How much racial resentment can be explained by ANES measures of antiBlack animus?

Racial resentment and symbolic racism are terms used to describe a set of measures used in racial attitudes research, including statements such as "Irish, Italians, Jewish and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors". This item and at least some of the other racial resentment items confound racism and nonracial ideology; in this "special favors" item, an individualist who believes that everyone should work their way up without special favors would select a response on the same side of the scale as an antiBlack racist who believes that only Blacks should work their way up without special favors.

Feldman and Huddy (2005) concluded that "racial resentment is an inadequate measure of prejudice because it confounds prejudice and political ideology" (p. 181), which is consistent with factor analysis of racial resentment items (Sears and Henry 2003: 271). Some research has addressed this confounding with what Feldman and Huddy (2005: 171) call the multivariate approach, in which the analysis includes statistical control for related ideological values. The logic of this multivariate approach is that racial resentment confounds ideology and antiBlack animus so that controlling for ideology should permit the residual association of racial resentment to be interpreted as the association due to antiBlack animus.

The analysis below approaches from the opposite direction: racial resentment confounds ideology and antiBlack animus so that controlling for antiBlack animus should permit the residual association of racial resentment to be interpreted as the association due to ideology. Moreover, if controls for ideology and for antiBlack animus are both included, then the association of racial resentment with an outcome variable should be zero. But this is not even close to being true, as illustrated below in a figure that reports the association of racial resentment with racial or possibly racialized outcome variables, using different sets of statistical control.

In each panel above, the top estimate indicates the association of racial resentment with the outcome variable controlling for only demographics. The second and third estimates respectively indicate the association of racial resentment with outcome variables after controls for demographics and racial attitudes and after controls for demographics and ideology. The fourth and fifth estimates respectively indicate the association of racial resentment with outcome variables after controls for demographics, ideology, and racial attitudes and after controls for demographics, ideology, and racial animus. The key comparison is between the third estimate and the fourth and fifth estimates: the measures of racial attitudes and racial animus had relatively little impact on the racial resentment estimate once the controls for ideology were included in the analysis. For example, in the top left panel, the coefficient for racial resentment was 0.51 controlling for demographics and ideology, was 0.48 controlling for demographics, ideology, and racial attitudes, and was 0.52 controlling for demographics, ideology, and racial animus. In a common racial resentment association analysis, the 0.51 coefficient controlling for demographics and ideology would be assigned to antiBlack animus, but the addition of seven racial attitudes controls accounted for only 0.03 of the 0.51 coefficient and the inclusion of six antiBlack animus controls did not even reduce the 0.51 coefficient. (see the Notes below for more description on the measurements).

A reasonable critique of the above analysis is that racial resentment taps a form of antiBlack racism that is not captured or is not well captured in the included measures of racial attitudes and racial animus. But, from what I can tell, that is an equally valid criticism of analyses that control for ideology: the nonracial ideology captured in racial resentment measures is not captured or not well captured in the included measures of ideology.

NOTES

1. The sample for the analysis was the 3,261 non-Hispanic Whites who completed face-to-face or online the pre- and post-election surveys, conducted between 8 September 2012 and 24 January 2013, and who were not listwise deleted from a model due to missing data for a variable. Each variable in the analysis was coded to range from 0 to 1. Linear regressions without weights were used to predict values of the outcome variables.

The racial resentment measure summed responses to the four ANES 2012 racial resentment items. Models included demographic controls for participant sex, marital status, age, education level, and household family income. Ideological controls were self-reported partisanship, self-reported ideology, an item about guaranteed jobs, an index of attitudes about the role of government, a moral traditionalism index, an authoritarianism index, and an egalitarianism index.

One set of models included seven controls for racial attitudes: a feeling thermometer difference of ratings of Whites and ratings of Blacks, a rating difference for Blacks and for Whites in general on a laziness stereotype scale, a rating difference for Whites and for Blacks in general on an intelligence stereotype scale, an item rating admiration of Blacks, an item rating sympathy for Blacks, an item measuring the perceived political influence of Blacks relative to Whites, and a difference in ratings of the level of discrimination in the United States today against Whites and against Blacks. Another set of models included six dichotomous controls that attempted to isolate antiBlack animus: a more than 20-point feeling thermometer rating difference in which Whites were rated higher than Blacks and with Whites rated at or above 50 and Blacks rated below 50, a rating of Blacks as lazier in general than Whites, a rating of Whites as more intelligent in general than Blacks, an indication of never feeling sympathy for Blacks, an indication that Blacks have too much influence in American politics but Whites don't, and an indication that there is no discrimination against Blacks in the United States today but that there is discrimination against Whites in the United States today.

2. Code for the analysis is here.

3. Results for the 2016 ANES are below:

4. Code for the 2016 ANES analysis is here.

5. Citations:

American National Election Studies (ANES). 2016. ANES 2012 Time Series Study. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2016-05-17. https://doi.org/10.3886/ICPSR35157.v1.

American National Election Studies, University of Michigan, and Stanford University. 2017. ANES 2016 Time Series Study. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2017-09-19. https://doi.org/10.3886/ICPSR36824.v2.