The main independent variables measure the race of the respondent. Each respondent was
asked, “What race do you consider yourself to be? White, Black or African American, Asian, or
Some other race?” Thirty-eight percent of respondents self identified as White, 5 percent self
identified as Black or African American, and one-third of the respondents self identified as
Latino/Hispanic, despite not being prompted to choose either pan-ethnic label. Twenty-two
percent of respondents self identified as some other unspecified race.
Whereas some studies
group together respondents who racially identify as “Latino” and “Some other race” (e.g. Logan
2003; Rodriguez 1992), I examine these groups separately so not to underestimate the complex
nature of racial identity formation in the Latino community and confound the results.
The model also includes several control variables.
Because education, income, age,
gender, martial status, language, and nativity are all significant predictors of Latino political
behavior (e.g. Hero and Campbell 1996; Hardy-Fanta 1993; DeSipio 1996b; Shaw et al. 2000;
Johnson et al. 2003), they are included in the model. Female, Married, and U.S. born are binary
variables, and Education, Income, and Age are all scaled from low to high. Spanish at home is
coded on a five point scale where lower values indicate that English is primarily spoken at home
and higher values indicate greater Spanish language use. It is expected that the primary use of
Spanish represents respondents’ access to social and community resources that enable political
participation (Johnson et al. 2003). Thus the relationship between Spanish at home and political
participation should be positive and statistically significant.
I also include two variables to measure political attitudes that predispose individuals to
overcome the costs of participation. Political efficacy is a necessary prerequisite for the exercise
of political and civic activities (Campbell et al. 1954; Abramson and Aldrich 1982). Efficacy is an
14
One respondent (.1%) identified as Asian. This group was excluded from the analysis because there were
too few to examine separately. In addition, 16 respondents who refused to answer the question (1.2%) were
also excluded from the analysis.
15
Clearly these two groups have in common a desire to choose culture as a basis for racial identification
rather than an intermediate racial category. Yet they choose different ways to express their rejection of
standard racial categories. Using multinomial regression, Stokes-Brown (2006) finds that there are
significant differences among the groups when predicting the acceptance of “Latino” and “Some other
race” as racial identities.
16
Checks for multicollinearity showed no large correlations between any two variables used in the
equations throughout the paper.
9
Convention