Get Out on Behalf of Your Group: Electoral Participation of Latinos and Asian Americans

For review, see de la Garza (2004) on political behavior of Latinos. On Asian-American politics, the literature is somewhat limited, in part due to scarcity of reliable survey data. For notable exception, see Lien et al. (2004).

On the last point, Gay (2001) reports a contrasting finding, showing that black elected officials rarely increase political engagement among blacks while they do negatively affect whites’ political involvement. However, Gay’s finding is based on the ecological inference from aggregate data, and thus may not be directly comparable to Bobo and Gilliam’s individual-level analysis. In addition, recent analysis by Banducci et al. (2004) and Barreto et al. (2004) generalizes the empowerment argument, by showing that descriptive representation secured through the creation of majority-minority districts increases political efficacy and participation of Maori voters in New Zealand and Latino voters in the United States, respectively.

At the same time, however, Browning et al. (1986) also finds that electoral mobilization of Latinos is greater where Latinos have previously experienced higher levels of non-electoral protest activities, which is a strong function of the absolute, rather than relative, size of the group.

On the other hand, Oliver (2001) finds that Latinos’ engagements in other types of civic activities, especially instrumental activities directed toward a particular goal such as contacting local officials or attending community board meetings, increase in non-white places than in predominantly white places, and interprets this as reflecting the combination of greater social problems and fewer public resources in racially mixed places.

In fact, the concept of relational goods plays an essential role in this paper as well, and the relationship between relational goods and the group size is explicitly derived from simple model of participation developed in the next section.

In fact, even in the tradition of rational-choice theory, the idea that people have two sets of preferences, one of which is “altruistic” preferences containing the individual’s perception of the utilities of others, is not new from its early generations (e.g., Harsany 1969, 1977; Margolis 1981) to more recent game-theoretic models (Coate and Conlin 2004; Feddersen and Sandroni 2006; Jankowski 2007. For review, see Feddersen 2004).

One should note that perceiving group-level benefits is different from the sense of duty to help others which does not depend on election outcomes. In fact, both sets of benefits here are of instrumental nature as people who care about the welfare of others will vote only if they think one of the alternatives is superior (Fowler 2006, p. 675).

For instance, consider the following thought experiment. Suppose that there are N = N ₁ + N ₂ +··· + N _G number of voters. As the probability of a vote being decisive is inversely proportional to N (Gelman et al. 1998), the utility of voting for a typical member i in group g can be written as \(U_{i0}=P\alpha\bar{B_{g}}N_{g}=(K/N)\alpha\bar{B_{g}}N_{g}\) for some constant K which represents the competitiveness of the election. Now, add s number of new voters to group g, holding everything else constant. In this situation, the new utility is given by \( U_{i1}=\{K/(N+s)\}\alpha\bar{B_{g}}(N_{g}+s)\). Then,Therefore, an increase in the group size provides a greater incentive for members to participate by increasing group-level benefits large enough to offset a decrease in the probability that one’s vote matters. In fact, this consideration should be particularly salient among Latinos and Asian: as many Latinos and Asian Americans are non-citizens and thus ineligible to vote, an increase in the number of Latinos and Asian Americans affects P to a lesser extent than a similar increase in the number of whites or African Americans does.

The median household income is one of the most direct measures of the aggregate socio-economic status, along with the educational attainment. However, the large proportion of immigrants among Latinos and Asian Americans makes the educational attainment potentially improper measure of the socio-economic status of their groups, as they are primarily educated outside of the U.S. and thus their educational attainment does not necessarily correspond to the extent to which they have advanced socio-economically in the U.S. society.

In this paper, we take county as a measure of context. Obviously, the size of a county varies across different areas, and county as a measure of context makes more sense in one place than in another. In fact, this is an inevitable problem in any contextual analysis, since every individual may perceive a context differently. Nevertheless, we believe that taking counties as a unit of analysis is a reasonable compromise given this paper’s national scope, especially considering data availability, without a strong theoretical reason that indicates otherwise.

Nevertheless, it is still questionable how representative these counties are. In fact, because of confidentiality restrictions, the CPS intentionally identifies only more populous, and thus more racially diverse, counties: for instance, the average population size of the counties identified in the CPS 2000 is about 9.5 times larger than that of unidentified counties; the counties identified in the CPS 2000 on average contain 97,607 Latinos and 31,883 Asian Americans, while unidentified counties contain only the average of 4,808 Latinos and 1,077 Asian Americans. Obviously, this skewed distribution of counties in the CPS data may bias the results reported in this paper to some extent, especially in a way that makes individual residents more responsive to racial contexts. If this is indeed the case, it is possible that the actual effect of racial contexts on individual participation can be smaller than what is reported in this paper. Therefore, the findings of this paper should be taken with caution and validated further with additional research. Nevertheless, the possible bias should be less significant in the CPS data with its broader geographic coverage than in other datasets that explicitly draw the sample from a limited number of places where a very large number of Latinos and Asian Americans are concentrated.

In fact, while the turnout rates calculated from the ANES 2000 is 76%, the turnout rates in the CPS 2000 is 67% and therefore much closer to the actual turnout rates.

For ease of interpretation, in the estimation, all of the group-level variables are centered around 0 by subtracting their respective means.

In addition, very high correlation between Group Size and Heterogeneity among Asian Americans can be another factor that makes it difficult to find significant interaction effects between the two: the correlation between Group Size and Heterogeneity is 0.79 among Asian Americans, compared to 0.55 among Latinos. On the other hand, the correlations between Group Size and Group Income are negligible for both Latinos and Asian Americans.

According to the Census 2000, the median household income of Latinos is $33,676, while that of Asian Americans is as high as $51,908. The median household income of Asian Americans is even higher than that of non-Hispanic whites, which is $45,367.

Specifically, our hypothetical Latino is 38-year-old married woman of Mexican origin, who is a native-born citizen. She is a high-school graduate, employed with family income of $30,000-$35,000, and owns a house where she has lived at least 5 years. Our hypothetical Asian American is 43-year-old married woman of Chinese origin, who is a naturalized citizen. She is employed with family income of $50,000–$60,000, has some college education (but no degree), and owns a house where she has lived at least 5 years.

We could not do a similar re-analysis with respect to Asian Americans, as Asian Americans include multiple language groups.

Although the number of counties covered by the VRA is less than 30% of all the counties considered here, more than three quarters of Latinos in the sample reside in those counties.