Racial attitudes and national anthem protests
Some research includes measures of attitudes about certain groups but not about obvious comparison groups, such as research that includes attitudes about Blacks but not Whites or includes attitudes about women but not men. Feeling thermometers can help avoid this, which I'll illustrate with data from the Democracy Fund Voter Study Group's Views of the Electorate Research (VOTER) Survey.
---
The outcome is this item, from the 2018 wave of the VOTER survey:
Do you approve or disapprove of football players protesting by kneeling during the national anthem?
I coded responses 1 for strongly approve and somewhat approve and 0 for somewhat disapprove, strongly disapprove, don't know, and skipped. The key predictor was measured in 2017 and is based on 0-to-100 feeling thermometer ratings about Blacks and Whites, coded into six categories:
* Rated Whites equal to Blacks
---
* Rated Whites under 50 and Blacks at 50 or above
* Residual ratings of Whites lower than Blacks
---
* Rated Blacks under 50 and Whites at 50 or above
* Residual ratings of Blacks lower than Whites
---
* Did not rate Whites and/or Blacks
The plot below controls for only participant race measured in 2018, with error bars indicating 83.4% confidence intervals and survey weights applied.
The plot suggests that attitudes about anthem protests associated with negative attitudes about Blacks and with negative attitudes about Whites. These are presumably obvious results, but measures such as racial resentment probably won't be interpreted as suggesting both results.
---
NOTE
1. Stata code and output. The output reports results that had more extensive statistical control.