Census Bureau

Are There Differences in Registration and Voting Behavior Between Naturalized and Native-born Americans?

Loretta E. Bass and Lynne M. Casper

Population Division
U.S. Bureau of the Census
Washington, D.C. 20233-8800

February 1999

POPULATION DIVISION WORKING PAPER NO. 28


This paper was presented at the Annual Meeting of the American Sociological Association (ASA), San Francisco, CA, August 1998.

This paper reports the results of research and analysis undertaken by Census Bureau staff. It has undergone a more limited review than official Census Bureau publications. This report is released to inform interested parties of research and to encourage discussions.


Table of Contents

Text
Abstract
Introduction
What Determines Whether an Individual Will Vote?
Data and Analytical Samples
Multivariate Methods
Descriptive Results
Multivariate Results
Discussion
Conclusions

References

Tables
1. Reported Voting and Registration Among Citizens, by Nativity Status: November 1996 (34k)
2. Odds of Reported Registration Among All Citizens and Naturalized Citizens: November 1996 (9k)
3. Odds of Reported Voting Among All Citizens and Naturalized Citizens: November 1996 (9k)

Appendices
1. Logistic Regression Results Predicting Reported Registration Among All Citizens and Naturalized Citizens: November 1996 (23k)
2. Logistic Regression Results Predicting Voting Among All Citizens and Naturalized Citizens: November 1996 (23k)

Figure
Registration and Voting Among Naturalized and Native-born Americans (24k)

Population Division Working Papers


Abstract

Relatively little is known about the differences in voting behavior between immigrants and native-born Americans, primarily due to a lack of good quality data on the national level. Using new data from the Voting and Registration Supplement to the November 1996 Current Population Survey (CPS), we first estimate logistic regressions to examine which socioeconomic and demographic factors are related to whether a citizen will 1) register to vote and 2) vote. Second, we also investigate whether native-born or naturalized citizens are more likely to register and vote, net of other factors. Third, we examine whether citizens born in specific regions of the world are more or less likely than those born in the U.S. to register and vote. Finally, we analyze the extent to which voting behavior among naturalized citizens is associated with region of origin and length of time spent in the U.S.

We find that citizens with more education and income, employed individuals, those who own their homes, those who have a longer length of time at current residence, professionals, women, older individuals, married individuals, and Blacks are more likely to register and vote. We find that naturalized citizens are less likely to register and to vote than native-born citizens, net of other effects. Citizens born abroad in Europe, Latin America, and Asia are less likely to register and those born abroad in Europe and Asia are less likely to vote than those born in the U.S. Among naturalized citizens, those who have a longer length of time at current residence and in the U.S., are older, with more education and higher income are more likely to register and vote. Region of origin was not significantly related to voting and registration among naturalized citizens, once duration in the U.S. was included in the model.


Introduction

The assimilation process of U.S. immigrants has generated much scholarly and popular discussion (for an overview, see Alba and Nee 1997; Smith and Edmonston 1997). An intriguing area of research is voting behavior, but we know relatively little about the differences in voting behavior between naturalized and native-born Americans, primarily due to a lack of good quality data on the national level. New data from the Voting and Registration Supplement to the November 1996 Current Population Survey (CPS) have recently become available and allow us to provide a profile of voting behavior by nativity status. Nativity status is defined by whether an individual is a native-born citizen or a naturalized citizen. The availability of this new data means that for the first time, we can answer questions such as: What factors are associated with whether or not a citizen registers and votes? Are there differences in the voting behavior of naturalized and native-born citizens? Does where a citizen is born affect whether or not she/he registers and votes? And, does time in the U.S. and region of origin affect the registration and voting behavior of naturalized citizens?

To answer these questions, we examine the propensity to register and vote by nativity status. First, we use logistic regressions to examine which socioeconomic and demographic factors are related to whether a citizen will 1) register to vote and 2) vote, and to investigate whether naturalized citizens are more likely to vote than native-born citizens once other factors have been taken into account. We then examine whether citizens born in the U.S. are more or less likely to register and vote then those born in various regions of the world. We also investigate the extent to which region of origin and length of time in the U.S. affect the likelihood of registration and voting among naturalized citizens.

What Determines Whether an Individual Will Vote?
Registration and Voter Turnout Among All Citizens

Previous research shows that "citizens of higher social and economic status participate more in politics. This generalization...holds true whether one uses level of education, income, or occupation as the measure of social status" (Verba and Nie 1972:125; see also Casper and Bass 1998; Leighley and Nagler 1992). Lewis, McCracken, and Hunt (1994) hold that those with higher stakes in society are the most likely to go to the polls -- older individuals, homeowners, and married couples.

The majority of these studies examine the bivariate relationships between voting and registration and other factors. However, evidence suggests that it is important to examine these relationships in a multivariate framework. Variations in income have been noted to have little relationship to voting once age and education are taken into account (Wolfinger 1994). In addition, Wolfinger and Rosenstone (1980) contend that although married men and women are more likely to vote than others, the effects among both men and women vary considerably with age and education. These results suggest that there is a need to consider these factors in a multivariate model to assess their relative importance, once other factors have been taken into account.

Leighley and Nagler (1992) tested whether demographic factors, like race and gender, are more important than socioeconomic factors like education in predicting voter turnout, and found that while it is important to include measures of demographic factors, education is a much stronger predictor of voter turnout. Likewise, Wolfinger and Rosenstone (1980) document the lower voter turnout among Blacks and Hispanics, but attribute this lower rate of voter turnout to lower educational levels and higher proportions of young and poor among minorities. Other results suggest that women are also more likely to register and vote (Jennings 1985, 1989, 1993).

Squire, Wolfinger, and Glass (1987) assert that it is important to consider the effect of residential mobility on registration and voting behavior. They found that movers resemble stayers on motivational factors related to voting and that this effect remains net of other socioeconomic and demographic factors. Therefore, they conclude that the requirement that citizens must register anew after each change in residence constitutes a key stumbling block in the path of a trip to the polls. Length of residence is a particularly important factor to consider in a highly mobile country like the U.S. where 16 percent of Americans move every year (Hansen 1997).

Another problem with some previous voting studies is that they are not always nationally representative of the voting population (Leighley and Nagler 1992; Verba and Nie 1972). Other research, such as that conducted by Squire, Wolfinger, and Glass (1987) and Wolfinger and Rosenstone (1980), used Current Population Survey data which was based on a nationally representative sample of the total civilian noninstitutionalized U.S. population. However, these two studies had to separate the citizen population from the total U.S. population using imperfect measures because citizenship status was neither directly asked nor fully edited in prior surveys. Also, the majority of studies are somewhat outdated; they are based on data collected in the 1980s.

In this study, we improve upon previous research in several ways. First, we use the most recent data available to examine voting and registration. Second, because of the availability of new, more accurate data on citizenship, we are able to restrict our analysis to the citizen population. Third, we use multivariate models to simultaneously control for many factors known to be related to voting and registration.

We expect that women, older individuals, married individuals, non-Hispanic whites, those with higher levels of education and income, those who are employed, homeowners, and those residing in their homes for longer periods of time will be more likely to register and vote, all else being equal. In addition, recent research has not investigated whether having children is associated with the registration and voting behavior of citizens.1 One might reason that those with children of school age would be more likely to be involved in their school districts and in local elections since the results are likely to affect the well-being of their children. We improve on previous research by including in our analysis a measure of whether or not an individual has related children under 18 in the household.

Registration and Voter Turnout among Naturalized Citizens

Nativity status and country of origin are important to consider in predicting the likelihood that an individual will register and vote. Different countries and regions of the world allow and exhibit different levels of political participation (Inglehart 1997). It is striking that voter participation in communist countries in the 1980s was much higher than in democratic countries of Western Europe and North America. Naturalized citizens from countries without the right to vote or from countries with one-party states might be more highly motivated to participate in the political process than native-born citizens who have had access to this right from birth.

Little research to date has been published on the differences in registration and voting behavior between naturalized and native-born citizens. Casper and Bass (1998) report in a bivariate analysis that native-born citizens are more likely to vote than naturalized citizens, except among Hispanics where the opposite is true. However, to the extent that naturalized Hispanics and native-born Hispanics have other characteristics that are positively associated with voting and registration, this effect may be spurious. For example, if naturalized Hispanics are more highly educated or earn more income, they may be more likely to vote than native-born Hispanics, not because of nativity status but because of these other factors.2 Thus, in ascertaining the relationship between nativity status and voting and registration, it is important to control for various social, economic, and demographic characteristics.

In this research, we use multivariate methods to investigate the differences in voting behavior between naturalized and native-born citizens. We expect that naturalized citizens may be more likely to vote than native-born citizens if they come from countries that didn't allow citizens the right to vote. They may also be more likely to vote to the extent that they place a higher value on the right to vote than do native-born citizens. On the other hand, if naturalized citizens are still emotionally tied to their countries of origin, they may be less likely to register and vote than native-born citizens. In addition, to the extent that native-born citizens are more integrated into their communities, are more familiar with the issues and candidates, and find the issues more applicable to themselves, we might expect native-born citizens to be more likely to register and vote than naturalized citizens.

Research also suggests that region of origin is related to voting and registration among naturalized citizens (Casper and Bass 1998). Bivariate results indicate that naturalized citizens from other North American countries, principally Canada, were found to be more likely to vote than naturalized citizens from Latin America and Asia. Naturalized citizens from other North American countries were also most likely to register, compared with those from other regions. Naturalized citizens from Asia were the least likely to register.

In this study, we examine how region of origin is related to voting and registration and expect that people immigrating from non-democratic countries may be more likely to register and vote than native-born individuals because the right to vote was not granted in their country of origin. In contrast, individuals from countries where voting rates tend to be higher than in the United States may be more likely to register and vote because voting is considered to be an important aspect of being a responsible citizen. In addition, people from different countries may be more or less likely to vote because of other cultural differences.

Length of time in the U.S. is an important factor to consider when analyzing voting participation among naturalized citizens because it is associated with the degree to which they have assimilated. For example, the 1990 decennial census indicates that overall only three-fifths of the immigrants who came 10 or fewer years ago spoke English well or very well; but among those who had been here 30 years or more, 97 percent reported that they could speak English well or very well (U.S. Bureau of the Census 1990). Clearly, assimilation and the acquisition of the English language is a gradual process which takes place over a number of years and may also have parallels in political life.

Here, we investigate whether duration in the U.S. is associated with the likelihood of voting and registration among naturalized citizens. Recently naturalized citizens had the opportunity to register at their swearing-in ceremony and therefore might be more likely to register and vote than others who have to register as a separate process. However, among the remainder of the naturalized citizens who did not receive their citizenship in the recent past, those who have lived here for longer periods of time have had longer to assimilate into their communities, to become familiar with the issues and candidates, and to develop their own set of special interests. Among this group, we therefore expect that those who have been in the U.S. for longer periods of time, will be more likely to register and vote.

Among naturalized citizens, evidence suggests that women may be more likely to register and vote than men. For example, Jones-Correa (1998) suggests that among Latin American immigrants, women draw on and become more invested in the political and governmental systems, while men remain less integrated and continue to rely on their old networks for many years after entering the U.S. He further suggests that Latin American immigrant women learn to use institutions and become politically integrated more quickly than men. In contrast, immigrant men cling to their old-country based institutions and clubs where they have higher social status based on the social structure they experienced in their country of origin.

In this study, we improve upon previous research by considering differences in registration and voting between naturalized and native-born citizens and among naturalized citizens in a multivariate framework. We expect the same factors considered for the overall citizen population to predict the propensity to register and vote among naturalized citizens age, gender, marital status, race and Hispanic origin, children in the household, education, income, employment status, homeownership, and length of current residence.

Data and Analytical Samples

The analysis in this paper is based on data from the November 1996 Current Population Survey (CPS) conducted by the U.S. Bureau of the Census for the Bureau of Labor Statistics. The November CPS is a nationally representative household survey of the civilian noninstitutionalized population of the United States based on a complex sample design. In 1996, approximately 48,000 households were sampled in 754 primary sampling units (PSUs) across the United States. The interviewed households contain about 95,000 people 15 years old and over.

The main purpose of the CPS is to collect labor force information to estimate the monthly national unemployment rate and other employment statistics. The CPS collects additional information on other social, demographic, and economic characteristics of people such as age, gender, race and Hispanic origin, marital status, household relationship, and educational attainment.

The November supplement, administered every other year (in congressional and presidential election years) since 1964, collects information on voting and registration behavior for all civilian household members of voting age (18 years and over since 1972). The November CPS is the major source of information regarding national voting and registration patterns in relation to the social, demographic, and economic characteristics of individuals. Because of the large sample size and the sampling frame, the CPS can also be used to provide voting and registration estimates on the state level.

In the 1996 Voting and Registration Supplement, respondents were asked if they voted in the election held on November 5, 1996. If they responded affirmatively, they were considered to have been registered, if not, they were asked an additional question about whether or not they were registered. For the first time in a Presidential election, the CPS began collecting detailed information on citizenship status. Previous research based on the CPS has generally been limited to examining the voting and registration behavior among the voting-age population (Jennings 1982, 1985, 1989, 1993; Leighley and Nagler 1992; Teixeira 1992; Wolfinger and Rosenstone 1980). Beginning in 1994, the CPS began asking direct questions on citizenship status and editing and allocating the responses for consistency and completeness.3 In addition to asking about each person's country of birth, new questions included country of birth of both parents, and citizenship and year of entry for those who were foreign-born.4 These additional questions allow us for the first time to examine voting and registration behavior among the citizen population as well as between native-born and naturalized citizens.

Estimates from the November 1996 CPS indicate that there were 179.9 million citizens 18 years and over who were eligible to register and vote of which 171.7 million were native-born and 8.2 million were naturalized. These populations constitute our analytical sample for the descriptive results presented in table 1 and are represented by 80,943 and 3,591 unweighted cases of native-born and naturalized citizens, respectively.

Multivariate Methods

Overall, we specify eight logistic regression models to examine registration and voting: six models for all citizens and two models for naturalized citizens. Starting with the equations to predict registration, the first model serves as our base model and predicts registration for all citizens as a function of gender, race and Hispanic origin, marital status, presence of related children under 18 in the household, educational attainment, employment, income, home ownership, length of residence, metropolitan residence, and region of residence. The second model includes all of the variables in the first model and adds a measure of nativity status to ascertain whether naturalized citizens are more or less likely to register than native-born citizens, controlling for all the variables in the base model. In the third model, we use dummy variables indicating the country of origin to investigate whether naturalized citizens from different regions of the world are more or less likely to register when compared with citizens born in the U.S., again controlling for all of the variables in the base model.

Our final goal is to investigate among naturalized citizens whether country of origin and duration in the U.S. affect the propensity to register and vote. The fourth model includes measures to indicate country of origin and duration in the U.S., in addition to the control variables used in the base model.

We then use the same four models to predict voting (models 5, 6, 7, and 8). Our sample for all citizens consists of those 18 years old and over for whom we have valid voting and registration data resulting in 78,195 unweighted cases for the registration models and 78,309 unweighted cases for the voting models.5 Our analytic sample for the naturalized citizen models includes all naturalized citizens 18 years old and over with valid voting and registration data, resulting in 3,200 unweighted cases for the registration model and 3,223 unweighted cases for the voting model.6

For the logistic regressions shown in this paper, the observations were first weighted, then divided by the average weight of the sample to approximate the actual number of cases in the regression. The CPS has a complex sample design, involving cluster, multistage sampling, and stratification. As a result, standard errors produced by SAS tend to be underestimated. To adjust for these sample "design effects" we multiply the standard errors produced in SAS by 1.20 (the square root of the design effects), in the models containing all citizens and 1.37 in the models containing naturalized citizens, and recalculate the significance levels based on the adjusted standard errors.

DEPENDENT VARIABLES

Registration and Voting

We use dichotomous variables to indicate whether or not an individual reported registering or voting in the 1996 election. We include all valid responses to this question whether the response was self-reported or reported by a proxy.

INDEPENDENT VARIABLES

Nativity Status

Based on our earlier discussion, we assume that nativity status matters in predicting the likelihood that an individual will register and vote. Our measure of nativity status is dichotomous indicating whether a citizen is naturalized or native-born; native-born is the omitted category.

Region of Origin

The propensity to register and vote is likely to be related to an individual's country of origin. Because the CPS sample is too small to allow us to conduct meaningful analyses on individual countries of origin, we create a set of dummy variables to represent five different regions of the world: Europe, Asia, Latin America, Other North American countries, and all other countries including all of the African countries, Australia, New Zealand, Figi, and the Pacific Islands.7 The omitted category is the U.S. and its territories when considering the citizen population overall and Other North American when considering the naturalized population. While these categories cannot be used to test explicitly our country-specific hypotheses, we believe they still provide meaningful information in a broader sense.

Duration in the U.S.

We use a set of dummy variables to ascertain whether duration in the U.S. is related to the likelihood that a naturalized citizen will vote. The categories are within 10 years, 11 to 25 years, 26 to 45 years, 46 to 65 years, and 66 years and over. The omitted category is within ten years.

Other Factors

To assess accurately whether naturalized citizens are more likely to report that they register and vote than native-born citizens, whether region of origin makes a difference, and whether duration in the U.S. makes a difference in registration and voting among naturalized citizens, it is vital to control for other factors related to registration and voting.

Controlling for various social, economic, and demographic characteristics is especially important when certain characteristics of subgroups vary in a systematic fashion and are related to the behavior being modeled. We include in all of our models several categorical variables known to be related to registration and voting: gender, race and Hispanic origin, marital status, presence of related children under 18 in the household, educational attainment, home ownership, length of time at residence, metropolitan residence, and region of residence. We also include two continuous variables for age and income.8

Descriptive Results

Overall, 71 percent of citizens reported that they were registered and 58 percent reported that they turned out to vote in the 1996 Presidential election (table 1). Figure 1 shows that native-born citizens were more likely to report that they registered and voted (71 percent and 59 percent, respectively) than naturalized citizens (63 percent and 53 percent, respectively). Across most social, economic, and demographic factors, native-born citizens were more likely to vote than naturalized citizens. However, among Hispanics and those with incomes less than $15,000, the reverse was true. Fifty-three percent of naturalized Hispanic citizens compared with 42 percent of native-born Hispanic citizens voted. Among those in the lowest income bracket (less than $10,000), 47 percent of naturalized citizens voted compared with only 41 percent of the native-born. However, to assess accurately these seemingly contrary bivariate relationships, it is important to control for other factors related to voting (i.e. education, age, occupation).

In general, the findings in table 1 replicate the results of previous research such as Casper and Bass (1998), Jennings (1993) and Lewis, et al. (1994), and indicate that previously noted relationships are consistent among both native-born and naturalized citizens. Those who are more established in society are the most likely to register and vote -- older individuals, homeowners, married couples, and people with more schooling, higher incomes, and good jobs. Much like the larger citizen population, naturalized and native-born citizens are more likely to register and vote the longer they have lived at their current residence. Also in both groups, people residing in the South are less likely to register and to vote compared with other regions. We next consider these variables in the multivariate analysis to ascertain whether these relationships hold once other variables that may be intercorrelated are taken into account.

Multivariate Results

Registration for All Citizens

Model 1 in table 2 shows that several social, economic, and demographic characteristics are significantly related to registration for all citizens. With increasing age, it is evident that individuals are more likely to register, net of other factors. Women are 23 percent more likely to register than men. By controlling for other factors, we see that Blacks are almost 1.5 times more likely than Whites to register, while Hispanics and people of other races are less likely than Whites to register.9 Note that the finding for Blacks was the reverse in the bivariate analysis (table 1).10

Married individuals are 27 percent more likely to register than never married citizens, while divorced, widowed and separated individuals are less likely to register than never married citizens. While we had hypothesized that individuals with related children in their household would be more likely to register, the contrary was found. Those citizens with related children in their household are less likely to register than those without children.

Citizens with more education and higher income levels are more likely to register than citizens with less education and lower income levels. These findings are consistent with earlier findings by Kelley, Ayres, and Bowen (1967). Employed and unemployed individuals are more likely to register than those not in the labor force. Homeowners are 34 percent more likely to register than those who rent their homes.

As hypothesized, length of time at current residence is a significant predictor of being registered. This finding supports earlier findings about length of residence (Squire, Wolfinger and Glass 1987). Those at their residence one to four years are nearly 30 percent more likely to register than those at their residence less than a year, and, further, those at their residence five or more years are nearly twice as likely to register as those at their residence less than a year. Surprisingly, those in the suburbs and nonmetropolitan areas are less likely than those in central cities to register, when controlling for other factors. Note that the bivariate relationship shown in table 1 indicates that suburbanites are more likely to register. Those in the Midwest are 20 percent more likely to register than those in the South.

Registration Among Naturalized and Native-born Citizens

The results in model 2 indicate that nativity status is related to registration once other factors are controlled for. Naturalized citizens are 35 percent less likely than native-born citizens to be registered, controlling for all the variables in the base model. Hispanic is no longer significant when nativity status is included in the model which suggests a confounding relationship between nativity status and the Hispanic race category. Other significant relationships appearing in the base model do not change significantly when nativity status is included in the second model.

The third model shows that region of origin is a significant factor among all citizens in predicting the likelihood that an individual will register. Those from Europe, Asia and the Pacific Islands, Latin America and "Other regions" of the world are less likely to register than those born in the United States, again controlling for the baseline variables used in this analysis. As one might expect, no significant difference in registration exists between Hispanics and Whites when region of origin is included in the model. In general, the odds ratios and the direction of the effects of the other variables remain the same when region of origin variables are included in the model.

Registration Among Naturalized Citizens

To address our final objective -- predicting registration among the naturalized citizen population -- in model 4 we used the base model again and included measures for duration in the U.S. and region of origin. We found that with increasing years spent in the U.S., individuals are generally more likely to register. However, when duration in the U.S. is included, region of origin no longer predicts the propensity to register. In contrast to the larger citizen population, marital status, gender, race and Hispanic origin, employment status, home ownership, occupation, metropolitan status, and region of residence are not significant in predicting registration among the naturalized citizen population.

Overall, the most prominent predictors of registration among naturalized citizens are duration in the U.S., educational level, and length of time at residence. Those who came to this country at least 66 years ago are more likely to be registered compared with those who came 10 or less years ago. Much like the citizen population as a whole, naturalized citizens with at least some college education are two-and-one-third times more likely to be registered than those with a high school or less educational level. Those with five years or more at their current residence are about 1.5 times more likely to be registered than those with less than one year at current residence.

Voting Among All Citizens

The results of the fifth model (table 3) -- the base model predicting the propensity to vote -- are very similar to model 1 (table 2). Women are 21 percent more likely to vote than men, net of other factors. With every year of age, individuals are slightly more likely to vote. Again, it is striking that Blacks are 1.5 times more likely to vote than Whites, while Hispanics and those of other races are less likely than Whites to vote. Married individuals are 30 percent more likely, and divorced, separated, and widowed individuals are 19 percent less likely to vote than never married individuals. Those with related children in the household are 14 percent less likely to vote than citizens without related children in the household.

Citizens with at least some college are two-and-two-thirds more likely to vote than those with a high school education or less. Those in professional occupations are 43 percent more likely to vote than individuals in other types of occupations. Employed individuals are more likely to vote than those who are not in the labor force. With increasing income, individuals are more likely to vote. Homeowners are nearly 40 percent more likely to vote than those who rent or do not own their homes.

With a greater length of time at the same residence, individuals are more likely to vote. For example, those who have been at their current address for more than five years are almost two times as likely to vote as those who have lived at their current residence for less than one year. Those in suburbs and nonmetropolitan areas are about 15 percent less likely to vote than those living in central cities. Those in the Midwest and the West are 26 percent and 30 percent, respectively, more likely to vote than those in the South. Individuals in the Northeast are also more likely to vote than those in the South.

Voting Among Naturalized and Native-born Citizens

When nativity status is added in model 6, we find that naturalized citizens are 26 percent less likely to vote than native-born citizens. Other parameter estimates remain about the same, suggesting that no masked effects of nativity status are manifested through the other variables in the model.

Model 7 shows that region of origin is a significant predictor of the propensity to vote, again controlling for all of the variables in the base model. Those of European origin are 27 percent less likely than those born in the U.S. to vote. Asians are about 40 percent less likely and those from other regions of the world are 34 percent less likely to vote than those of U.S. origin. Compared to the base model, women are still 21 percent more likely to vote than men. The remaining control variables continue to have similar significant effects.

Voting Among Naturalized Citizens

Examining the voting behavior among naturalized citizens in model 8, we considered the base model and included region of origin and duration in the U.S. to understand their effects on the propensity to vote. As expected, duration in the U.S. is positively associated with voting behavior. Our findings show that those who have spent 26 to 65 years in the U.S. are more likely to vote than those who entered the U.S. ten or less years ago. However, contrary to our expectations, the region of origin variables do not remain a significant predictor of the propensity to vote among naturalized citizens.

Some of the other variables in the model behaved much the same as they did in the other models. With increasing age and income, naturalized citizens are more likely to vote. Among naturalized Citizens, those with professional occupations are 35 percent more likely to vote than those with other occupations. Education is a strong predictor of voting among naturalized citizens where those with at least some college education are over two times more likely to vote than those with a high school degree or less. The longer a naturalized citizen has resided at her/his current residence, the more likely s/he is to vote. Those who have more than five years at their residence are 68 percent more likely and those who have one to four years at their residence are 40 percent more likely to vote than those who have resided at their residence less than one year. Region of the U.S. has a significant effect on the propensity to vote among naturalized citizens. Those from the West are more likely to vote than those from the South.

Some of the variables in model 8 did not have the expected effects. Contrary to the citizen population, race and Hispanic origin, marital status, gender, employment status, home ownership, and metropolitan residence status are not significant predictors of the propensity to vote among naturalized citizens.

Discussion

Even when controlling for other variables in the model, among the citizen population those who are more established in society are the most likely to register and vote -- older individuals, homeowners, those with a longer length of time at current residence, married couples, and people with more schooling, higher incomes, and good jobs. Race is also an important factor related to whether a citizen will register and vote. These findings are consistent with previous studies on voting participation, and, further, show the importance of social, economic, and demographic characteristics in predicting voting behavior. Among naturalized citizens, education, length of time at current residence, and length of time in the U.S. are the most prominent predictors of registration and voting. The main determinant of voting behavior among both naturalized and native-born citizens is educational level.

Overall, we find that naturalized citizens are 35 percent less likely than native-born citizens to register and 26 percent less likely than native-born citizens to vote. This may be because naturalized citizens have not developed strong ties within their communities or do not relate to the issues or candidates. One might expect that the more time that naturalized citizens spend in the U.S., the more assimilated they become, building stronger ties to community and becoming more integrated into U.S. institutions and social customs. The duration in the U.S. variable among naturalized citizens provides some evidence of this; naturalized citizens that have been in the U.S. a longer length of time are more likely to register and vote.

Our results indicate that different factors are important in predicting the propensity to register and vote between naturalized and native-born citizens. While some factors, like education, income, and age are significant predictors of the propensity to register and vote in both groups, other factors like marital status, region of origin, labor force status, home ownership, metropolitan residence, and gender are not significant predictors of whether an individual will register and vote among the naturalized citizen population.

As stated, evidence suggests this may be because the duration in the U.S. among naturalized citizens is one of the most important factors, indicating that the assimilation argument has significant merit.

All voting studies based on recent data have reported that women are more likely to register and vote than men. Qualitative research suggests that naturalized Hispanic women might be more likely to vote than naturalized Hispanic men because immigrant women are more likely to be politically active (Jones-Correa 1998). When we tested whether this holds among all naturalized women and men, regardless of region of origin, we found that naturalized women are no more likely than naturalized men to vote (results not shown). Note that this remains the case even after controlling for region of origin and race, suggesting that regardless of where naturalized women come from they are no more or less likely to vote than naturalized men. However, when an interaction term for being Latin American and female is included in the model, the relationship is significant (results not shown). Therefore, among the naturalized population women from Latin America are unique in that they are more likely to vote than others. This finding further supports the qualitative findings on the expected political participation of immigrant Latin American women.

In all of our models, we had expected that those with related children in their household would be more likely to register, since they may be more concerned about the community and demonstrate greater civic responsibility. However, we found the contrary; those with related children in their household are less likely to vote than those without children, across the regression models. This may be the case because those with children may be busier than those without, and consequently, have less time to get to the polls. Other research has indeed shown that the primary reason given for not voting among those who were not registered was that they were too busy or could not get time off from work (Casper and Bass 1998). In fact, among those with related children under 18 in their household, 27 percent reported they were too busy to vote compared with only 17 percent of those with no related children in their household.

It is interesting that Blacks are 1.5 times more likely to register and to vote than Whites, when nativity status and baseline variables are included in the model. This finding further delineates the race and class interaction in predicting voting behavior. While Leighley and Nagler (1992) found that education was much more important than race in affecting voter turnout, we find that race and education are both very strong predictors of the propensity to register and to vote -- at least among the citizen population. Among naturalized citizens race did not seem to matter, perhaps because race does not have the same social connotations in the countries of origin of naturalized citizens as it does in the U.S.

This analysis provides strong evidence of the importance of simultaneously controlling for several factors that are related to voting and registration behavior. Many of the bivariate relationships that have been reported do not hold up or have been reversed in a multivariate framework. For example, bivariate results indicate that Whites are more likely to register and vote than Blacks (see table 1), however, when controlling for other factors, we find that Blacks are actually more likely to register and vote. Also, many of the bivariate relationships are better explained in a multivariate framework. For example, bivariate results indicate that native-born citizens are more likely to vote than naturalized citizens, but that native-born Hispanics are less likely than naturalized Hispanics to vote. To understand more fully this relationship, we ran the base model adding an interaction term for Hispanic and naturalized, and found the interaction to be significant (results not shown). Therefore, Hispanic naturalized citizens continue to be more likely to vote than native-born Hispanics even when other factors in the model are controlled for.

Conclusions

Voting research conducted in the 1970s and 1980s demonstrated that those who are more established in society are the most likely to register and vote. Our research using November 1996 Current Population Survey data shows that this trend continues in the 1990s. The major findings based on multivariate models are that citizens with more education and income, employed individuals, those who own their homes, those who have a longer length of time at current residence, professionals, women, older individuals, married individuals, and Blacks are more likely to register and vote.

Another major conclusion from our research is that voting behavior between naturalized and native-born citizens differs. Overall, naturalized citizens are less likely to register and vote than native-born citizens. Citizens born abroad in Europe and in Asia are 30 to 40 percent less likely to vote than those born in the U.S. While socioeconomic factors are prominent determinants of voting behavior among the native-born population, these factors are less important among the naturalized population. Among naturalized citizens, those with more education, a longer length of time at current residence, and a longer length of time in the U.S. are more likely to register and vote. The length of time spent in the U.S. is a prominent factor associated with both registration and voting among naturalized citizens. Why might this factor be so important in determining voting and registration among naturalized citizens?

We suspect that a major part of the explanation lies with the assimilation process of U.S. immigrants. The amount of time that a naturalized citizen has spent in the U.S. is a unique predictor of voting participation because, as discussed earlier, naturalized citizens who have been here longer are generally more attached and integrated into their communities and more accepting of U.S. institutions and social customs. However, outside of this classical assimilation paradigm, recent public debate over access to public services and citizenship also may have affected the propensity to vote among all naturalized citizens. For example, if the CPS had data on why people voted, we may have found that some naturalized citizens who sympathize with other immigrants may be more likely to vote than native-born citizens because they feel they must vote to protect the rights and public resources available to new immigrants who have not yet attained citizenship. We also may have found that the associations between other socioeconomic and demographic factors and voting and registration may have been significantly diminished or even eliminated if we could consider the reasons people vote. Other survey instruments which provide additional questions that ask why people voted may augment our understanding of these processes and further define the factors that influence voting participation among the naturalized citizen population.

Finally, this research provides a baseline for future analysis on the voting behavior of naturalized citizens and the differences in voting behavior between naturalized and native-born citizens.


1 Jennings (1983) found in bivariate analysis that parents with their own children under 18 years of age are less likely (47.3 percent) than other adults to vote (53.9 percent) in 1982.

2 Note this relationship would operate much like the relationship Wolfinger and Rosenstone (1980) found among Whites, Blacks, and Hispanics discussed previously.

3 Prior to 1994, questions regarding citizenship status were not asked regularly and were not available for every month. In addition, citizenship data were neither edited, nor allocated, nor consistently asked the same way in different survey years.

4 In their report, "The Foreign-born Population: 1996", Hansen and Faber discuss these data in more detail.

5 6339 cases or 7.5 percent of the unweighted cases were excluded from the total citizen analysis due to missing registration data and 6225 cases or 7.4 percent were excluded due to missing voting data.

6 391 cases or 10.9 percent of the unweighted cases were excluded from the naturalized citizen analysis due to missing registration data and 368 cases or 10.2 percent were excluded due to missing voting data.

7 These regions and countries did not have enough cases to allow for separate categories. The "Latin America" region includes Mexico. The "Other North America" region excludes Mexico and the United States and its territories.

8 Income for related individuals living in family households is family income, while income for unrelated individuals is personal income. Missing income information is coded to the mean income category and we add a dummy variable to indicate whether income data are missing.

9 We use the terms "Black" and "White" to refer to the joint race and ethnicity classifications of Black, not Hispanic and White, not Hispanic, respectively.

10 To our knowledge, no one has considered race in a multivariate model predicting registration. However, Leighley and Nagler (1992) found that Blacks are more likely to vote than Whites of similar socioeconomic status.


References

Alba, Richard and Victor Nee. 1997. "Rethinking Assimilation Theory for a New Era of Immigration." International Migration Review. Vol. 31:4. (Winter).

Callis, Robert. 1997. "Immigration Bolsters U.S. Housing Market." Census Brief. CENBR/97-4. Washington, DC: U.S. Government Printing Office.

Casper, Lynne M. and Loretta E. Bass. 1998. "Voting and Registration in the Election of November 1996." Current Population Reports P-20-504. Washington, DC: U.S. Bureau of the Census.

Hansen, Kristin A. 1997. "Geographical Mobility: March 1995 to March 1996." Current Population Reports P20-497. Washington, DC: U.S. Bureau of the Census.

Hansen, Kristin A. and Carol S. Faber. 1997. "The Foreign-Born Population: 1996." Current Population Reports P-20-494. Washington, DC: U.S. Government Printing Office.

Inglehart, Ronald. 1997. Modernization and Postmodernization Cultural, Economic, and Political Change in 43 Societies. New Jersey: Princeton University Press.

Jennings, Jerry T. 1993. "Voting and Registration in the Election of November 1992." Current Population Reports P-20-466. Washington, DC: U.S. Government Printing Office.
_____________. 1989. "Voting and Registration in the Election of November 1988." Current Population Reports P-20-440. Washington, DC: U.S. Government Printing Office.
_____________. 1985. "Voting and Registration in the Election of November 1984." Current Population Reports P-20-405. Washington, DC: U.S. Government Printing Office.
_____________. 1983. "Voting and Registration in the Election of November 1982." Current Population Reports P-20-383. Washington, DC: U.S. Government Printing Office.

Jones-Correa, Michael. 1998. "Different Paths: Gender, Immigration and Political Participation." International Migration Review. Vol. 32: 2. (Summer).

Kelley, Stanley Jr., Richard E. Ayres, and William G. Bowen. 1967. "Registration and Voting: Putting First Things First." American Political Science Review. 61: 359-79.

Kim, Jae-On, John R. Petrockik, and Stephen N. Enokson. 1975. "Voter Turnout Among the American States: Systemic and Individual Components." American Political Science Review. 69: 107-31.

Leighley, Jan E. and Jonathan Nagler. 1992. "Individual and Systemic Influences on Turnout: Who Votes?" Journal of Politics. 54: 718-40.

Lewis, Pierce, Casey McCracken, and Roger Hunt. 1994. "Politics: Who Cares?" American Demographics. Vol. 16, No. 10.

Smith, James P. and Barry Edmonston. 1997. National Academy Press. The New Americans: Economic, Demographic and Fiscal Effects of Immigration.

Squire, Peverill, Raymond E. Wolfinger, and David Glass. 1987. "Residential Mobility and Voter Turnout." American Political Science Review. Vol 81: 1. (March).

Teixeira, Ruy. 1992. The Disappearing American Voter. Washington, DC: The Brookings Institution.

U.S. Bureau of the Census. 1990. Census of the Population Social and Economic Characteristics. CP-2-1. Washington, DC: U.S. Government Printing Office.

Verba, Sidney and Norman H. Nie. 1972. Participation in America. New York: Harper and Row.

Wolfinger, Raymond. 1994. "Improving Voter Participation." in What to Do: Recommendations for Improving the Electoral Process. by Paul E. Frank and William G. Mayer. Boston: Northeastern University Press.

Wolfinger, Raymond and Steven J. Rosenstone. 1980. Who Votes? New Haven and London: Yale University Press.

Source: U.S. Census Bureau, Population Division,
Fertility & Family Statistics Branch

Authors: Loretta E. Bass and Lynne M. Casper
Last Revised: July 09, 2008 at 02:52:19 PM