Political scientists have long noted several gender gaps in the attitudes of Americans (see, e.g., Box-Steffensmeier, De Boef, and Lin Reference Box-Steffensmeier, Boef and Lin2004; Chaney, Alvarez, and Nagler Reference Chaney, Alvarez and Nagler1998; Conover Reference Conover1988; Hatemi, Medland, and Eaves Reference Hatemi, Medland and Eaves2009; Kaufmann and Petrocik Reference Kaufmann and Petrocik1999; Manza and Brooks Reference Manza and Brooks1998; Norrander Reference Norrander1999, Reference Norrander and Whitaker2008; Welch Reference Welch1977). One set of gaps indicate that compared with men, women are more likely to be liberal and to support women's and compassion issues such as welfare and education because they are either predisposed or socialized to be nurturing and caregivers, have a feminist consciousness, or more likely to have had experiences such as poverty, discrimination, and motherhood. Women are also more likely than men to be Democrats, in part because the Democratic Party is more likely to take positions that women support and prioritize. This body of literature is well known, so I will not elaborate on it.
One of the problems with the literature described here, however, is that it lumps all men and women together regardless of race, ethnicity, nationality, class, or other social groupings. It is likely that the gender gaps are unique within each group. In fact, the literature on the gender gaps among Hispanics, African Americans, and Asian Americans suggests some variation by race and ethnicity (Bedolla, Monforti, and Pantoja Reference Bedolla, Jessica L. and Pantoja2006; Bejarano Reference Bejarano2014; Bejarano, Manzano, and Montoya Reference Bejarano, Manzano and Montoya2011; Harell and Panagos Reference Harell and Panagos2013; Montoya Reference Montoya1996). Some research on Hispanics finds fairly weak gender gaps, although in the same direction as Anglos (Montoya Reference Montoya1996; Welch and Sigelman Reference Welch and Sigelman1992), while other research finds larger gender gaps (Bedolla, Monforti, and Pantoja Reference Bedolla, Jessica L. and Pantoja2006). A reason for the variation in findings may be that the size of the gender gap among Hispanic Americans depends on levels of acculturation. Compared with Latinos, the more acculturated Latinas become, the more egalitarian they are, and this contributes to a gender gap (Bejarano Reference Bejarano2014; Bejarano, Manzano, and Montoya Reference Bejarano, Manzano and Montoya2011). Among African Americans, the direction of the gender gap tends to be similar to that of Anglos, but the size tends to be smaller (Welch and Sigelman Reference Welch and Sigelman1992). Lien (Reference Lien1998) finds that among Asian Americans and African Americans, the gaps tend to be smaller than among whites, or in some cases they are slightly reversed. The little research that has examined the gender gaps within groups has virtually ignored American Indians. Harell and Panagos (Reference Harell and Panagos2013) offer the only published examination of gender gaps among American Indians, and their study is limited to examining voter turnout and choice among Aboriginals in Canada.Footnote 1
The goal of this article is to help fill this void by examining the gender gaps in party, ideology, women's/compassion issues, and American Indian identity among American Indians in the United States. Then, given the importance of party identification in U.S. politics, it further explores the party gender gap among American Indians.
Harell and Panagos (Reference Harell and Panagos2013) find that although the gender gap in levels of participation among Aboriginal Canadians is similar to that of non-Aboriginal Canadians, the gap in vote choice is greater. Aboriginal Canadians are more likely to vote for the New Democratic Party, which takes the strongest pro-Aboriginal positions, than non-Aboriginal Canadians. This was most pronounced for Aboriginal women. The reason for the large gender gap, they conclude, is that colonization had a greater effect on Aboriginal women than men, which they term the “colonialism hypothesis.”
Considering American Indians (in Canada and the United States), it is likely that colonialism was most harmful for women. Prior to colonialism, women's roles were seen as equal to, although different from, men's roles, and women's economic contributions to their tribes and families were acknowledged and substantial. In some tribes as well, women held leadership roles, and some tribes were matrilineal. However, with colonialism, women's economic contributions were often supplanted; in a society dominated by European Christian values, women and their work became less valued by the tribe (Bonvillian Reference Bonvillian1989; Chiste Reference Chiste1994; Denetdale Reference Denetdale and Goldstein2014; LaFromboise, Hyley, and Ozer Reference LaFromboise, Hyley and Ozer1990). It is not that American Indian men were not harmed by colonialism; they were, but while they were not powerful in the colonized state, they did gain power in the tribe. As Barker notes, “With few opportunities for political power and economic self-sufficiency off the reserve, heterosexual status Indian men were given opportunities in band [tribal] government and reserve life. Unfortunately, they took advantage of these opportunities and even came to feel empowered and entitled to them” (Reference Barker2008, 263). Thus, although both male and female American Indians were harmed by colonialism, only women lost power inside the tribe as well as outside of it.
The foregoing discussion leads to the first expectation: (1) compared with American Indian men, American Indian women are likely to prioritize their American Indian identity over their colonialized identity as a U.S. citizen since colonialism had a disproportionately negative effect on them. While the term “identity” has been used in different ways by scholars (see, e.g., Stryker and Burke Reference Stryker and Burke2000, 284), I use the term “American Indian identity” to refer to the strength of one's self-identification as an American Indian. Phinney's (Reference Phinney1992) third component of identity, ethnic identity achievement, suggests that history is a significant influence on identity strength. Phinney states that “the process of ethnic identity formation appears to involve an exploration of meaning of one's ethnicity (e.g., its history and traditions) that leads to a secure sense of oneself as a member of a minority group” (Reference Phinney1992, 160). Similarly, while examining American Indian identity, Markstrom states that “[e]thnic identity of American Indians is embedded within the local cultural milieu and encompasses an array of dimensions organized around identification, connection, and culture/spirituality” (Reference Markstrom2011, 519). Within the culture/spirituality dimension, Markstrom discusses the significance of history and its understanding in the formation of American Indian identity. Jervis et al. (Reference Jervis, Beals, Croy, Klein and Manson2006), too, find that historical consciousness among American Indians is correlated with the strength of their American Indian identity.
While the expectation is that women have stronger American Indian identity because of their understanding of tribal history and traditions, there are interpretations of tribal history and traditions that reinforce colonialized gender roles. Some traditions that have evolved in recent generations have incorporated the colonizer's gender roles. The effects of colonialism on gender power and roles were strong, and many tribes have incorporated European Christian gender roles and patriarchy (Barker Reference Barker2008, Reference Barker2011; Denetdale Reference Denetdale and Goldstein2014). As gender roles and power have changed since the beginning of colonialism, many tribes and tribal leaders have developed new traditions that reinforce the European colonizer's gender roles (Barker Reference Barker2008, Reference Barker2011; Denetdale Reference Denetdale and Goldstein2014). In her discussion of marriage equality debates among the Cherokees and Navajo, for example, Barker (Reference Barker2011, chap. 6) notes that both sides of the debate used arguments based on tradition and sovereignty to support their positions. Same-sex marriage supporters noted that both tribes recognized a third gender and supported equality prior to colonialism. Opponents relied on recent traditions: those that developed after colonialism as a way to gain acceptance in the larger political, economic, and social culture and are consistent with conservative Christian values.
Another reason women may have a stronger American Indian identity than men is that some men have a stronger identification with the United States through their service in the U.S. military. American Indian men have a strong history of service in the U.S. armed forces that likely has strengthened their identity as U.S. citizens.Footnote 2 Demonstrating the vast presence of American Indians in the military, the ratio of the percentage of American Indians in the military to that in the population is 2.96, while it is 1.05 for whites, 1.04 for African Americans, and .57 for Asian Americans/Pacific Islanders (Wilkins and Stark Reference Wilkins and Stark2011, 210–11). This service has been thought to increase the patriotism of American Indians (Wilkins and Stark Reference Wilkins and Stark2011, 211), as well as the acculturation of other ethnic minorities, specifically, Latinos (Leal Reference Leal2003).
If American Indian women, compared with men, have a stronger American Indian identity, it follows that they would identify more with the Democratic Party. Ethnic identity can affect party identification through people's attachments to social groups (Green, Palmquist, and Schickler Reference Green, Palmquist and Schickler2002; Mangum Reference Mangum2013). Mangum (Reference Mangum2013) uses identity theory to argue that people categorize themselves as part of a race/ethnicity, learn the interests of that race/ethnicity, internalize those interests, and then identify with the party that reflects those interests. Further, he argues that this is likely because the parties take different positions on issues important to a social group. He looks at three aspects of racial/ethnic psychological attachments: categorization (simply seeing oneself as having a particular race/ethnicity), identification (how close one is to that race/ethnicity), and group consciousness (is that race/ethnicity treated fairly?). He finds that categorization and identification are strongly related to party identification even after other issues and ideology are controlled.
Based on this, for American Indian identity to increase individuals' identification with the Democratic Party, the Democratic Party would have to take positions and be seen as the party that supports American Indians. This seems likely. Since the 1960s, the Democratic Party has become associated with minority groups in the United States (Carmines and Stimson Reference Carmines and Stimson1989), and this likely includes American Indians. Although historically it was felt that American Indian issues were nonpartisan in the United States (Turner Reference Turner2005 Tyler Reference Tyler1964), the parties have diverged on American Indian issues such that the Democratic Party is more pro-Indian (Conner Reference Conner2014; Turner Reference Turner2005). For example, analysis of roll call votes and committee work in Congress over the last half of the twentieth century finds strong evidence that since the 1970s, Democratic members were more likely to support pro–American Indian legislation than Republicans (Conner Reference Conner2014; Turner Reference Turner2005).
Determining what is pro–American Indian policy is complicated. The way Turner dealt with the complication was to divide American Indian legislation into three categories: land, rights, and funding. Legislation that increased American Indian lands, rights, and funding was coded as more liberal or pro-American Indian. In addition, recent research suggests that most American Indians are Democrats (McClain and Stewart Reference McClain and Stewart2014, 100; Min and Savage Reference Min and Savage2012, Reference Min and Savage2014; Ritt Reference Ritt1979). This is the case for American Indian leaders as well. For example, Stubben (Reference Stubben2006, 168) finds that in the mid-1990s, 61% of American Indian leaders were Democrats and only 10% Republicans. This leads to the second and third expectations: (2) compared with American Indian men, American Indian women are more likely to be Democrats; and (3) the reason for the gender gap is that American Indian women are more likely to prioritize their American Indian identity.
Another reason that a partisan gender gap may exist among American Indians is that the Democratic Party is associated with supporting the concerns of women. As noted earlier, among Americans generally, women support women's/compassion issues more strongly than men because of socialization or experience, and support for these issues helps explain the party gender gap (see, e.g., Chaney, Alvarez, and Nagler Reference Chaney, Alvarez and Nagler1998; Kaufmann and Petrocik Reference Kaufmann and Petrocik1999). In addition, Winter (Reference Winter2010) finds that Americans see the parties in gendered terms, with the Democratic Party being female and the Republican Party male.
For women's/compassion issues to effect the party gender gap among American Indians, there would need to be a gender gap in their support for women's/compassion issues. The little research that has been done on American Indian women's support for these issues has focused on political leaders, and generally it finds that American Indian women care about women's/compassion issues (Prindeville Reference Prindeville2000; Prindeville and Gomez Reference Prindeville and Gomez1999; but see McCoy Reference McCoy1992). Although McCoy (Reference McCoy1992) finds that women leaders' policy priorities do not center on women's issues and are similar to men's priorities, other researchers have found otherwise. For example, Prindeville and Gomez conclude that “American Indian women leaders pursue political goals and promote policies and programs to improve the economic opportunities and social well-being of their people -- particularly to enhance the lives of women, children and the aged” (Reference Prindeville and Gomez1999, 28).
Similarly, after examining testimony of Aboriginal women in Canada, Chiste (Reference Chiste1994) suggests Aboriginal women have some similarities to mainstream women's groups, but they differ in ways that are consistent with supporting American Indian interests. For example, like other women activists, Aboriginal women are highly concerned about domestic violence; however, instead of punishing the perpetrator, they want to heal the family. Or, while they are concerned about reproductive services, they are less concerned about abortion and more concerned about fetal health and adoption of Aboriginals. There is also evidence that women leaders believe that women bring different perspectives to governing that advance women's interests (Prindeville Reference Prindeville2004; Prindeville and Gomez Reference Prindeville and Gomez1999). Many of these studies do not compare women leaders with men leaders, so they do not indicate whether women are more or less interested in these issues than men. In addition, since they interview only leaders, it is unclear what the views of American Indian women generally would be. Nevertheless, this research implies that American Indian women are highly supportive of women's/compassion issues.
The foregoing discussion leads to the fourth and fifth expectations: (4) compared with American Indian men, American Indian women are expected to be more supportive of women's/compassion issues; and (5) the gender gap in women's/compassion issues contributes to the gender gap in party.
RESEARCH DESIGN
In order to test the five expectations, I use two datasets. Two datasets are used because one does not include questions about compassion issues and the other does not include questions that tap American Indians' prioritization of their American Indian identity. An additional value of using two datasets is that I can replicate the findings in the areas of party and ideology.
The first dataset comes from a survey I conducted with Jeanette Mendez that used an opt-in panel sample maintained by Survey Sampling International (SSI). SSI has 15 years of experience conducting online surveys and works with other organizations to ensure the representativeness of its panels. Although opt-in panel studies are somewhat new to the social sciences, they are quickly becoming widely accepted. Ansolabehere and Schaffner's (Reference Ansolabehere and Schaffner2014) examination of opt-in internet surveys found relatively few differences in participants across survey mode on a variety of political and demographic traits. Of particular importance for this project is that panel surveys provide a means to identify a national sample of American Indians.
The survey was conducted in March 2016 and resulted in 301 questionnaires completed by American Indians over the age of 18. In other work, I, along with others, examine the representativeness of this sample by comparing these results with those of American National Election Studies (ANES) and the Current Population Survey (CPS) (Herrick, Mendez, Davis, and Pryor Reference Herrick, Mendez, Davis and Pryor2016). We find that once we weight the data using CPS estimates of American Indians' levels of education and gender, the sample is comparable to the other studies of American Indians in gender, race, voter registration rates, levels of political trust, party identification, and political interest, although the sample respondents have disproportionally high levels of education.
One advantage of the SSI survey is that it was conducted to examine the group consciousness of American Indians, so there are questions that can measure gender differences in American Indian identity. The key question asked, “People identify with many different groups in society, which of the following identity is most important to you?” (your tribe, Native/American Indian, U.S. citizen). Those who said U.S. citizen were coded 0, and those who said their tribe or Native/American Indian were coded 1. This question has face validity.
Another question concerned whether respondents prioritize coethnic candidates: “People can prefer a candidate for a variety of reasons. How important is it for you that a candidate is Native/American Indian?” (very important, somewhat important, or not important). If American Indian women place more value on their American Indian identity than men, they should see voting for an American Indian as more important. This follows from research indicating that an individual's identity affects his or her voting behavior, including preferences for coethnic candidates (Schildkraut Reference Schildkraut2013; Stokes-Brown Reference Stokes-Brown2006). The survey also asked an ANES-style question that results in a seven-point party identification scale (1 = strong Democrat to 7 = strong Republican). Although the survey did not include questions concerning compassion issues, it did ask about ideology with a five-point self-identification question (1 = liberal to 5 = conservative).Footnote 3 The assumption is that liberals are more supportive of compassion issues than conservatives.
The key independent variable in the analyses is the respondent's gender (women = 1, men = 0). Since individual traits other than gender can affect political attitudes and behaviors, several control variables are included in the analyses: education, southern state, and age. In addition, the amount of time respondents spent on a reservation while growing up is controlled for.Footnote 4 This is used as an indicator of unique socializing experiences and conditions thought to affect American Indian identity (Liebler Reference Liebler2010; Liebler, Bhaskar, and Porter Reference Liebler, Bhaskar and Porter2016; Peroff and Wildcat 2002).Footnote 5
The other dataset is the 2014 Cooperative Congressional Election Study from Harvard University (Schaffner and Ansolabehere Reference Schaffner and Ansolabehere2015). This internet survey is highly respected; individuals wanting more information on it can go to its website.Footnote 6 The survey had 414 respondents who described themselves as American Indian. As with the SSI data, the primary independent variable is gender. Unlike the SSI survey, there are no questions asking respondents about the importance of their American Indian identity to them. There are, however, questions that create a seven-point party identification scale (1 = strong Democrat to 7 = strong Republican) and a five-point self-placement ideology scale (1 = very liberal to 5 = very conservative). The CCES data also include responses to several questions that can tap whether men and women differ in their views on women's/compassion issues. These issues are gun control, abortion, support for the Affordable Care Act (ACA), concern for the environment, affirmative action, and gay marriage.Footnote 7 In the analysis of the CCES data, several control variables are used: education, age, church attendance, family income, and southern state.Footnote 8
Table 1 lists the variables from both surveys and gives their ranges, means, and standard deviations. In doing the analysis, ordered logistic regression is used when the dependent variable is ordinal and logistic regression is used when the dependent variable is dichotomous.Footnote 9
Table 1. Means, standard deviations (SD), and ranges of key variables
FINDINGS
Analysis of the survey data identified some significant gender gaps in the political attitudes of American Indians. First, Table 2 reports the results for the SSI survey. It offers general but mixed support for the expectation that compared with men, women are more likely to prioritize their American Indian identity. On the positive side, compared with American Indian men, American Indian women were significantly less likely to identify as U.S. citizens when given the choice between U.S. citizen, tribal member, or American Indian. Based on these results, the probability of a woman identifying as a U.S. citizen was .38 (confidence intervalFootnote 10 [CI] = .26–.50) compared with .58 (CI = .37–.79) for a man. However, women were neither more nor less likely to think it important to vote for American Indian candidates. This may be because most American Indian leaders have been men, so both men and women are equally likely to want to have American Indian leaders.
Table 2. Gender gap in identity, ideology, and party identification: SSI data
Note: See Table 1 for description of variables. Numbers in Parenthesis are standard errors.
*** p < .01; ** p < .05; + p < .10 (using a two-tailed test).
The results from Table 2 also suggest that there was not a gender gap in ideology among American Indians. The coefficient was not statistically significant. However, the data do support the expectation that women are more supportive of the Democratic Party. Women had a .27 (CI = .19–.35) probability of being a strong Democrat compared with .19 (CI = .10–.28) for men. The probabilities of being a strong Republican were .09 (CI = .04–13) and .13 (CI = .07–.19), respectively. Thus, overall, the results support the expectations that American Indian women have a stronger American Indian identity (although they are not more likely to support coethnic candidates) and are stronger Democrats, but they are not more liberal.
Table 3 reports the findings with regard to the CCES data. It indicates that American Indian women were significantly more supportive of women's/compassion issues than their male counterparts. Compared with American Indian men, American Indian women were significantly more supportive of pro-environment policies, gun control, affirmative action, and gay marriage. Similarly, they were significantly less pro-life than American Indian men. The only policy area in which there was no significant gender gap was in support of the ACA. However, even here, women were more supportive, and the size of the relationship neared statistical significance.
Table 3. The gender gap in women's/compassion issues, ideology and party: CCES data
Note: See Table 1 for description of variables.
*** p < .01; ** p < .05; + p < .10 (using a two-tailed test).
To offer some idea of the substantive size of the gaps, I present some probabilities. The probability that a woman strongly supported affirmative action was .09 (CI = .04–.15), and the probability for a man was .06 (CI = .02–.09); the probabilities for strongly opposing it were .35 (CI = .23–.46) and .46 (CI = .36–.57), respectively. The probability that a woman supported gay marriage was .58 (CI = .46–.71), and for a man it was .42 (CI = .30–.54). The probability that a woman scored 0 on the environment scale was .19 (CI = .10–.27), and for a man it was .32 (CI = .20–.45), while the probabilities of scoring 4 were virtually the mirror image: .30 (CI = .22–.39) for women and .17 (CI = .11–.23) for men. The difference was even larger for gun control. The probability of a woman scoring the highest value on the scale was .53 (CI = .40–.66), and for a man it was .24 (CI = .13–.35).
The CCES results, however, offer little support for the idea that there was a gender gap in ideology: the coefficient was not statistically significant. This is consistent with the SSI data, and given the findings concerning compassion issues, it may be that ideology is unrelated to support for compassion issues among American Indians. The CCES and the SSI data are also consistent in their confirmation of a gender gap in party identification. Based on the CCES data the probability that a woman was a strong Democrat was .16 (CI = .09–.23) compared with .09 (CI = .05–.17) for a man. The probabilities of being a strong Republican were .11 (CI = .05–.13) and .20 (CI = .12–.27), respectively.
So far, I have just tested for the existence of the gender gap among American Indians but have not indicated whether the gaps were larger or smaller among American Indians than other Americans. Fortunately, the CCES data have results for whites as well. A table reporting the size of these gaps for non-Hispanic whites (Anglos) can be found in the online appendix. With a couple of exceptions, these results are very similar in that the same relationships were statistically significant and in the same direction regardless of whether respondents were white or American Indian. One exception was that among American Indians, the difference between men and women just neared statistical significance on support for the ACA, whereas it was significant among Anglos. That many American Indians qualify for a government-run health care program may help explain the lack of a gender gap here. A clearer difference between Anglos and American Indians was that among Anglos, there was a clear ideology gap, with women being more liberal, whereas one did not exist among American Indians. Although the size of the coefficients for the gender variable were much larger when examining American Indians than when examining Anglos, interactions between race/ethnicity and gender were not statistically significant (data not shown).Footnote 11
EXPLAINING THE GENDER GAP IN PARTY IDENTIFICATION
The data used for the foregoing analysis can be used to test two potential explanations for the gender gap in party identification: variation in American Indian identity and support for women's/compassion issues. The SSI data can be used to test whether variations in identity effect the gender gap in party. One way to test this is to simply control for respondents' identity (measured as U.S. identity and preference for coethnic candidate). These analyses offer limited support for the expectation that the variation in identity explains the party gender gap. Although the gender coefficient was no longer statistically significant with these variables in the model, it still neared statistical significance, and the size of the coefficient was similar to the one in Table 2. In addition, neither of the coefficients for the American Indian identity variables were statistically significant, although they neared statistical significance.
Another way to examine whether there is something about the American Indian identity that could explain the gender gap in American Indian's party identification is to examine the effects of an interaction between American Indian identity and gender on party identification. I do this by creating an interactive variable by multiplying identity by gender and including it in the model. These results indicate there is a significant interaction effect (see Table 4). However, the gender gap is reverse for those with an American Indian identity. To help make sense of this finding, Table 5 lists the probabilities for different combinations of gender and identity. These results demonstrate that women who identify as American Indian were actually somewhat less Democratic than those identifying as U.S. citizens. For example, the probability of a woman identifying as an American Indian being a strong Democrat was .24, while the probability for a woman identifying as a U.S. citizen was .32. The probabilities of being a strong Republican went from .09 to .06. However, among men, identifying as American Indians increases the strength of their Democratic identity. While the probability of an American Indian man who identifies as a U.S. citizen being a strong Democrat was .26, it was .12 for a man who primarily identifies as a U.S. citizen. The probabilities of a man being a strong Republican went from .08 to .19. These data suggest that among American Indians, the correlates of party identification differ by gender. Only among men does American Indian identity increase the strength of their Democratic identity.
Table 4. Explaining party identification: Women's/compassion issues and identity
Note: See Table 1 for description of variables. Numbers in Parenthesis are standard errors.
*** p < .01; ** p < .05; + p < .10 (using a two-tailed test).
Table 5. Estimated probabilities for party identification
Note: Numbers in parenthesis are confidence intervals.
Given these interesting findings, I replicate these findings using variation in time spent on a reservation as a measure of identity. American Indians who grew up on a reservation tend to have higher levels of American Indian identity. Geographic location is thought to have a significant effect on American Indian identity (Liebler Reference Liebler2010; Liebler, Bhaskar, and Porter Reference Liebler, Bhaskar and Porter2016; Peroff and Wildcat 2002). Many cultural traditions are tied to a location. In addition, living on Indian territories likely means people come in contact with their heritage in ways that reinforce their American Indian identity. A relationship may also exist between living on tribal lands and American Indian identity because people who want to be close to their heritage are more likely to choose to live on American Indian land, or their parents had strong ties and chose to live on Indian lands (Liebler Reference Liebler2010).
To conduct these analyses, I created an interaction variable by multiplying gender by percent of time spent on the reservation as a child and added this variable to the model. These results support the conclusion that the effects of identity on party are gender specific. For example, about 30% of women were strong Democrats and about 7% to 8% were strong Republicans, regardless of how much time they spent on the reservation as children. However, for men who spent no time on a reservation as a child, the probability of being a strong Democrat was .16 (CI .08–.24); it was .59 (CI .27–.92) for men who spent at least 75% of their childhood on a reservation. The probabilities of a man being a strong Republican went from .05 (CI = .08–.22) to .02 (CI = .01–.06).Footnote 12
Growing up on a reservation and having a strong American Indian identity increase the likelihood that men are Democrats but have an inverse effect on women's Democratic identity. This means that women's higher levels of American Indian identity cannot explain why women are more likely to support the Democratic Party. Another explanation for the gender gap in party has to do with the gender gap in women's/compassion issues.
Although the SSI data did not include questions about compassion issues, the CCES data did. Analyses of these data found that when compassion issues were included in the model explaining party identification, the gender variable was no longer statistically significant (see Table 4). In addition, when compared with the size of the gender coefficient in Table 3, the coefficient here was clearly smaller, and the coefficients for most of the compassion issues were statistically significant and in the expected direction. Thus, it appears that variation in support for women's/compassion issues largely explains the gender gap in party identification.
CONCLUSION
This article set out to see whether there are gender gaps among American Indians that parallel those found among Americans generally. It found that the gender gaps are similar to those among Americans generally in some ways, but in other ways they are not. Like Americans generally, among American Indians, women are more likely to be Democrats and more likely to support compassion issues, and the gender gap in compassion issues helps explain the gender gap in party. However, unique to American Indians is that there is not an ideology gender gap and American Indian women identify more strongly as American Indians (or less strongly as U.S. citizens). This finding confirms the results of others who have examined Americans with a double minority status (female and a racial/ethnic): being female does not diminish their identity as racial/ethnic minorities but rather enhances it (Simien Reference Simien2005; Simien and Clawson Reference Simien and Clawson2004). Also unique to American Indians is the effect that identity has on the gender gap in party. While among women, American Indian identity has a negligible effect on party identification, among men, those with a strong American Indian identity have a stronger Democratic identification.
This last finding is intriguing given concerns about how to determine who is a “real” American Indian (McCool Reference McCool1982; Peroff and Wildcat 2002; Strum Reference Strum2010). For example, Strum (Reference Strum2010) notes that there are Americans who start to identify as American Indians as adults. These “race shifters,” as Strum identifies them, discover they have some American Indian ancestry and alter their identity. Although this ancestry may or may not be substantiated, many attempt to reclaim their American Indian culture. However, they likely appear white, grew up as white, and have lived as white people up until that point. Thus, race shifters are unlikely to differ from other white Americans in their political attitudes. This would help explain why there would be different political attitudes among self-identified American Indians depending on the strength of their American Indian identity, with those with a weak American Indian identity being similar to white Americans.
However, this does not explain why American Indian identity has gender-specific effects on party identification. Assuming that party differences on racial/ethnic issues and images affect party identification, it may be that among American Indian men, those who prioritize their American Indian identity are more likely to see themselves as a racial/ethnic minority. Among these individuals, the parties' differences on racial/ethnic matters have a stronger effect. But for women, because of the gendered nature of parties and their differences in compassion issues, they are inclined to be Democrats without an American Indian identity.
The findings also offer little support for Harell and Panagos's (Reference Harell and Panagos2013) colonialism hypothesis, since the identity gender gap does not explain the party gender gap. One interpretation of this finding is that Harell and Panagos (Reference Harell and Panagos2013) overestimated the effects of colonialism on the partisan gender gap, which is possible given different understandings of tribal history and traditions. Another possibility is that attitudes toward American Indians play more of a role in explaining the gender gap in party in Canada than in the United States. This, too, is possible since the New Democratic Party is more clearly aligned with Aboriginal interests than the Democratic Party in the United States is with American Indians. It is also possible that the measures of American Indian identity did not directly test the colonialism hypothesis. Unfortunately, I lack the data to test these possibilities. Nevertheless, these findings do not support the expectation that gender gap in identity explains the gender gap in party in the United States.
These results, coupled with the results of many other studies, confirm a tendency for women to hold different political positions than men. This suggests the importance for women to be fully engaged in politics. Without women's full engagement, their unique positions will be underheard. However, this study also notes some differences among American Indians in the gender gaps confirming the importance of considering the intersectionality of gender with other social/demographic groups. Thus, not only is it important that women participate in large numbers but also that the group of women who participate is diverse.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S1743923X17000344
