Ecological Assimilation: Race, Ethnicity, and the Inverted Gap of Environmental Concern

Abstract

Utilizing data from the 2010 General Social Survey, three theoretical perspectives are used to explain differing levels of environmental concern and sustainable practices among native- and foreign-born race and ethnic groups in the United States. The most prominent pattern resulting from the analysis is that of ecological assimilation among the Mexican-origin population, with the foreign-born portion of this group being much more likely than their native-born counterparts or whites to both engage in sustainable practices and express a willingness to make sacrifices for the sake of the environment. There is evidence for a concern gap between African Americans and whites in the responses to questions tied to economic progress and standard of living reductions, though the inverted concern gap implied by the environmental justice literature is also apparent in responses to questions about perceived environmental risks where the majority of non-whites in the survey showed higher levels of concern than whites.

Keywords:

• environmental attitudes and concerns
• environmental justice
• ethnicity
• immigration and the environment
• race and the environment

Notes

¹Listwise deletion of our initial 1430 cases reduced sample sizes to near 900 in a few of our models. Subsequent t-tests on dependent variables showed significant differences between group means based on missing and nonmissing values in independent variables with large numbers of missing values such as household income. Rather than rely on listwise deletion, we ultimately chose to use multiple imputation to replace missing values. We performed imputation with chained equations using Stata 13. We made the missing at random (MAR) rather than the more restrictive missing completely at random (MCAR) assumption, and used a massive rather than constrained approach which predicts missing values on the basis of all other variables, including dependent variables. Ten imputed data sets were created. As is common practice, we dropped cases missing on the dependent variable but used imputed scores for all missing values in the independent variables (von Hippel 2007; White, Royston, and Wood 2010). This is evident in Table 1, where all independent variables have an N of 1430, and all dependent variables have Ns reflecting cases with nonmissing values for that variable in the data set.

²The GSS environmental module also includes four dichotomous activism questions regarding membership in an environmental organization, giving money to an environmental cause, signing an environmental issue petition, and having attended an environmental demonstration over the past five years. Only one of these had a sufficient number of affirmative responses to allow us to perform meaningful logistic regression analysis: signing an environmental issue petition. Using a model similar to the OLS and ordered-logit regressions in Tables 234, we found that only two of our race and ethnic categories had statistically significant outcomes: US-born African Americans and foreign-born born other race/ethnicities were both less likely to have signed an environmental issue petition over the past five years. Whether this result reflects a lack of interest in petition signing among certain groups or the avoidance of some groups by people requesting signatures is unclear, and for this reason we opted not to include it along with our principal findings.

³As a point of contrast, we ran iterations of all the models in Tables 3 and 4 that included the three environmental concern composite variables featured in Table 2 and three additional single indicator variables: a measure of postmaterialism; a gauge of agreement with the statement “environmental issues directly affect me”; and a question about concern over population growth. In no case did this appreciably change the magnitude of race and ethnic odds ratios, and in only four cases did this result in a change in coefficient p values: coefficients for US-born other race/ethnicities, Mexican immigrants, and foreign-born Latinos in the buy chemical-free produce model and for Mexican immigrants in the drive less model all exceeded the p < .05 significance cutoff.