Which of the following is a factor that is least likely to influence whether a person votes or not

This study uses national survey data in federal election years during 1996-2004 to examine voter registration and voting. It shows that racial/ethnic disparities in socio-economic resources and rootedness in the community do not explain overall group differences in electoral participation. It contradicts the expectation from an assimilation perspective that low levels of Latino participation are partly attributable to the large share of immigrants among Latinos. In fact net differences show higher average Latino participation than previously reported. The study focuses especially on contextual factors that could affect collective responses of group members. Moving beyond past research, significant effects are found for the group's representation among office holders, voting regulations, and state policies related to treatment of immigrants.

Keywords: Political Participation, Race, Registration, Voting, Voter ID, Immigration, Group-Resources

An established research tradition demonstrates the strong effects of an array of personal background characteristics on political participation, such as age, education, and residential stability. Such characteristics, which are often described as indicators of “resources and rootedness,” are fungible: they provide a boon to political participation regardless of group membership. This paper focuses instead on how group membership, defined by race, ethnicity, and nativity, structures political participation above and beyond such personal characteristics. Unlike most previous studies, we use a large national sample, compare four major racial/ethnic groups, and test for differences between immigrants and natives in every group, including non-Hispanic whites. This design allows us not only to identify persistent differences by race but also to document the unexpected high level of participation for Latino immigrants, after controlling for other factors. The analysis also draws attention to several dimensions of the political context—which we term the “group context of participation.” In addition to examining measures that have been used in past research, this is the first nationally representative study to show evidence of increased participation for blacks where there are co-ethnic officials (i.e., political empowerment) and evidence that state voter ID policies dampen turnout for blacks and Hispanics and for non-Hispanic whites.

A long-standing tradition of research seeks to explain group differences in participation – especially the typical finding of lower participation by Latinos and Asians – through differences in the resources and community rootedness of group members (Uhlaner, Cain and Kiewet 1989; Leighley 2001; Leighley and Vedlitz 1999; Antunes and Gaitz 1975; Ramakrishnan 2005; Ramakrishnan and Espenshade 2001; Martinez 2005). For example, Citrin and Highton (2002) argue that low Latino voting in California can be accounted for by Latinos’ lower citizenship rate, their relative youth, and their lower socioeconomic status. But researchers have been particularly hard-pressed to come up with explanations for why, given their more favorable socioeconomic position, Asian-American citizens still show depressed voting (Verba, Schlozman and Brady 1995; Citrin and Highton 2002). The standard individual-level model argues that political participation is rewarding and comes with few costs for persons with the resources of time, political experience, information, education, money, and knowledge (Verba et al 1993). Scholars have also found that people who are rooted in their local context are more likely to participate in politics (Putnam 1995). Rootedness in this sense is typically measured through indicators such as older age, marriage, residential stability, larger family size, and living with children (Bueker 2006, Rosenstone and Hansen 1993, Ramakrishnan and Espenshade 2001, Highton 2000, Timpone 1998, Wolfinger & Wolfinger 2008).

Asians have higher than average resources and they appear to be strongly rooted in their residential communities. Something else about “being Asian,” not reducible to individual-level processes, seems to make a difference. Scholars have offered several suggestions about this group-specific effect, pointing to Asian-Americans’ geographic dispersion across the country, cultural factors such as a “community norm to avoid political involvement or the learned attitude that electoral politics are a waste of time,” lack of political leadership, and experience of discrimination in the U.S. (Uhlaner, Cain and Kiewet 1989, p. 217).

From the perspective of assimilation theory (Alba and Nee 2003), a likely suspect is recent immigration. Both Asians and Latinos – because they include many recent immigrants – are expected to be incorporated slowly into mainstream society. But the second and later generations of group members, reflecting their economic, cultural, residential and linguistic assimilation, should progressively participate on a more equal footing in key institutions, including civic affairs (Skerry 2004). Some researchers do report that foreign-born persons (regardless of time in the U.S.) are less likely to vote than persons in the second generation (Cho 1999; DeSipio 1996). But there is also contradictory evidence (Lien 2004). In fact it has been suggested that those who choose to become citizens and therefore are eligible to register and vote are an especially motivated, self-selected subset of immigrants who are therefore more likely than natives to participate in the political process (Segal 2002, Barreto et al. 2005, Pantoja et al. 2001, Cassel 2002, Garcia and Arce 1988). Ramakrishnan and Espenshade show that generational differences in participation vary by racial group (Ramakrishnan and Espenshade 2001; Ramakrishnan 2005; Lien 2004). Only for Asians is there a linear progression of increasing participation by generation. By contrast, there is a decline across generations among Latinos. And among whites, the 2nd generation is more likely to vote than either of the other generations.

We expect to find that group differences cannot be accounted for by compositional effects (reflecting individual-level resources or rootedness), and that generational differences among members of largely immigrant groups will not conform to the assimilation model. This expectation leads us to focus on collective factors that might lead to racial and ethnic differences. In the literature on immigrants, the key concept is the “context of reception” that a newcomer encounters (Portes and Rumbaut 1996). We apply this same concept to all racial/ethnic groups, using the more general term “group context of participation.”

Collective conditions could either depress or enhance participation, perhaps especially for members of minority racial and ethnic groups. The obstacles to voting by African Americans continue to be the grounds for litigation and policy debate. More broadly the political environment may be perceived as discouraging or even threatening (see Pantoja, Ramirez and Segura [2001] on the threat perceived by Latinos in California in the mid-1990s). On the other hand, in the wake of the voting rights movement, blacks have been found to have distinctively high rates of voting, despite deficits in socioeconomic and other resources (Tate 1991, 1993). This phenomenon has been attributed to group consciousness and mobilizing institutions specific to African American communities such as the black church (Harris 1994; Brown & Brown 2003) and civil rights groups like the NAACP, Urban League, or the Southern Christian Leadership Conference (Antunes and Gaitz 1975). Similar processes that could channel group-relevant information, shape group political consciousness, encourage the formation of racial or ethnic group identity, or invite group-specific mobilization have been discussed for other groups (Timpone 1998; Cho 1999; Jones-Correa 2005; Ramaskrishnan and Espenshade 2001; Bueker 2005; Leighley 2001; Gay 2001; Cho et al 2006).

One hypothesis derived from this reasoning that we can test directly is “political empowerment” (Bobo and Gilliam 1990) – the argument that voting is encouraged by increased numbers of black elected officials, actors who can promote mobilization, demonstrate its efficacy, and enhance feelings of solidarity and pride. Washington (2006) found a positive impact of black candidates for the House of Representative on voting among blacks, but a countervailing increase in white turnout voting against the black candidate. Pantoja and Segura (2003) also found modest support for this hypothesis in their analysis of 1997 data, which found that greater numbers of Latino legislators slightly decreased political alienation among Latinos in California and Texas. We test whether political representation of this sort affects voter registration and voting not only for blacks but also for Latinos and Asians.

We consider two other aspects of the political environment that are particularly likely to affect Latinos and Asians and have not previously been studied at the national level. One factor is general public attitudes toward immigrants. Van Hook, Brown, and Bean (2006) found higher rates of naturalization in states with more positive attitudes (i.e., more welcoming states). We expect similar effects for political participation by Latinos and Asians. A second factor is the degree to which state policies restrict immigrants’ access to welfare services that are typically considered part of the welfare safety net. Van Hook showed that immigrants were more likely to become citizens in states with more restrictive policies, suggesting an instrumental reason to naturalize in order to gain better access to services. As applied to participation, however, the effect could be to promote registration and voting, in the same way that more welcoming public attitudes may do.

The black experience of political marginalization by discriminatory voting policies calls attention to other ways that the group context of participation could affect other groups. First, participation could be enhanced by voting rights legislation designed to mitigate or remove obstacles to voting by linguistic minorities. Up to now analyses of these provisions have yielded mixed results. Jones-Correa (2005) reported from his analysis of the Current Population Survey from 1996 and 2000 that Asians and Latinos were more likely to vote in states that offer voting and registration materials in the respondents’ native languages, and that this was especially true for Latinos (both immigrants and native-born) while the effect varied among Asian groups. By contrast, Ramaskrishnan and Espenshade (2001), in their analysis of CPS data from 1994, 1996, and 1998, found that minority rights provisions did not significantly impact voting among first generation Latinos.

Aside from regulations specifically oriented to minorities, the political system's general “rules of the game” can be more or less restrictive, thus encouraging or discouraging participation. Here we examine three aspects of these rules. First is regulation of voter registration, specifically how long before an election a person must register in order to be eligible to vote. Past studies have had mixed results. Using the CPS, Jones-Correa (2001) found no significant impact, but Burden et al (2010) found a strong negative effect of restrictive policies. Studies using the NES (Timpone 1998; Leighly and Nagler 1992) showed that individuals in more restrictive states with earlier closing dates for registration were slightly less likely to be registered. Second is provision for absentee voting. Ramakrishnan and Espenshade (2001) and Jones-Correa (2001) both found that higher restrictions on absentee voting decrease the probability of voting by registered voters. In places where more people are allowed to mail in ballots, there is increased voter turnout. But Burden et al (2010) found that absentee voting (or other forms of potential “early voting”) reduces turnout.

Recently a third type of regulation has become prominent in policy debates: the requirements for voters to show identification prior to voting. Provisions for photo identification are now widely discussed, and came under review by the U.S. Supreme Court during its 2007-2008 session. A report by researchers at the Eagleton Institute of Politics, Rutgers University, and the Moritz College of Law, Ohio State University (2006) found that strict voter identification requirements depressed voting turnout in 2004, and that this effect was especially pronounced for minority voters (Eagleton Institute 2006, but cf Muhlhausen and Sikich 2007, Burden et al 2010).

Early research on political participation concentrated on blacks and whites. As the Hispanic and Asian populations have grown, more attention has been given to immigrant minorities, but many new studies have looked at selected states or regions, especially areas of high Latino population like California, Florida, and Texas (Uhlaner et al 1989; Portes and Mozo 1985; Shaw, de la Garza and Lee 200; Cho 1999; Pantoja, Ramirez and Segura 2001; Antunes and Gaitz 1975; Barreto 2005). Our purpose requires us to use a nationally representative sample that includes members of all of the major racial/ethnic groups (like Bass and Casper 2001; Ramakrishnan 2005; Bueker 2005). We rely on the Current Population Survey in the years 1996, 1998, 2000, 2002, and 2004. In these years the November survey included a voting and registration supplement. Since the CPS also contains questions that allow identification of the nativity and citizenship status of respondents, it is well-suited for analyses of the registration and voting of immigrants, members of the “second generation,” and U.S.-born persons of native parents. When data from all of these years are pooled together the CPS includes adequate samples of naturalized immigrants for each of four major racial/ethnic categories. Other common sources of political participation data, such as the National Election Survey, contain much smaller samples of foreign-born members of non-white racial groups.

The CPS also includes information about the geographic location of households, which makes possible examination of contextual effects that we add to the dataset from census and other sources. The state of residence is provided for all cases. The smallest identifiable geographical unit of residence is the Metropolitan Statistical Area (MSA/PMSA). For this study, contextual variables constructed from the Census 2000 summary tape files and other sources were merged with CPS data based on state or MSA/PMSA of residence. Recent changes in the CPS and Census definition of metropolitan boundaries create challenges for consistent construction of metropolitan variables over time. MSAs were redefined by the Office of Management and Budget in 2003, and as a result the definitions used in the 2004 CPS data differ from those used in prior years. To deal with this change, the 2004 MSA definitions have been recoded here in a way that would correspond as closely as possible to the MSA definitions from earlier CPS years, as well as to definitions used in the Census 2000 summary files.

Table 1 provides a summary of variable definitions from all data sources. The key outcome variables are self reports from the CPS of voting or being registered for the November election of the year of data collection. The specific question about voting is: “In any election some people are not able to vote because they are sick or busy or have some other reason, and others do not want to vote. Did (you/name) vote in the election held on Tuesday, November _?” This question is only asked about citizens aged 18 and above. A follow-up question is asked of those who were eligible to vote, but did not report having voted: “(Were you/Was ‘name’) registered to vote in the November __ election?”

List of variables included in the analysis

The wording of these questions is designed to diminish stigma associated with non-voting or non-registration (Bueker 2006, p. 21). Presser, Traugott and Traugott (1990) have shown that the CPS contains less misreporting than other surveys such as the NES. Ramakrishnan (2005, p. 22) validated data for white immigrants in the CPS with comparable data collected from the National Election Survey (NES), and showed that 2nd generation whites under-reported their voting in the CPS compared to other generations. Accordingly the effects we find for whites, if anything, underestimate the differences between first and second generation whites. Other studies of voting behavior data employing validated data, such as the three-state study conducted by the Thomas Rivera Policy Institute in 1996 (cited by Ramakrishnan 2005), found no significant differences by nativity or generation status in the reliability of self-reported voting. However, Shaw, de la Garza, and Lee find evidence of Latino over-reporting in a validated study of voting in Texas, California and Florida (2000).

Race and ethnicity are represented by four broad categories constructed from two different questions posed in the CPS basic survey, the race question and the Hispanic origin question. The race question in the basic CPS allows respondents to select one of five racial categories, with the option to choose ‘other’ or ‘something else’. Specifically, respondents are asked to answer the question “What is your race? Are you White, Black, American Indian, Aleut or Eskimo, Asian or Pacific Islander or something else?” Unlike the 2000 decennial census, respondents were not given the option to select multiple race categories except in the 2004 CPS. A separate CPS question about Hispanic origin or descent allows individuals to identify as: Mexican American; Chicano; Mexican; Puerto Rican; Cuban; Central Or South American; other Spanish; All Other; or Don't Know.

We combined data from these two questions to construct four broad race/ethnic categories that structure much of our analysis: non-Hispanic white, non-Hispanic black, Asian/Pacific Islander and Latino (because of their small number “other” race respondents have been excluded from the sample). A respondent who indicated any category of Hispanic origin is treated as “Latino” regardless of race. Non-Hispanics are categorized as white, black, or Asian/Pacific Islander based on the race question. Information from the 2004 CPS was recoded to match these categories as closely as possible, as was the 2000 census data used in constructing contextual variables.

In these analyses race/ethnicity is treated in conjunction with generation in the U.S. based on information contained in the Basic CPS about birthplace and parental birthplace. Those born outside of the U.S. are considered 1st generation. Those born in the U.S. with at least one parent born outside of the U.S. (in a non-U.S. territory) are classified as members of the 2nd generation. The remaining 3+ generation cases are individuals born in the U.S. whose parents were also born in the U.S. (or in U.S. territories). We also attempted to use country of origin to better identify immigrants’ ethnicity, but we found that the sample could not support such fine-grained distinctions.

Table 2 provides a starting point for the analysis based on our pooled sample of citizens aged 18 and above in the Current Population Survey conducted in the five national election years during 1996-2004. They show a marked disparity in voting between Latinos and Asians, on the one hand, and blacks and non-Hispanic whites, on the other, reflecting what some authors call a “turnout gap” (Citrin and Highton 2002). This gap is due mainly to differences in voter registration (10-13 points), with an additional deficit in voting by Latino registered voters (5 points less than Asians and 7 points less than blacks and whites). The table suggests that racial and ethnic differences are not due to variations in group composition by nativity. In fact the influence of nativity in its own right appears to be small except among Asians, for whom immigrants and the second generation children of immigrants are 10-12 points less likely to vote than those in the third and later generations. The opposite effect of nativity is found among Latinos, with modestly higher voting turnout (due to greater likelihood that registered voters actually vote) among foreign-born citizens than in the 2nd and 3+ generations. This finding is a first example of effects that turn out to be contingent on group membership.

Registration and Voting by Race/Ethnicity and Generation in U.S., 1996-2004 (citizens aged 18+)

Our strategy is to elaborate upon these initial findings in multivariate analyses where registration and voting are treated in turn as dependent variables. We present one set of models in which members of all four racial/ethnic groups are pooled, which allows us to identify the group differences net of variations in other individual and contextual influences, and to show how the population as a whole is affected by contextual factors like electoral rules that are not group-specific. Surprisingly, one key result is to upend the established hierarchy of group participation, placing first generation Latinos above non-Hispanic whites. We then estimate separate models for each group, adding variables representing important elements of their context of participation. The contribution is intended to be two-fold: demonstrating the importance of contextual conditions and showing which of their effects are group-specific.

The multivariate models include several indicators of resources and rootedness deemed important by the standard voting behavior literature. These are listed in Table 1. Educational attainment, family income, and home ownership are the socioeconomic status indicators. Demographic variables include age, marital status, number of children under 18 in the household, gender, and residential mobility (years at the current address). (Note that, because education is reported only for persons aged 25 and above, the age categories of 5-15 and 16-24 must be interpreted both as categories of age and as persons without a reported education.) Another variable was constructed from information on household members: the number of children under age 18 living in the household.1 All indicators of greater resources and rootedness (higher education and income, home ownership, older age, being married, female and having kids) are hypothesized to lead to increased political participation.

To examine the assimilation hypothesis we constructed three categories of generation in the United States: first generation immigrants (born abroad, not a native citizen), second generation (born in the U.S. and at least one parent born abroad), and 3+ generation (born in the U.S. and both parents born in the U.S.). In the pooled analysis, generation is combined with race/ethnicity to create a series of 12 dummy variables. Another related indicator is linguistic isolation. This is the only language measure available in the November CPS, and it indicates whether a person lives in a household where only Spanish is spoken. We used this household variable in our models for Latinos, based on the assimilation hypothesis that linguistic isolation would reduce participation.

A weakness of the CPS questionnaire for our purpose of studying group differences in participation is the lack of explicit indicators of group consciousness or other racial/ethnic or political attitudes. We therefore turn to other sources to create variables that reflect various aspects of the context of participation.

We use Census 2000 data (Summary Files 1 and 3) to capture key information about demographic context at the level of the metropolitan region.2 The year 2000 is at the midpoint of our 1996-2004 CPS data, and the Census is the most reliable source for metropolitan-level population variables. The metropolitan variable reported here is the ratio of the median household income of each racial group in the MSA to the median household income of non-Hispanic whites. These data are derived from Summary File 3, where the household is categorized by the race/ethnicity of the household head. Data for non-Hispanic whites, blacks, and Asian/Pacific Islanders refer to households headed by a person who listed only that race. This is a measure of relative affluence or poverty that could show whether (net of their own socioeconomic standing) members of relatively poorer groups would participate less.

Measures of political context were drawn from a variety of sources. The test of the empowerment thesis for Latinos, Asians and Blacks is based on data on office holding at the level of metropolitan regions. Information about Latino co-ethnic office holding is from the 2000 directory prepared by the National Association of Latino Elected and Appointed Officials (NALEO). The NALEO directory includes elected and appointed public officials at all levels, and provides their official postal address (NALEO 2000). Black co-ethnic official data, including only elected officials (at Federal, State and local levels), was provided by the Joint Center for Political and Economic Studies. Data on Asian elected and appointed officials was obtained from the National Asian Pacific American Political Almanac 2001-2002 provided by the UCLA Asian American Studies Center. We linked zip codes of office holders in these files to the Metropolitan Statistical Areas identified in our Current Population Survey Data file, drawing on the linkage between zip code areas and MSAs developed by the Missouri Census Data Center Geographical Correlation Engine (Missouri Census Data Center 2007).

We examined several ways to code these data, treating the total number of co-ethnic political officials within each MSA as an interval scale or as a set of up to ten categories to test for non-linear effects. We found that a simple dichotomy adequately reflects the observed relationships: 0-5 (the reference category) or more than 5 representatives. The measure for all groups is for the year 2000, midway in the 1996-2004 period that we study.

Other indicators of the context of participation are measured at the level of states. Some of these are rules about voting and registration that could affect political participation by members of all racial/ethnic groups. A key variable of this type that has rarely been studied before is voter identification policy, which is hypothesized to reduce participation. The question is whether or not a respondent lives in a state requiring prospective voters to show some form of personal identification before casting a ballot. Forms of identification required or requested may include photo or non-photo ID. There also may be variation in policies that govern what a poll worker should do in case a prospective voter does not have an ID. States that do not require or request any form of ID have varying requirements and procedures for verifying the identity of potential voters, including having them state their name, sign their name, or matching a signature with a signature on record with election officials. We draw here mainly on the classification of policies by state from reports published by the Election Reform Information Project (2002 and 2006) conducted in conjunction with Electionline.org and The Constitution Project and the Eagleton Institute and Moritz College of Law (2006), with corrections from a report of the Heritage Center for Data Analysis (Muhlhausen and Sikich 2007). We use information about state level policies in 2000 as indicators of state level policies in 1998 and 1996, since significant legislative and political activity to shift voter ID policies took place only after 2000 (Election Reform Information Project 2006).

The Election Reform Information Project and the Eagleton Institute both utilized five original categories of voter ID policy (though the reports used slightly different classification systems). We collapsed these into a simple dichotomy based only on the maximum requirement: does the state request documentary evidence at the polls of the prospective voter's identification? These categories do not take into account other procedures (sometimes called “minimum requirements”) to be employed when voters do not meet “maximum” identification requirements, nor do they account for rules specific to absentee voters, election-day registrants (where allowed), first time voters or those who have registered by mail (see Eagleton and Moritz 2006 for discussion of minimum vs. maximum requirements). They also do not reflect variation in voter identification policies that may exist at local (sub-state) jurisdictions.

Two additional indicators of electoral rules are drawn from Hansen and the Task Force on the Federal Election System (Hansen 2001). These are the availability of early voting and the flexibility of absentee voting policies, both of which are expected to facilitate participation. Tucker and Espino (2006) also provide information about the number of counties within each state that are covered by bilingual voting ballot provisions, as mandated by the Federal Voting Rights Act, potentially increasing participation by Latinos and/or Asians. We used these data at the state level to construct three categories reflecting the amount of “coverage” by these provisions: 1) no coverage (meaning no counties are required to comply with the bilingual ballot provisions); 2) some coverage (meaning that some counties offer bilingual ballots by law); and 2) full coverage (meaning that all counties offer bilingual ballots by law). One qualification of this variable is that our source does not identify which specific language groups were covered by the requirement. The ballot variable indicates the significant presence of non-English speaking populations and a political-institutional climate formally designed to promote linguistic minority participation.

Additional variables measure features of the political context at the state level that are particularly relevant to immigrants. First is a measure of “immigrant receptivity” that is intended to capture public attitudes toward immigrants. The original scale was developed for metropolitan areas by DeJong and Tran (2001) and DeJong and Steinmetz (2004), using data from the General Social Survey in the years 1995 to 1997. This measure was expanded to the state level and converted to standardized scores by Van Hook, Brown, and Bean (2006). Second is a measure of the “social policy safety net” for immigrants, an index developed to capture the spread of social services and welfare available to non-citizens by Zimmerman and Tumlin (1999) in a working paper of The Urban Institute. This index is compiled to capture immigrant access to benefits in 12 separate social policy areas, including post-1996 access to TANF, Medicaid and Food Stamps. The original measure categorizes states into four levels of safety net availability. We collapse these into two categories, the least restrictive states (where the safety net is “most” and “somewhat” available) vs. the most restrictive (“less” and “least” available).3

The election year is included as dummy variables to account for the historical particularities of each electoral contest. Presidential election years (1996, 2000 and 2004) are expected to have higher turnout. One additional measure of the impact of group-specific mobilization, or group specific reaction to racial discrimination, is a dummy variable identifying 1996 cases in which the respondent lived in California. California's Proposition 187 was enacted in 1994 to deny undocumented immigrants’ access to social services, health care, and public education. This dummy variable is intended to test speculation that anti-immigration legislation would have a mobilizing effect on naturalized citizens in the subsequent 1996 election (Ramakrishnan and Espenshade 2001) and that, in general, 1996 mobilizing efforts targeted at Latinos were more extensive than in previous years (Shaw, de la Garza, and Lee 2000).

Finally, we include variables indicating whether the person lives in an identifiable metropolitan area, lives in a non- metropolitan region, or lives in a metropolis for which we could not match the CPS's MSA code. These dummy variables are included simply as control variables, and we do not seek to interpret these coefficients.

We report the results of logit models of both voting and registration.4 The analysis is conducted with pooled data from all five years (1996, 1998, 2000, 2002 and 2004). Dummy variables indicate the year in which the respondent was interviewed. We select only potential voters: citizens (including both U.S. born and naturalized foreign born) aged 18 and above in the year of the survey. In order to correct for autocorrelation between household members, we then randomly select one person per household (for the group-specific models this is done after selecting by race/ethnicity). The sample for voting analyses included only individuals who report having registered.

We first present results for a model that combines persons from all racial/ethnic groups (non-Hispanic white, non-Hispanic black, Asian and Latino). We then present separate models for each race/ethnicity. In all analyses, cases are weighted by the CPS “second stage/final step” weight (PWSSWGT). Because these weights artificially inflate the overall sample size, we divide them by 1,000 so that cases are weighted properly in relation to one another but the overall sample size is close to the original unweighted number.

There are two other concerns with our estimates from logistic regression models. First, it is necessary to correct the standard errors of coefficients for contextual variables to account for the clustering of sampled cases by metropolitan area or state. We explored two procedures for this correction: the Huber/White procedure (the robust cluster option in Stata) and a full multilevel model. Some results are the same using either option, but more consistent effects for absentee voting and the welfare safety net are found with the Huber/White approach. Because these two contextual variables are substantively important, we report these coefficients but with the proviso that the same effects are not found in a multilevel model. The second potential problem arises from the large number of predictors that are defined at the state level. With only 51 cases (50 states plus the District of Columbia), it is desirable to limit the number of these variables. We accomplish this by a stepwise procedure, beginning by entering each state-level variable into the model by itself (with the full range of individual-level predictors), then examining models with two, three, or more of these variables in combination. The most robust results are for voter ID requirements and group political representation. The models reported below include only those predictors that had significant effects in a consistent direction after these two variables were entered.

The overall levels of registration and voting by race/ethnicity and generation were presented in Table 2. We turn now to multivariate analysis to discover whether differences by race/ethnicity persist after introduction of controls, what other resource, political context, and group-specific factors affect these outcomes, and how these may vary by race/ethnicity.

Table 3 presents models for registration and for voting (among registered voters only) in which persons of all racial/ethnic categories are pooled together and with the Huber/White correction. The models in Table 3 do not include variables that are defined only for specific groups (the Spanish language variable, relative group income level, and political representation). Among the variables for combinations of race/ethnicity and nativity, non-Hispanic whites in the 3+ generation are taken as the reference category. Separate dummy variables identify 1st and 2nd generation non-Hispanic whites, as well as blacks, Latinos, and Asians of each generation.

Logistic Regression Model for Registration and Voting: All Races, 1996-2004

Net of other factors, are there racial and ethnic differences, and is there a significant difference in political participation across generations? Let us first compare these groups in the 3rd generation. With the non-Hispanic white 3rd generation as the reference category, 3+ generation blacks are substantially and significantly more likely to register and to vote. The coefficients are large, representing odds of both registering and voting that are more than 50% higher than those of whites. Latinos in the 3+ generation are moderately but significantly less likely to register and vote than whites. Asians in this generation are much less likely to register (with odds only two thirds as high as whites). Once registered they are not significantly less likely than whites to vote, but their coefficient (−.095) is similar to the coefficient for Latinos (−.151). Hence among members of the “native” generation and controlling for other variables, the racial/ethnic hierarchy of participation has blacks at the highest level, followed by whites, then Latinos and Asians.

These models do not provide significance tests of differences between groups within the 1st and 2nd generation, but the size of the coefficients offers a guide to how they stand. In the 2nd generation the results are generally consistent with those for the 3+ generation. Whites are most likely to register (b=.137), followed by blacks (b= −.067), Latinos (b= −.224), and Asians (b= −.616). For voting the relative positions of blacks and whites are switched, again followed by Latinos and Asians.

It is in the 1st generation that there is a more surprising result. All coefficients for registration are negative, meaning that first-generation immigrants of all racial/ethnic background are less likely to register than are third-generation whites. The least negative coefficient is for blacks, followed by Latinos, whites, and Asians, in that order. For voting the relative ranking is the same. The positive coefficients for first-generation blacks (.262) and Latinos (.214) mean that these immigrants are even more likely to vote (once registered) than third-generation whites. The white coefficient is negative (−.216), and Asians again have the strongest negative coefficient.

These results in Table 3 can also be read in terms of the effect of generation within each group, but a better test is provided in the group-specific models in Tables 4-​5 where additional contextual variables are taken into account and all parameters are allowed to vary across groups. Table 4 presents results for every group for registration; Table 5, for voting. The results are much more complex than anticipated by assimilation theory.

Logistic Regression Models for Registration, 1996-2004

Logistic Regression Models for Voting, 1996-2004

For registration (Table 4), blacks, Latinos, and Asians show the expected “assimilation” pattern in which the 3+ generation is most likely to register and the 1st generation is least likely. But among whites there is what could be called a “2nd generation surge.” The 2nd generation is most likely to register, while the 1st generation is least likely. This same 2nd generation surge is found for white voting. For Asian voting, the assimilation pattern is repeated. Among blacks there is no linear generation effect (there is some indication of a second generation surge for voting but no significant difference between 1st and 3rd generation). The generation effects among Latinos are unusual but consistent with what was shown in Table 3 – here it is the 1st generation that is most likely to vote, while there is no significant difference between the 2nd and 3+ generations.

These findings show interaction effects between race/ethnicity and generation that support no simple theoretical model. In most comparisons, apparently whites’ overall parity with blacks that we saw in Table 2 is due to their advantages in other background characteristics that have been controlled in the multivariate models. All else equal, blacks participate more in terms of both registration and voting than do whites. Asians’ higher overall participation than Latinos is due to the same compositional differences. But all else equal, Asians in every generation are less likely to register than Latinos. If registered, they are less likely than Latinos to vote except in the 3+ generation.

Also relevant for the assimilation perspective are the findings for Latinos on linguistic isolation. We find that those who live in Spanish-speaking households are less likely to register; but surprisingly they are substantially more likely to vote.

The pooled models in Table 3 also offer evidence of the overall effects of the context of participation. None of these measures of context, except for voting year, show significant effects for registration in the all-races model (as noted above, for the sake of parsimony Table 3 omits a number of non-significant predictors). Voter ID requirements sharply reduce the odds of voting. People in states that allow absentee voting are substantially more likely to vote (a 25% increase in the odds). And residents of states with a stronger immigrant safety net are also significantly more likely to vote.5

The election year also has strong effects. Registration and voting are both lower in 1998 and 2002, the non-Presidential years, than in 2000. Considering only the Presidential election years of 1996, 2000, and 2004, there is a clear upward trend in registration and voting. California in 1996 had a significantly higher registration than in 2000, but lower voting turnout – not consistent with the expectation of heightened participation in that year with highly publicized debate on immigrant issues.

Tables 4 and ​5 allow us to see how the context of participation may differentially affect members of different racial-ethnic groups, and also to introduce contextual variables that are specific to each minority group.

The black, Latino, and Asian models include a measure of relative group economic position in the metropolitan context: the group's median income in comparison to that of non-Hispanic whites in the metropolis. Controlling for individuals’ own socioeconomic status, we expected participation of minorities to be depressed in metropolitan areas where their group's income (as a ratio to whites) is low. However, strikingly, for blacks, Latinos and Asians the coefficient for registration is negative. And there are no significant effects on voting.

Similarly the measure of co-ethnic political representation is included only in the models for the three minority groups. The effect of having more than five co-ethnic public officials in the metropolitan area is positive and very strong for blacks, resulting in an increase of over 30% in registration and over 40% in voting. There are no significant effects for Latinos (though in the multilevel model there is a positive effect on Latino voting). For Asians, there is an unexpected negative effect on registration and no impact on voting. One possible interpretation of the positive effects for blacks is that higher levels of voting by group members contribute to greater co-ethnic representation (a reversal of the causal order). This interpretation is buttressed by the fact that higher residential isolation (which implies that minority voters are likely to be more concentrated in certain electoral districts) is associated with higher political representation and voting. This is why residential isolation could not be included in these models. However these are results at the individual level, which are net of several strong control variables. They can be translated only indirectly into associations at the level of electoral districts (at the local, state, or federal level).

Other measures of political context are included in the models for all four groups. Voter identification requirements have a substantially negative impact on the voting of all groups except for Asians (though there are no significant impacts for registration for any group). Particularly strong negative effects are seen for blacks and Hispanics: a decrease in voting by 18% and 22% respectively. Even whites show dampened turnout associated with voter ID policies. Yet for Asians, strikingly, voter ID has the opposite effect, boosting turnout by nearly 30%. This is an intriguing instance in which Asian participation patterns markedly differ from that of other groups.

More liberal absentee voting policies increase the odds of voting for whites, Latinos and Asians, though there is no effect for blacks. The only significant effect on registration is a surprising negative coefficient for Latinos.

Greater immigrant access to a social service safety net is the other state-level predictor that has some significant effects. These are positive for white, Latino and Asian voting. There are also strong positive effects shown for Asian registration: an increase in the odds of registering by almost 30%. There is no significant effect on registration for any other group and no significant effects at all for blacks.

The last measure of political context included in the analysis is the dummy variable for the 1996 general election in California (note that there is a 1996 effect shared by all states, and this dummy variable represents additional variation for California residents). This is a special case often cited in the research literature as an instance when the threat of anti-immigrant legislation was likely to mobilize participation by Latinos and perhaps by Asians. Apparently the conditions present in California in 1996 were associated with higher levels of white and Latino registration, but this variable shows countervailing negative effects on voting for both groups. Only Asians showed the expected boost in voting while no effects were seen for black registration or voting.

Table 6 summarizes the findings on contextual effect for all races and for each group-specific model. It includes variables that were omitted from Tables 4-​5 that we examined and found to have no significant effects: early voting, bilingual ballot provisions, and immigrant receptivity in public opinion.

Summary of findings for contextual effects

Although the effects of what we term resources and rootedness are not the main concern of this study, they are clearly important predictors of political participation. Aside from generation and race/ethnicity, Table 3 shows significant and theoretically important effects of people's resources, and these are confirmed in the group-specific models. As found in previous research, having more resources and indicators of stronger connections to the local community increases the propensity of registering and voting for all groups. There is a uniform relationship between socioeconomic status and registration/voting for all groups. The higher a person's education or income level, the more likely the person is to register and vote. Compared to renters, homeowners are also more likely to register and vote in almost every model (the exception is the model for Asian voting where the coefficient is insignificant).

Older people register and vote at higher rates. Marriage generally enhances registration and voting (the exception is voting among blacks where the coefficient is insignificant). However, for all groups (except for Asians in the voting model), having more children in the household depresses registration and voting. This suggests that children in this context are not social connectors but perhaps a time demand that conflicts with political participation. For the most part, women are more likely to register and vote than men, with some variation (among whites, men are more likely to vote and there is no significant difference between Asian men and women in voting). People who have lived longer in their current place of residence are much more likely to register and vote.

This study has examined a very wide range of factors that contribute to variations in political participation. The results confirm that resources and rootedness based models go a long way toward explaining the likelihood of participation. Some aspects of the political context are also significant. The main contribution here, though, is to explore group-specific effects. This has been accomplished in several ways: by pinpointing the net differences across groups after controlling for other factors, by testing whether resource and political context operate similarly for each group, and by including nativity and several contextual measures that are specific to each minority group. Although electoral participation is ultimately something that people do in isolation in a voting booth, we have emphasized that it is also a collective act. The significant associations shown here between these individual behaviors and indicators of the things group members have in common support the conclusion that the group context of participation influences choices to register and vote.

The pooled analyses presented in Table 3 offer the best evidence of net differences across racial/ethnic groups. Like previous studies we have shown that all else equal, blacks register and vote at higher rates than whites. Among the largely immigrant groups with lower levels of participation, Latinos register and vote at higher rates than Asians. Unexpectedly, though, we showed that these group differences are conditional on nativity, because among immigrants Latinos participate more than either white or Asians and almost as much as blacks.

Looked at another way, the effects of nativity are contingent on race and Hispanic origin. Though there has been speculation that the high share of immigrants in the voting-eligible Latino and Asian populations could help to explain their lower political participation, the impact of nativity is not uniform across groups and does not account for the differences between groups in participation. For whites it is the 2nd generation that is more likely to register and vote. For other groups it is the 3+ generation that is more likely to register, but in terms of Latino voting it is the immigrant generation that stands out.

In this respect the assimilation model, which has proved useful in studying other aspects of social and economic life and which posits a general direction of incorporation across generations, is only partly right. Even language (measured here as linguistic isolation), which is a strong predictor of such outcomes as occupational achievement among immigrants, has mixed effects on Latinos in this study. Living in a Spanish-speaking household reduces the likelihood of registering but increases voting. Race, Hispanic origin, and immigration status apparently combine to produce distinctive collective influences on people's understanding of the political system and their engagement in it.

Other aspects of group members’ shared situation also affect participation. This study included no direct measure of group consciousness or mobilization. It would be desirable to have direct measures of organized efforts to mobilize voter turnout, such as voter registration drives or campaigns on specific issues that could stimulate greater participation. We introduced the “California 1996” variable in hopes of tapping such activity, especially among Latinos. Our results confirm that Latino and white participation were boosted, but only for registration and surprisingly with the opposite effect on voting. Minority political representation (our measure of co-ethnic public officials in the metropolitan region) is a related factor, and we found strong positive effects for blacks along with some evidence that there may be an effect also for Latinos. Although the direction of causality in this finding is not certain and the Asian results run in the opposite direction, these findings should encourage further efforts to bring measures of group-based organizational activity into analysis of individual political behavior.

State voting rules are especially important because these are amenable to change, and we examined a wide range of these policies. There is a consistent effect for voter ID requirements. Some states have recently introduced new identification requirements and others are considering it. The evidence here suggests that this policy will depress white, black and Latino participation in electoral politics—and the effect is especially strong for blacks and Latinos. On the other hand liberal absentee voting policies lead to higher voter turnout except, surprisingly, for blacks. Finally there is some evidence that a stronger immigrant service safety net is associated with greater political participation—an effect which is particularly clear for Asians for both registration and voting—but again not for blacks.

Other contextual variables, such as the requirement of bilingual ballots in some states, availability of early voting, receptivity of public opinion to immigrants., and the relative income of group members in relation to non-Hispanic whites, are not significant in any of the models that we examined for voter turn-out (and they are not included in the models reported here). Another important finding is that relative group income is surprisingly negative for all minority groups for registration. This suggests that minorities in a position of lesser economic disadvantage relative to white counterparts in a given MSA may be less likely to register. From an assimilation perspective, one would have expected the opposite effect, since higher income at the individual level is associated with higher likelihood of registration.

It is valuable to learn which aspects of the policy or political context make a difference. Perhaps more important the variations in how different groups respond to their community context remind us how little we still know about the group basis of political behavior and group solidarity (Junn 2006). This study has pinpointed several specific ways in which patterns of political participation for Asian Americans from other groups. All else equal, our study shows that Asians respond differently to co-ethnic representation (when it comes to registering), voter ID policy (for voting turnout), and the environment of “political threat” present in California in 1996. Moreover, Asian registration, unlike for the other minority groups, is positively affected by a robust social policy safety net for immigrants.

We have uncovered original evidence of an “empowerment effect” for African Americans along with hints of the opposite effect for Asian Americans. Future research is required to pinpoint the social processes (and the direction of the “causal arrow”) that underlies this relationship. Future work might analyze the relationship between coethnic representation and political participation over time or space (with attention to sub-state variation) and might consider additional variables such as co-ethnic group size. Furthermore, given that our multilevel analyses showed hints of a positive impact of co-ethnic officials on Latino (which were not robust and thus not reported in our tables), future research should probe for these effects for Latinos.

It is natural to find variations in coefficient estimates when many predictors are introduced in models for four different groups. We believe, however, that there are real differences here that remain to be explained. The challenge for researchers (from this study as well as many prior studies that allow group differences to be revealed) is to understand the specific circumstances of each group's arrival and incorporation into American society. This study shows that group differences are not solely a function of the resources and rootedness of group members or a consequence of the high proportion of immigrants among Latinos and Asians. Attention now needs to be focused on the contexts of participation faced by each group, and how their participation is facilitated or discouraged by their shared conditions in the communities where they live.

A previous version of this paper was presented at the 2007 annual meeting of the Eastern Sociological Society. This research was supported by the Russell Sage Foundation and by the research initiative on Spatial Structures in the Social Sciences at Brown University. The authors appreciate the contributions of several reviewers to the improvement of this study.

1One potentially relevant variable not used in the analysis is length of residency in the U.S. This variable could be important for the study of immigrant assimilation, but it would only be defined for 1st generation persons. It is also logically linked to the residential mobility indicator. Therefore including years in the U.S. could only be accomplished by creating a complex new variable that represents years, generation, and residential stability. We choose instead to focus on generational differences within categories of race and ethnicity.

2A theoretically important factor of this type that we studied but do not report here is racial isolation, based on indices calculated from tract level summary data from SF1. The Isolation Index measures the racial/ethnic composition of the census tract in which the average group member lives, and specifically the proportion of same-group members in the tract. Based on the studies cited above, we expected isolation to enhance political participation by blacks, but possibly to reduce participation by Latinos and Asians. In fact, the effects turned out to be positive for both blacks and whites, while negative or mixed for Latinos and Asians. Unfortunately isolation is very highly associated with group political representation (described below) – with correlations above .70. This is not unexpected, since communities with large minority populations are more likely to elect minority officials. But multicollinearity requires that we include only one of these predictors in the models, and we choose to present the more explicitly political of the two variables.

3Tests for multicollinearity show that variables measured at the state level are not strongly enough associated with one another to negatively affect the coefficient estimates.

4We also estimated multivariate probit models of voting and registration (with and without selection bias). Either type of model is appropriate for dichotomous outcome variables. Timpone (1998) has argued that selection is an important consideration because voting is contingent on registration, and he recommends correction for selection bias using a probit model discussed by Dubin and Rivers (1989, available in STATA as Heckprob). In such a model, the selection equation (registration) should contain at least one variable that is not in the outcome equation (voter turnout). To implement this procedure for the registration model, we included an indicator that is likely only to affect registration (how close in time to an election a person is allowed to register). For the voter turnout model, we added the voting policy variables that indicate states’ early voting and liberalized absentee voting policy. (All the personal characteristics shown to be strong predictors of registration are also expected to influence voting, so these were included in both models.)

The estimated correlations (rho) between the errors in the registration and voting equations for the Latino, Asian, non-Hispanic black, and non-Hispanic white sub-samples are negative. They are also statistically significant, except in the Latino model. These correlations imply the counter-intuitive results that, after controlling for measured characteristics, those who did not register would have been more likely to vote (had they registered) than those who did register. This counter-intuitive result reduces our confidence in the selection model. We then compared the magnitudes and signs of the estimated coefficients in the logit voting turnout equation (estimated only for registered voters) with coefficients in probit models with and without correction for selection bias. We found no significant differences. Therefore we conclude that the logit approach, which has the advantage of easier interpretability of model coefficients, provides a sound basis for analysis in this case.

5The multilevel models for all races replicate these findings for the impact of voter ID and absentee voting policies on voting. However, by contrast the multilevel models show no significant effects for immigrant safety net on voting while also showing negative effects of voter ID on registration.

John R. Logan, Brown University.

Jennifer Darrah, Brown University.

Sookhee Oh, University of Missouri-Kansas City.

• Alba Richard, Nee Victor. Remaking the American Mainstream: Assimilation and Contemporary Immigration. Harvard University Press; Cambridge, MA: 2003. [Google Scholar]
• Antunes George, Gaitz Charles M. Ethnicity and Participation: A Study of Meican- Americans, Blacks, and Whites. The American Journal of Sociology. 1975;80(5):1192–1211. [Google Scholar]
• Barreto MA, Ramirez R, Woods ND. Are naturalized voters driving the California Latino electorate? Measuring the effect of IRCA citizens on Latino voting. Social Science Quarterly. 2005;86(4):792–811. [Google Scholar]
• Bass LE, Casper LM. Differences in registering and voting between native-born and naturalized Americans. Population Research and Policy Review. 2001;20(6):483–511. [Google Scholar]
• Bobo L, Gilliam FD. Race, Sociopolitical Participation, and Black Empowerment. American Political Science Review. 1990;84(2):377–393. [Google Scholar]
• Brace K, Handley L, Niemi RG, Stanley HW. Minority Turnout and the Creation of Majority-Minority Districts. American Politics Quarterly. 1995;23(2):190–203. [Google Scholar]
• Brown KR, Brown RR. Faith and Works: Church-Based Social Capital Resources and African American Political Activism. Social Forces. 2003;82(2):617–641. [Google Scholar]
• Bueker Catherine Simpson. Political incorporation among immigrants from ten areas of origin: The persistence of source country effects. International Migration Review. 2005;39(1):103–140. [Google Scholar]
• Bueker Catherine Simpson. From Immigrant to Naturalized Citizen. LFB Scholarly Publishing; El Paso, Teas: 2006. [Google Scholar]
• Burden Barry C., Canon David, Mayer Kenneth, Moynihan Donald. Election Laws, Mobilization, and Turnout: The Unanticipated Consequences of Election Reform. University of Wisconsin; Madison: 2010. [November 2010]. https://mywebspace.wisc.edu/bcburden/web/bcmm2010.pdf. [Google Scholar]
• Cassel CA. Hispanic turnout: Estimates from validated voting data. Political Research Quarterly. 2002;55(2):391–408. [Google Scholar]
• Cho WKT. Naturalization, socialization, participation: Immigrants and (non-) voting. Journal Of Politics. 1999;61(4):1140–1155. [Google Scholar]
• Cho WKT, Gimpel JG, Dyck JJ. Residential concentration, political socialization, and voter turnout. Journal of Politics. 2006;68(1):156–167. [Google Scholar]
• Citrin Jack, Highton Benjamin. A report of the Public Policy. Institute of California; San Francisco, CA: 2002. How Race, Ethnicity, and Immigration Shape the California Electorate. [Google Scholar]
• De Jong GF, Steinmetz M. Receptivity Attitudes and the Occupational Attainment of Male and Female Immigrant Workers. Population Research and Policy Review. 2004;23(2):91–116. [Google Scholar]
• De Jong GF, Tran QG. Warm Welcome, Cool Welcome: Mapping Receptivity Toward Immigrants in the U.S. Population Today. 2001;291(8):4–5. [Google Scholar]
• DeSipio L. Making citizens or good citizens? Naturalization as a predictor of organizational and electoral behavior among Latino immigrants. Hispanic Journal of Behavioral Sciences. 1996;18(2):194–213. [Google Scholar]
• Eagleton Institute of Politics and Moritz College of Law . Report to the U.S. Election Assistance Commission on Best Practices to Improve Voter Identification Requirements Pursuant to the Help America Vote Act of 2002 Public Law 107-252. The State University of New Jersey and The Ohio State University; 2006. [Accessed March 2007]. [Google Scholar]
• Election Reform Information Project and The Constitution Project (“Electionline.org”) Election Reform Briefing: Voter Identification. Washington DC.: 2002. [April 2007]. http://www.electionline.org/Portals/1/Publications/Voter%20Identification.pdf. [Google Scholar]
• Election Reform Information Project (“Electionline.org”) Election Reform: What's Changed, What Hasn't and Why, 2000-2006. Washington, D.C.: 2006. [October 2007]. http://www.electionline.org/Portals/1/Publications/2006.annual.report.Final.pdf. [Google Scholar]
• Gay Claudine. A report by the Public Policy. Institute of California; San Francisco, CA: 2001. The Effect of Minority Districts and Minority Representation on Political Participation in California. [Google Scholar]
• Garcia John A., Arce Carlos H., Garcia F. Chris. Latinos and the Political System. University of Notre Dame Press; 1988. Political Orientations and Behaviors of Chicanos: Trying to Make Sense out of Attitudes and Participation. pp. 119–125. [Google Scholar]
• Hansen JM. Early Voting, Unrestricted Absentee Voting, and Voting by Mail. Task Force on the Federal Election System. 2001 [Google Scholar]
• Harris FC. Something within - religion as a mobilizer of African-American political activism. Journal of Politics. 1994;56(1):42–68. [Google Scholar]
• Highton B. Residential mobility, community mobility, and electoral participation. Political Behavior. 2000;22(2):109–120. [Google Scholar]
• Joint Center for Political and Economic Studies Black Elected Officials (BEO) Roster-Year 2000 crosstabs for entire USA all offices by State, City and zip code. (Database supplied to authors by The Joint Center) 2000.
• Jones-Correa Michael. Different paths: Gender, immigration and political participation. International Migration Review. 1998;32(2):326–349. [PubMed] [Google Scholar]
• Jones-Correa Michael. Institutional and Contextual Factors in Immigrant Naturalization and Voting. Citizenship Studies. 2001;5(1):41–56. [Google Scholar]
• Jones-Correa Michael. Language provisions under the Voting Rights Act: How effective are they? Social Science Quarterly. 2005;86(3):549–564. [Google Scholar]
• Junn Jane. Mobilizing Group Consciousness: When Does Ethnicity Have Political Consequences. In: Lee Taeku, Karthick Ramakrishnan S, Ramirez Ricardo., editors. Transforming Politics, Transforming America: The Political and Civic Incorporation of Immigrants in the United States. University of Virginia Press; 2006. pp. 32–50. [Google Scholar]
• Leighley JE, Nagler J. Individual and systemic influences on turnout - Who votes 1984. Journal of Politics. 1992;54(3):718–740. [Google Scholar]
• Leighley JE, Vedlitz A. Race, ethnicity and political participation: Competing models and contrasting explanations. Journal of Politics. 1999;61(4):1092–1114. [Google Scholar]
• Leighley Jan. Strength in Numbers. Princeton University Press; Princeton, NJ: 2001. [Google Scholar]
• Lien P. Asian Americans and voting participation: Comparing racial and ethnic differences in recent US elections. International Migration Review. 2004;38(2):493–517. [Google Scholar]
• Martinez Lisa M. Yes We Can: Latino Participation in Unconventional Politics. Social Forces. 2005;84:135–55. [Google Scholar]
• Missouri Census Data Center [April 2007];MABLE: Geographic Correspondence Engine with Census 2000 Geography. 2007 http://mcdc2.missouri.edu/websas/geocorr2k.html.
• Muhlhausen David B., Sikich Keri Weber. New Analysis Shows Voter Identification Laws Do Not Reduce Turnout. [October 2007];A Report of the Heritage Center for Data Analysis. 2007 http://www.heritage.org/Research/LegalIssues/cda07-04.cfm.
• National Association of Latino Elected and Appointed Officials (NALEO) 2000 Directory of Latino Elected Officials. NALEO Educational Fund. 2000 (Database supplied to authors from NALEO) [Google Scholar]
• Pantoja AD, Ramirez R, Segura GM. Citizens by choice, voters by necessity: Patterns in political mobilization by naturalized Latinos. Political Research Quarterly. 2001;54(4):729–750. [Google Scholar]
• Pantoja AD, Segura GM. Does Ethnicity Matter? Descriptive Representation in Legislatures and Political Alienation Among Latinos. Social Science Quarterly. 2003;84(2):441–459. [Google Scholar]
• Portes A, Mozo R. The political adaptation process of Cubans and other ethnic-minorities in the United States: a preliminary analysis. International Migration Review. 1985;19(1):35–63. [Google Scholar]
• Portes A, Rumbaut RG. Immigrant America: A Portrait. University of California Press; Berkeley, CA: 1996. [Google Scholar]
• Presser Stanley, Traugott Michael W., Traugott Santa. Vote ‘Over’ Reporting in Surveys: The Records or the Respondents. Technical Report No. 39, National Election Study. 1990 www.umich.edu/~nes.
• Putnam Robert D. Tuning In, Tuning Out: The Strange Disappearance of Social Capital in America. PS. 1995;28:664–83. [Google Scholar]
• Ramakrishnan SK, Espenshade TJ. Immigrant incorporation and political participation in the United States. International Migration Review. 2001;35(3):870–909. [Google Scholar]
• Ramakrishnan S. Karthick. Democracy in Immigrant America :changing demographics and political participation. Stanford University Press; Stanford, CA: 2005. [Google Scholar]
• Rosenstone Steven J., Hansen John Mark. Mobilization, participation, and democracy in America. Macmillan Pub. Co.; New York, NY: 1993. [Google Scholar]
• Segal Adam. A report of the Hispanic Voter Project. John Hopkins University; 2002. Records Broken: Spanish-language Television Advertising in the 2002 Election. [Google Scholar]
• Shaw D, de la Garza RO, Lee J. Examining Latino turnout in 1996: A three-state, validated survey approach. American Journal of Political Science. 2000;44(2):338–346. [Google Scholar]
• Skerry Peter. The Political Assimilation of Today's Immigrants. In: Jacoby Tamar., editor. Reinventing the Melting Pot: The New Immigrants and what it Means to be American. Basic Books; New York, NY: 2004. pp. 221–235. [Google Scholar]
• Tate Katherine. Black Political Participation in the 1984 and 1988 Presidential Elections. American Political Science Review. 1991;85:1160–1176. [Google Scholar]
• Tate Katherine. From Protest to Politics. Harvard University Press; Cambridge, MA: 1993. [Google Scholar]
• Timpone RJ. Structure, behavior, and voter turnout in the United States. American Political Science Review. 1998;92(1):145–158. [Google Scholar]
• Tucker James, Espino Rodolfo. Minority Language Assistance Practices in Public Elections. 2006. Reprinted in Voting Rights Act: Evidence of Continued Need, H.R. Rep No. 109-103, at 2094.
• UCLA Asian American Studies Center National Asian Pacific American Political Almanac. 2000.
• Uhlaner Carole J., Cain Bruce E., Kiewiet D. Roderick. Political participation of ethnic minorities in the 1980s. Political Behavior. 1989;11(3):195–231. [Google Scholar]
• Van Hook J, Brown SK, Bean FD. For Love or Money? Welfare Reform and Immigrant Naturalization. Social Forces. 2006;85(2):643–666. [Google Scholar]
• Verba S, Schlozman KL, Brady H, Nie NH. Race, Ethnicity and Political Resources - Participation in The United-States. British Journal of Political Science. 1993;23:453–497. [Google Scholar]
• Verba Sidney, Schlozman Kay Lehman, Brady Henry. Voice and Equality: Civic Voluntarism in American Politics. Harvard University Press; Cambridge, MA: 1995. [Google Scholar]
• Washington Ebonya. How Black Candidates Affect Voter Turnout. The Quarterly Journal of Economics. 2006;121:973–98. [Google Scholar]
• Wolfinger NH, Wolfinger RE. Family Structure and Voter Turnout. Social Forces. 2008;86(4):1513–1528. [Google Scholar]
• Zimmermann W, Tumlin K. Patchwork Policies: State Assistance for Immigrants Under Welfare Reform. The Urban Institute; Washington, D.C.: 1999. [Google Scholar]