of analyses can produce meaningful and interpretable results. Methods The analysis is divided into two major parts. First, we look at the prevalence and the determinants of endogamy and intermarriage among the three racial groups. We adopt the concepts and principles that have been widely applied in mobility table analyses to estimate measures of the prevalence of endogamy and intermarriage that are unaffected by the distribution of men and women in the three racial groups. These methods have also been used to study intermarriage (Alba and Kessler, 1979; Shavit, 1978; Travis, 1973). Our approach follows the procedures developed by Hauser (1978, 1980; Featherman and Hauser, 1978). The models we estimate are of the form

E[Mijl = aPirj&j,

(1)

where Mij is the observed frequency in cell ij of the cross-classification of wife’s race (i) by husband’s race (j), cx is the grand mean, pi is the vector of row effects, rj is the vector of column effects, and 6, is a set of interaction effects. The structure of 6ij can be specified to reflect the

356

SANDEFUR

AND MC KINNELL

hypotheses outlined above, and tests of the model in Eq. (1) with alternative specifications of 6, can be used to determine which hypotheses can be rejected. The second part of the analysis involves examination of the characteristics of endogamous and intermarried couples. We test for the significance of differences in characteristics through a multivariate analysis of variance. RESULTS Prevalence of Endogamy and Intermarriage

Table 1 contains the proportions of men and women of each group who are intermarried or endogamous. Panel A contains the results for Indian states and Panel B contains the results for non-Indian states. The most obvious difference between the two is the difference in the proportions of American Indians who are endogamous and married to whites. In Indian states, approximately 60% of American Indian men and women are endogamous, whereas in non-Indian states approximately 30% are endogamous. In Indian states, close to 40% of American Indian men and women are married to whites, whereas in non-Indian states the comparable figure is over 60%. In addition, there are higher rates of intermarriage of Indians with blacks and others in non-Indian states than in Indian states. The differences in marriage patterns for whites and blacks across the two sets of states are very slight, especially when compared to the Indian differences. Unfortunately, these results could be misleading, since the proportions in Table 1 are affected by the size of each group and by the sex ratio of each group (Besancency, 1965). An alternative way of measuring prevalence is to compute what Hauser (1978) has referred to as “new” mobility ratios. In this paper they become endogamy and exogamy ratios.’ A major advantage of this approach is that it provides a statistical framework for testing the hypotheses outlined above, and then provides a mechanism for examining the prevalence of intermarriage based on assumptions that are supported by the data. This is done by specifying interactions (6ij from Eq. (1)) among the race of husband and the race of wife that reflect hypotheses to be tested. For example, to test whether the level of endogamy varies across the three groups we may compare a model that constrains endogamy to be equal across groups to a model that allows it to vary. ’ There are at least two other ways to adjust for group size and sex ratio. One is to compare the number of endogamous marriages to the number of intermarriages between two groups (Gurak and Fitzpatrick, 1982; Parkman and Sawyer, 1967). We followed this approach and discovered that. after adjusting for size, the level of intermarriage involving whites and blacks is much lower than the level of intermarriage involving whites and American Indians. A second way is to look at measures of endogamy that adjust for the sex ratio and group size (Rust and Seed, 1985). We followed this approach and found that after adjusting for possible differences due to group size and the sex ratio, American Indians are much less endogamous than are whites or blacks.

AMERICAN

Proportions

INDIAN

357

INTERMARRIAGE

TABLE 1 of Endogamy and Exogamy among Three Races, 1979 Women

Men

White

Black

A. Indian states (N = 7765 couples) 1. Proportions of women marrying White ,986 Black .002 Indian ,004 Other a08 Total

1.000

Indian

Other

Total

.014 .014 .022

1.000 1.001 l.ooo

.014 .008 .074

l.ooo .999 .999

men of different races .013 .382 ,977 ,014 .OOl ,583 ,009 ,021 1.000

1.000

2. Proportions of men marrying women of different races White .981 .OOl .004 Black .034 -950 .003 Indian ,372 .007 ,599 B. Non-Indian states (N = 17,806 couples) 1. Proportions of women marrying men of different races White .985 .008 .608 Black .002 .989 .024 Indian Xl02 .OOl .290 Other .Oll .003 ,077 Total

1.000

1.001

.999

2. Proportions of men marrying women of different races White .983 .OOl ,002 Black .021 .969 .OOl Indian .616 ,023 .286 Note. Source: Public Use Microdata

Sample from 1980 Census.

If the latter model represents a statistical improvement, this suggests that there are differing levels of endogamy. Once the most appropriate model is discovered, the coefficients of this model may be used to construct intermarriage ratios that are unaffected by the size and sex composition of the racial groups in the table. These ratios are computed as follows: Rij

=

6,

Eij,

where 6, comes from Eq. (1) and Eij is equal to the observed frequency in a cell divided by the frequency predicted by the model (Featherman and Hauser, 1978). The 6, may also be used to compare the levels of endogamy and intermarriage. Table 2 contains the design matrices for Hypotheses 1-4 and Table 3 contains some test statistics for models that reflect the hypotheses. The design matrix for Hypothesis 1 (QJ indicates that the prevalence of endogamy should be different from the prevalence of exogamy or inter-

3.58

SANDEFUR

Alternative

AND MC KINNELL

TABLE 2 Specifications of Endogamy and Intermarriage

Parameters

Women Men

White

Black

A. Qz Endogamy and exogamy differ (Hypothesis I) White I 2 Black 2 I Indian 2 2 Other 2 2 B. Qg Racial differences in endogamy (Hypothesis 2) White 1 4 Black 4 2 Indian 4 4 Other 4 4

intermarriage 4 2 4 4

Other

2 2 1 2

2 2 2

4 4 3 4

C. Q5 Indian intermarriage with whites differs (Hypothesis 3) White I 4 Black 4 2 Indian 5 4 Other 4 4 D. Q6 Gender differences in black/white White 1 Black 6 Indian 5 Other 4

Indian

-

4 4 4 -

5

4 3 4

(Hypothesis 4) 5 4 3 4

4 4 4

-

Nore. Qs refer to design matrices; subscripts indicate the number of levels in the matrix.

marriage, i.e., the diagonal elements are I and the off-diagonal elements are 2. The subscript 2 in Q2 refers to the number of levels in the design matrix. We could have chosen any other two integers for the design matrix, since it specifies nominal rather than ordinal or interval independent variables. The diagonal element for Other/Other is excluded since this category is not used in the analyses. Design Matrix Q4 represents Hypothesis 2. This matrix specifies that the prevalence of endogamy for the three racial groups will be different, but that the prevalence of all types of exogamy will be equivalent. Design Matrix Qs, representing Hypothesis 3, assumes that the prevalence of black/white intermarriage does not differ from that of the other types of intermarriage (an assumption that we can test), but that Indian/white intermarriage does. More specifically, the hypothesis is that Indian/white intermarriage is more frequent than black/white intermarriage. Design Matrix Q6 represents Hypothesis 4 for black/white intermarriage. It assumes that Hypothesis 3 is supported, and allows intermarriage between black men and white women to differ from the other types of intermarriage. A test of Hypothesis 4 for white/Indian intermarriage involves specifying

AMERICAN

INDIAN

359

INTERMARRIAGE

TABLE 3 Test Statistics for Models of Intermarriage Model 1. Conditional independence (MSKFS)

L’ __~--

29,974

df

L’ldf _ ---___~

LZH/LIT ..~~

16

1873.38

100.00

2. Hypothesis 1 Equal endogamy, equal exogamy (MSNFS)(Q,)

5,138

15

342.53

17.14

3. Hypothesis 2 All endogamy differs, equal exogamy WWSWQ,)

294.1

13

22.62

0.98

4. B/W differs from other intermarr.

293.8

12

24.48

0.98

112.6

12

9.38

0.38

71.26

11

7.02

0.25

7. Hypothesis 4 Gender differences in I/W

73.95

10

7.40

0.24

8. MWWQ,,)

17.69

6

2.95

0.06

5. Hypothesis 3 I/W differs from other

(MWFWQ,) 6. Hypothesis 4 Gender differences in B/W

OW(FS)(Q,)

Note. F = female of a particular race, M = male of a particular race, S = state of residence, Qs as defined in Table 2.

a separate level parameter for intermarriage between Indian men and white women. Table 3 contains a summary of the statistical tests of Hypotheses I4. L2 can be interpreted as a x2 statistic. The second column gives the degrees of freedom. The third column contains the L’ divided by the degrees of freedom and gives some indication of the strength of the model. The final column can be interpreted as the percentage of the L’ in the baseline model that is unexplained by a particular model. Since we are interested in the impact of Indian vs non-Indian state of residence on intermarriage patterns, we begin with a model of conditional independence as the baseline model. This model assumes that the distribution of men and women of the four different racial groups varies across the

360

SANDEFUR

AND MC KINNELL

two sets of states, but that there is no association between the race of the husband and the race of the wife. The second model in Table 3 is based on the addition of Design Matrix Q2 to the model of conditional independence. The test statistics for this model indicate that it represents a substantial improvement over Model 1; the prevalence of endogamy differs from that of intermarriage. The third model in Table 3 is based on the addition of Design Matrix Q4 to the model of conditional independence. This model provides a substantial improvement over Models 1 and 2, leaving only 0.98% of the L2 for the model of conditional independence unexplained; this indicates that the prevalence of endogamy varies across whites, blacks, and Indians. Model 4 is not a test of one of the four hypotheses, but of whether black/white intermarriage varies significantly from other types of intermarriage. A comparison of this model to Model 3 indicates that there is not a significant difference (at the .05 level). Since these are nested models, we can compare the models by subtracting the L2 for Model 4 from that for Model 3 (294.1 - 293.8 = 0.3, with 1 degree of freedom). Model 5 is a test of Hypothesis 3 and is based on the addition of Design Matrix Qs to the model of conditional independence. This model does achieve an improvement over Model 3 and leaves only 0.38% of the L2 in the initial model unexplained. This suggests that Indian/white intermarriage does differ significantly from other types, including black/white intermarriage. Models 6 and 7 are tests of gender variations in intermarriage with whites for blacks and Indians, respectively. The test statistics for Model 6 indicate that the prevalence of intermarriage between black men and white women does differ significantly from the prevalence of intermaniage between black women and white men, whereas the test statistics for Model 7 indicate that there is not a significant difference (at the .05 level) in the prevalence of intermarriage between Indian men and white women on the one hand and Indian women and white men on the other. Thus, the traditional pattern of gender differences in black/white intermarriage is supported by 1980 Census data, but does not seem to apply to Indian/white intermarriage.* The final model, Model 8, tests for an interaction between the best fitting previous model (Model 6) and the locational variable. The test statistics indicate that there is a significant interaction. Model 8 leaves only 0.06% of the initial L* unexplained. We believe that Model 8 is an appropriate model to use to examine the endogamy and intermarriage parameters (6ij), for the following reasons: (1) the parameters in the model ’ We also tested each of hypotheses separately in the two groups of states; these tests indicated support for Hypotheses l-3 and gender variations in black/white intermarriages in both sets of states, but no support for gender variations in Indian/white intermarriages in either set of states.

AMERlCAN

INDlAN

INTERMARRIAGE

361

are suggested by previous theoretical and empirical work on racial intermarriage; (2) given the number of cases in the analysis (25,571) the model does a good job of explaining the variation in endogamy and intermarriage patterns; (3) examination of the residuals indicates that they are very small in the categories of interest (endogamy and intermarriage between blacks and whites, whites and Indians). There are, however, some reservations to keep in mind in interpreting the parameters yielded by this model. First, the L* for Model 8 is still significant between the .Ol and .OOl levels. Second, the residuals for a few of the intermarriage categories represented by Level 4 in the design matrix of Model 8 suggest that these categories vary from the other types of intermarriage with which they share Level 4. We analyzed models that included separate level parameters for these types; these results did not alter our conclusions about endogamy, intermarriage between blacks and whites, and intermarriage between Indians and whites.’ The estimated endogamy and intermarriage parameters from Model 8 are displayed in Table 4. Within and across the two groups of Indian and non-Indian states, the level parameters specified in Design Matrix Q6 differ significantly (at or below the .OS level) from each other. Each parameter reflects the prevalence of that type of intermarriage, adjusting for the number of married men and women in each racial group. A positive number indicates that a type of endogamy or intermarriage is more prevalent than one would expect, and a negative number indicates that a type of endogamy or intermarriage is less frequent than one would expect if individuals chose their marriage partners without regard to race. Since the values in Table 4 are the effects on the log of the odds, the exponent of the difference between two coefficients gives the ratio of the prevalence of one type of marriage to another. In terms of the general pattern that applies to both sets of states, these results indicate that black endogamy is most prevalent, followed by white endogamy and Indian endogamy. In Indian states, black endogamy is approximately 2.14 times as likely as white endogamy [exp(3.744 3.004) = 2.141, and white endogamy is approximately 2.05 times as likely as American Indian endogamy. In non-Indian states, black endogamy is ’ Since there are two groups of states and 15 types of intermarriage in each group of states, there are 30 categories of interest. Eighteen of these categories are in Level 4. Featherman and Hauser (1978) suggest that if the log of the observed frequency in a cell divided by the expected frequency does not lie outside the bounds of ~0.20, the residuals can be interpreted as proportionate errors. This condition does not apply to 5 categories. These categories are BF/WM in non-Indian states (0.22), BF/OM in non-Indian states (-0.40). IF/BM in non-Indian states (-0.32), IF/OM in non-Indian states (0.21) and BF/IM in Indian states (-0.33). A negative residual indicates that the type of intermarriage is less frequent than suggested by the level parameter, and a positive residual indicates that the type of intermarriage is more frequent than suggested by the level parameter.

362

SANDEFUR

Endogamy and Intermarriage

AND MC KINNELL TABLE 4 Parameters [Based on (MS)(FS)(Q,S)l Women

Men

White

Black

Indian

Other

A. Indian states White Black Indian Other

3.004 - 0.458 1.452 - 1.278

- 1.278 3.764 - 1.278 - 1.278

1.452 - 1.278 2.286 - 1.278

- I.278 - 1.278 - 1.278 -

B. Non-Indian White Black Indian Other

2.950 -0.186 0.807 - 1.102

-1.102 4.977 - 1.102 -1.102

0.807 -1.102 0.561 - 1.102

- 1.102 - 1.102 - 1.102 -

states

approximately 7.59 times as likely as white endogamy, and white endogamy is approximately 10.9 times as likely as American Indian endogamy. The prevalence of endogamy obviously varies more across the racial groups in non-Indian states than in Indian states. The next most prevalent type of marriage is that between whites and Indians. Each type is more prevalent than one would expect, given the distribution of races. Intermarriage between black men and white women is more prevalent than the remaining types of intermarriage, but it is less prevalent than one would find if marriage partners were chosen without regard to race. Next we examine the differences between the two sets of states. First, the results indicate that white endogamy is 1.05 times more prevalent in Indian states than in non-Indian states, whereas black endogamy is 3.36 times more prevalent in non-Indian states than in Indian states. This may be because the non-Indian states include most of the South. Second, Indian endogamy is 5.61 times more prevalent in Indian states than in non-Indian states. This is what we would expect, given the consistency of Indian identity in Indian states as opposed to non-Indian states. Third, the results indicate, surprisingly, that marriage between whites and Indians is more prevalent in Indian states than in non-Indian states. This suggests that part of the high observed rate of white/Indian marriage in nonIndian states in Table 1 is due to the lack of availability of Indian partners in these states. Also, the other types of intermarriage involving American Indians are more prevalent in non-Indian states.” However, the contrast between Indian endogamy and Indian/white intermarriage is higher in I” A test of locational differences in the prevalence of Indian/other marriage between states indicates that these types of intermarriage are significantly more prevalent in nonIndian areas than in Indian areas.

AMERICAN

INDIAN

INTERMARRIAGE

363

Indian states, where Indian endogamy is 2.3 times more likely than Indian/white intermarriage; in non-Indian states, Indian endogamy is 0.78 times less likely than Indian/white intermarriage. Characteristics of Endogamous and Intermarried Couples

Although it is clear that the level of intermarriage with whites is higher for American Indians than for blacks, we do not yet know whether intermarried couples differ in substantively and statistically significant ways from endogamous couples, and whether endogamous black and Indian couples differ significantly from each other. The possibility of differences between intermarried and endogamous couples is especially important in the case of American Indians because intermarried Indians make up such a large proportion of the total Indian population. Table 5 displays selected characteristics of seven types of couples. Table 6 summarizes the tests of significance for differences in characteristics among these types of couples. For this analysis we did not make a distinction between Indian and non-Indian states.” These tests are based on a multivariate analysis of variance. The four contrasts in Table 6 are orthogonal, but they are not a complete set of orthogonal contrasts. One way to approach the differences among these types of couples is to ask whether the characteristics of endogamous couples differ very much from those of intermarried couples. The first contrast in Table 6 (column 1) compares endogamous white, black, and Indian couples on the one hand to intermarried white, black, and Indian couples on the other. Table 6 shows that these two groups of couples differ significantly in every characteristic except household income and per capita household income. The descriptive statistics in Table 5 show that intermarried couples are younger and more educated than endogamous couples. The differences in age are probably due to a cohort effect: more recent cohorts in American society are probably less opposed to interracial contact, including intermarriage, than are older cohorts. The educational levels of intermarried couples are higher than those of endogamous couples, and the black female/white male couples have the highest level of education. Endogamous white couples, however, have the next highest level of education. Couples involving Indians are less educated, and endogamous Indian couples have a startling low level of education--less than 10 years of completed schooling. The significant difference in household size is largely due to the comparatively large households of endogamous black and Indian couples. Endogamous white couples have the smallest households. The language ” In other analyseswith the 1980 Census data we have found that Indians in non-Indian states are economically better off than those in Indian states. The economic situation of Indians in non-Indian states is not as good as the economic situation of whites. See (Sandefur and Sakamoto, 1986) for a discussion of these differences.

Education of wife (years) Household size

3.3 (1.4)

(2.7)

12.1

12.4 (3.5)

Education of husband (years)

43.9

Age of wife

(15.3)

46.7 (15.7)

Age of husband (years)

(years)

WFjWM

Variables

11.2 (3.4) 3.9 (1.8)

10.4 (4.1)

42.3 (14.8)

45.5 (15.5)

BF/BM

(2.2)

4.6

9.8 (4.4)

39.5 (14.3)

42.6 (14.8)

lF/IM

3.6 (1.5)

12.5 (3.0)

(2.9)

12.7

33.3 (10.2)

37.8 (12.0)

WF/BM

3.6 (1.5)

3.4 (1.5)

13.2

(2.6)

(2.5)

(3.6)

13.0

36.8 (13.8)

40.0 (14.2)

BF/WM

Couples

11.9

11.9 (3.4)

37.0 (13.4)

39.8 (13.7)

WF/IM

TABLE 5 Selected Characteristics of Endogamous and lntermarried

3.6 (1.5)

(2.8)

11.6

11.9 (3.1)

37.6 (13.4)

40.1 (14.3)

IF/WM

3.9 (1.7)

11.9 (3.5)

(3.8)

12.3

35.1 (12.0)

37.3 (12.3)

Other

F

2 m

5 E

i%

2

kz

Note.

Poverty

F = female;

rate

Per capita HH income (annual gross)

M = male;

5%

(602%

8436

24,845 (15,283)

7%

Wife uses non-English language

Household income (annual gross)

6%

Husband uses non-English language

1 = Indian;

14%

5644 (4358)

19.358 (12,787)

4%

3%

W = white;

25% B = black.

4261 (3837)

16,760 (13,046)

46%

46%

7%

6710 (4525)

21,043 (12.516)

18%

9%

10%

6694 (5152)

21,149 (14,018)

5%

8%

6%

(6901)

7921

23,666 (17,533)

10%

13%

10%

6594 (4853)

20,775 (12,942)

10%

6%

10%

6461 (4733)

21,797 (14,065)

44%

41%

366

SANDEFUR

AND MC KINNELL

TABLE 6 Tests of Significance for Selected Contrasts among Exogamous and Endogamous Couples

Variables Age of husband Age of wife Education of husband Education of wife Household size Husband uses non-English language Wife uses non-English language Household income Per capita HH income Poverty rate

Endogamous vs exoga.

Contrasts -...-- -.-~~~~~. Endogamous whites Endogamous black Exogamous black vs end. minority vs end. Indian vs exo. Indian

*** ***

** ***

***

***

***

***

***

***

***

***

***

***

***

* **

***

***

***

***

*** ***

* ***

* .Ol < p c .05. ** ,001 < p < .Ol. *** p s .OOl.

variables refer to whether the individual sometimes uses a non-English language in the home. This clearly occurs most often among endogamous Indians, but intermarried couples are generally more likely to do so than are endogamous whites and blacks. Finally, the significant differences in the level of poverty are due to the comparatively high incidence of poverty among endogamous blacks and Indians. The second column and contrast in Table 6 tests for significant differences between endogamous whites on the one hand and endogamous blacks and Indians on the other. This contrast shows that these two groups differ significantly on all the characteristics in Table 5. The third column and contrast tests for significant differences between endogamous blacks and endogamous Indians. It shows that these two groups differ in wives’ education, in household size, language usage, per capita household income, and the prevalence of poverty. Considered together, the two contrasts in columns 2 and 3 show that endogamous whites are older than endogamous blacks and Indians; they are also more educated. The average level of

AMERICAN

INDIAN

INTERMARRIAGE

367

education of endogamous Indian women is lower than that of endogamous black women. The households of endogamous whites are smaller than those of endogamous blacks and Indians, and the households of endogamous Indians are larger than those of endogamous blacks. We discussed the language use of endogamous Indians above. The household income of endogamous blacks and Indians is significantly lower than that of endogamous whites, and when we adjust for household size, the per capita income of endogamous Indians is significantly lower than that of endogamous blacks. Finally, endogamous whites are least likely to live below the poverty line and endogamous Indians are significantly more likely than endogamous blacks to live below the poverty line. The last column and final contrast in Table 6 compares intermarried black/white couples with intermarried Indian/white couples. These couples differ significantly only in the education of the wife. Women in black/white couples have a significantly higher level of education than women in Indian/white couples. The results in Tables 5 and 6 clearly indicate that endogamous Indian couples have a substantially lower standard of living than other couples. Endogamous black couples are significantly better off, but are not as well off as endogamous white couples and intermarried couples. It is, of course, important not to make too much out of these results, since these analyses do not allow us to attribute causality to endogamy and intermarriage, i.e., we are in no position to say that marrying a white person improves one’s life chances. It is likely that intermarriage and the other factors in Table 6 are connected in a number of ways that are difficult to study with cross-sectional data. It is particularly important to be cautious about our finding of very low socioeconomic status among endogamous Indian couples. This may very well be due to what is known as the “reservation effect.” Endogamous Indians are probably more likely to live on reservations and in isolated rural areas with quite limited contact with whites, but also with few opportunities for obtaining good jobs. Thus, it is probably their isolation that explains both endogamy and low incomes. SUMMARY

AND CONCLUSIONS

Our results indicate that blacks are more likely to marry blacks than whites are to marry whites and that Indians are the least racially endogamous. Indians in “Indian” states-those with traditionally higher concentrations of Indians-are more endogamous than those in nonIndian states. Indians are, however, more likely to intermarry with whites in Indian states than in non-Indian states, after adjustments are made for the number of men and women in the different racial groups. Finally, the gender effect (variation in frequency of intermarriage depending on gender of majority and minority group member involved) in black/white

368

SANDEFURANDMCKINNELL

marriages is evident in the 1980 Census data, but this effect is not present in the case of Indian/white marriages. These results are based on an analysis of married couples only. Future research should examine the incidence of intermarriage and examine the covariates (e.g., education, age) of endogamy and intermarriage between whites and Indians. In addition, future research should examine intermarriage among different minority groups. The results also indicate that the characteristics of endogamous and intermarried white, black, and American Indian couples vary considerably. Endogamous Indian couples are poorer and less educated than even endogamous black couples. Black/white couples and white/Indian couples vary only in that the level of education of women in black/white marriages is higher. Of the three racial groups considered in this analysis, intermarriage is most prevalent among the American Indian population. In order to understand better the significance of intermarriage for American Indians, it is necessary to consider it within the context of other changes in their population in recent years. First, there has been an increase in the number of people who identify themselves as American Indians in the census. Most of this increase seems to be in those areas where Indians have not traditionally lived. Second, the American Indian population is becoming increasingly urbanized. In 1980, for the first time in American history, approximately 50% of American Indians lived in urban areas (U.S. Bureau of the Census, 1983). The continuing urbanization of the Indian population is important, since the identity of individual Indians has always revolved around tribal or subtribal affiliations rather than “Indian-ness.” Thus, urbanization has altered the close ties to community and tribe that have been the traditional basis of Indian identity, and has facilitated intermarriage of Indians with other groups. Intermarriage, changes in the self-identification of individuals from white to Indian, and urbanization are likely to have a significant impact on American Indians in the future. At least three distinct possibilities come to mind. First, American Indians may become more and more an ethnic group of mixed racial descent, and less and less a racial group (Thornton, in press). Although the Bureau of Indian Affairs recognizes only those with one-quarter or more Indian blood as eligible for its services, there is no restriction on the ability of Indian tribes to define their own criteria for membership, and those criteria vary widely. Some tribes require only some Indian blood from that specific tribe; at least one requires five-eighths Indian blood from that tribe, but no tribe requires more than that (U.S. Bureau of Indian Affairs, 1981). Tribal definitions of membership often reflect the history of the tribe’s relations with whites. The tribes that continue to be isolated and have experienced little in-

AMERICAN

INDIAN

INTERMARRIAGE

369

termarriage tend to have more restrictive definitions of who is an Indian than those which have had a history of intermarriage. A second possibility is that interest in ethnicity may subside in the future, and those individuals who changed their self-identification from non-Indian to Indian may revert to their previous non-Indian racial identity. This will reduce the prevalence of Indian/white intermarriage, though it should have little impact in Indian states where approximately 40% of Indians are married to whites. Although “lifelong” Indians have always been concerned about “instant Indians”-individuals who assert their Indian identity when it is to their benefit to do so (Yinger and Simpson, 1978)-and at least one respected national Indian leader has called for the exposure of “fakes” who try to influence the course of Indian affairs (Deloria and Lytle, 1984), it is not clear how involved the “new” Indians are in Indian alfairs, nor how seriously they take their own Indian identity. It is also not clear how the children of current mixed marriages will perceive their racial identity in the future.” Third, continuing intermarriage, urbanization, and changes in self-identification may create a division or divisions within the American Indian population similar to the class division among blacks described by Wilson (1978). This division, however, would be along a different set of dimensions-intermarriage, mixed racial descent, and urbanization as well as class. Since Indians are such a small group (0.6% of the population in 1980), these changes”and this division might go relatively unnoticed by the nation at large. In some parts of the country, however, segments of the public and some politicians have begun to question the nation’s promises and commitment to its indigenous peoples. The danger is that those Indians on reservations, those in isolated rural areas, and poor urban Indians (regardless of their “degree” of Indian blood) may be neglected and ignored if comparatively well-off urbanized and intermarried Indians capture the attention of the nation. In sum, it is not clear what impact intermarriage will have on American Indians, but it is clear that a substantial segmentof the Indian population continues to be economically disadvantaged and in need of assistance even as rates of intermarriage with whites increase. REFERENCES Alba, R. D., and Kessler, R. G. (1979). “Patterns of interethnic marriage among Catholic Americans,” Social Forces 57, 1124-l 140. Berry, B. (1%3). Almost White: A Study of Certain Racial Hybrids in the Eastern United States. Macmillan Co., New York. Besancency, P. H. (l%S). “On reporting rates of intermarriage,” American Journal of Sociology 70, 717-721. ‘* The issue of the rights and identity of “mixed blood” Indians is not a new one. Mixed bIood Indians were of concern to 1890 census ofhcials who felt that people with little claim to being Indians were recording themselves as such (Thornton, in press).

370

SANDEFUR

AND MC KINNELL

Beveridge, A. (1919). The Life of John Marshall. Houghton Mifffin, New York, Vol. 1. Blau, P. M. (1977). Inequality and Heterogeneity. Free Press, New York. Bogardus, E. S. (1968). “Comparing racial distance in Ethiopia, South Africa and the United States,” Sociology and Social Research 52, 149-156. Carter, H., and Ghck, P. C. (1970). Marriage and Divorce: A Social and Economic Study. Harvard Univ. Press, Cambridge, Mass. Chickasaw Times (1985). “BIA release new statistics,” Chickasaw Times, Vol. 14, No. 3, p. 13. Davis, K. (1941). “Intermarriage in caste societies.” American Anthropologist 43, 376395. Deloria, V., and Lytle, C. (1984). The Nations Within: The Past and Future of American Indian Sovereignty. Pantheon Books, New York. Dobyns, H. F. (1966). “Estimating aboriginal American population: an appraisal of technique with a new hemisphere estimate,” Current Anthropology 7, 395-416. Featherman, D. L., and Hauser, R. M. (1978). Opportunity and Change. Academic Press, New York. Gibson, A. M. (1971). The Chickasaws. Univ. of Oklahoma Press, Norman. Gurak, D. T., and Fitzpatrick, J. P. (1982). “Intermarriage among Hispanic ethnic groups in New York City,” American Journal of Sociology 87, 921-934. Hauser. R. M. (1978). “A structural model of the mobility table,” Social Forces 56, 91953. Hauser, R. M. (1980). ‘Some exploratory methods for modeling mobility tables and other cross-classified data,” in Sociological Methodology, 1980 (Karl F. Schnessler, Ed.) Jossey-Bass, San Francisco. Heer, D. (1974). “The prevalence of black-white marriage in the United States, 1960 and 1970,” Journal of Marriage and the Family 36, 246-258. Johnston, J. H. (1970). Race Relations in Virginia and Miscegenation in the South, 17761780, Univ. of Massachusetts Press, Amherst. Kikumara, A., and Kitano, H. L. L. (1973). “Interracial marriage: picture of the JapaneseAmericans,” Journal of Social Issues 29, 67-81. Lauber, A. W. (1913). Indian Slavery in Colonial Times, within the Present Limits of the United States, Columbia Univ. Press, New York. Lieberson, S., and Waters, M. (1985). Ethnic Mixtures in the United States, paper presented at 1985 annual meeting of the Population Association of America, Boston, Mass. Maury, A. (1872). Memories of a Hugenot Family. Putnam. New York. Merton, R. K. (1941). “Intermarriage and the social structure: fact and theory,” Psychiatry 4, 361-374. Office of Indian Affairs (1832). Letter Book. Washington, DC. Park, R. E. (1950). Race and Culture: The Collected Papers of Robert Ezra Park (E. C. Hughes, C. S. Johnson, J. Masuoka, R. Redheld. and L. Wirth, Eds.). The Free Press, Glencoe. Parkman, M. A., and Sawyer, J. (1967). “Dimensions of ethnic intermarriage in Hawaii,” American Sociological Review 32, 593-607. Passe], J. S. (1976). “Provisional evaluation of the 1970 Census count of American Indians.” Demography 13, 397-409. Passel, J. S., and Berman, P. A. (1985). Quality of 1980 Census Data for American Indians, paper presented at 1985 annual meetings of the American Statistical Association, Las Vegas, Nev. Rust, P. F., and Seed, P. (1985). “Equality of endogamy: statistical approaches.” Social Science Research 14, 57-79. Sandefur, G. D., and Scott, W. J. (1983). “Minority group status and the wages of Indian and black males,” Social Science Research 12, 44-68.

AMERICAN

INDIAN

INTERMARRIAGE

371

Sandefur, G. D., and Sakamoto. A. (1986). “American Indian Household Structure and Income,” presented at the American Sociological Association Annual Meeting, August. 1986. Schoen, R., and Cohen, L. E. (1980). “Ethnic endogamy among Mexican American grooms: a reanalysis of generational and occupational effects.” American Journal of Sociology 86, 359-366. Schoen, R., and Wooldredge, J. (1985). “Who marries whom: marriage patterns in California, 1970,” Working Papers in Population Studies, No. 8501, University of Illinois, UrbanaChampaign. Shavit, Y. (1978). “Ethnic intermarriage in the United States,” Center for Demography and Ecology, Working Paper 78-l 1, University of Wisconsin-Madison. Stevens, G. (1985). “Nativity, intermarriage and mother-tongue shift,” American Sociological Review 50, 74-83. Thornton, R. (in press). As Snow Before a Summer Sun: The Demographic Destruction and Survival of North American Indians (tentative title). Univ. of Oklahoma Press, Norman. Travis, H. P. (1973). Religious In-marriage and Inter-marriage in Canada, 1934-1969. masters thesis, University of Wisconsin-Madison. U.S. Bureau of the Census (1975). Historical Statistics of the United States. U.S. Govt. Printing Office, Washington, DC. U.S. Bureau of the Census (1979). The Social and Economic Status of the Black Population in the United States: an Historical View, 1970-1978, Series P-23, No. 80, U.S. Govt. Printing Office, Washington, DC. U.S. Bureau of the Census (1983). Ancestry of the Population by State: 1980. Supplementary Report, PC80-Sl-10. U.S. Govt. Printing Office, Washington, DC. U.S. Bureau of Indian Affairs (1981). BIA Profile: The Bureau of Indian Affairs and American Indians, U.S. Govt. Printing Office. Washington, DC. Wagley, C., and Harris, M. (1964). Minorities in the New World. Columbia Univ. Press, New York. Wilson, W. J. (1978). The Declining Significance of Race. Univ. of Chicago Press, Ill. Wirth, L., and Goldhamer, H. (1944). “Legal restrictions on Negro-white intermarriage.” in Characteristics of the American Negro (0. Kleindenz, Ed.), Harper, New York. Wright, J. L., Jr. (1981). The Only Land They Knew: The Tragic Story of the American Indians in the Old South Free Press, New York. Yinger, J. M., and Simpson, G. E. (1978). “The integration of Americans of Indian descent,” The Annals 436, 137-151.