"Anti-Racist" = Anti-White
Consider which cells are racist in the “Is it racist?” table:
Which cell would “anti-racists” most agree upon being racist? We don’t have to speculate, we can just run the study. I collected 1,014 responses in a Prolific sample of Americans representatively normed in terms of race, age, sex, and political views [6]. Participants were first asked “Is racism a strong synonym for evil?” and “Do you consider yourself to be an antiracist?” before randomly being assigned one of four follow-up questions:
“Suppose somebody asserted a generalization which portrays White people negatively. Would this be racist?”
“Suppose somebody asserted a generalization which portrays White people positively. Would this be racist?”
“Suppose somebody asserted a generalization which portrays Black people negatively. Would this be racist?”
“Suppose somebody asserted a generalization which portrays Black people positively. Would this be racist?”
Questions were assigned randomly to prevent people from trying to appear to be morally consistent. Responses were yes/no, with the order in which order “yes” and “no” were listed being randomized. Once participants answered a question and moved onto a subsequent one, participants were prevented from changing their previous answers. The question “Is racism a strong synonym for evil?” was coded as “evil”. The question “Do you consider yourself to be an antiracist?” was coded as “antiracist”. Follow-up question 1 was coded as “WN”. Follow-up question 2 was coded as “WP”. Follow-up question 3 was coded as “BN”. Follow-up question 4 was coded as “BP”.
The overall results are as follows:
With both groups lumped together, positive generalizations were rated as racist at the same rate as negative generalitions were rated as non-racist (odds ratio = 1.04, p > .25). As such, for the first analysis, each group had the “No” responses to their negative generalizations lumped with “Yes” responses to their positive generalizations and vice versa ( i.e. the four cases were column 1, column 2, column 3, & column 4). From there, the odds ratio of (column 2 / column 1) / (column 4 / column 3) was computed: This yields an odds ratio of 1.67, p < 0.001 (a significant europhobic bias in what is rated as being racist).
I then tried breaking it down by type of generalization. Negative generalizations were the primary concern, and negative generalizations of Whites had over twice the odds of being rated as non-racist (OR = 2.11, p < 0.001). As a point estimate, positive generalizations of Whites had 1.35 times the odds of being considered racist, but the p value is 0.094031. This is likely a type II error, as A) there was not the statistical power to detect an odds ratio of 1.35; and B) the difference between the two odds ratios did not reach statistical significance. The log odds of 2.11 (ln(2.11) = 0.74668794748) had a standard error of 0.2191442 while the log odds of 1.35 (ln(1.35) = 0.30010459245) had a standard error of 0.214825, with the difference in log odds (ln(2.11) - ln(1.35) = 0.44658) having a standard error of sqrt(0.2191442 ^ 2 + 0.214825 ^2) = 0.3068782. Dividing the difference in log odds by its standard error yields a z score of 1.45524626718, which integrates to give us a p-value of 0.072807.
All p-values reported above are one-sided.
Anti-Racists:
Among respondents who self-identified as anti-racists, negative generalizations had 3.279402 times the odds (p < 0.00001, standard error of log odds = 0.2736651) of being rated as racist if they applied to Blacks. Among participants who answered “no” when asked the same question, there was an insignificant effect in the other direction, with the odds ratio falling to 0.751263 (p > 0.1, standard error of log odds = 0.4022907). The difference in log odds (1.473661) had a standard error of 0.4865495, thus yielding a z score of 3.0288, which integrates to give us a p-value of 0.0012277050; self-described “anti-racists” had a significantly-more anti-White double standard in what negative generalizations they considered to be racist.
Still, the 0.75 odds ratio might be regarded as some amount of a double standard even if it isn’t in the same direction. Suppose we took its reciprocal (1 / 0.751263 = 1.331091); Is there a significant difference between groups in the absolute degree of a double standard? Well, the standard error of the difference in log odds actually stays the same while the difference in log odds falls to, 0.9016612, which divided by its standard error of 0.4865495 yields a z score of 1.853175, which integrates to give us a p-value of 0.0319285778.
Democrats, Republicans, & Indendents:
There were three analyses I could’ve done here; Democrats compared against Republicans, Democrats compared against non-Democrats, and Republicans compared against non-Republicans. I chose the third option because it was the one with the most statistical power; for comparing Democrats against Republicans, the difference in log odds had a standard error of 0.5936847; for comparing Democrats against non-Democrats, this standard error was 0.5251003; and for comparing Republicans against non-Republicans, this standard error was 0.4688662.
Again, Republicans had an insignificant anti-Black double standard on the order of an odds ratio of 0.696774 (p > 0.1) while non-Republicans had 3.76 (p < 0.00001) times the odds of rating a negative generalization as racist when they’re directed at Blacks instead of at Whites. Dividing the difference in log odds (1.686564) by the standard error of the difference yields a z score of 3.597112; this integrates to give us a p-value of 0.0001609223. That is, non-Republicans had a significantly-more anti-White double standard in what negative generalizations they considered to be racist.
Suppose however that we took the reciprocal (1 / 0.696774 = 1.43518558385) of the Republican odds ratio: Is there a significant difference between groups in the absolute degree of a double standard? Well, the standard error of the difference in log odds stays the same again while the difference in log odds falls to, 0.9639758, which divided by its standard error of 0.4688662 yields a z score of 2.055972, which integrates to give us a p-value of 0.0198925403.
White Supremacy Vs Asian Supremacy:
Suppose somebody asserts that “Asians are smarter than non-Asians.” Is this an example of an Asian-Supremacist worldview, or is it an example of a “model-minority myth” which is inadvertently harmful to Asians by suggesting that they don’t need any help in school due to their intelligence? I asked four more questions:
"Would assertions of White intelligence be examples of White-Supremacist views?"
"Would assertions of Asian intelligence be examples of Asian-Supremacist views?"
"Would it, in some way, be inadvertently harmful to Asians to generalize Asians as intelligent?"
"Would it, in some way, be inadvertently harmful to Whites to generalize Whites as intelligent?"
Q5 was coded as “WS”. Q6 was coded as “AS”. Q7 was coded as “AH”. Q8 was coded as “WH”. Of questions 5-8, participants were again randomly assigned one of the four so as to preclude trying to appear fair. Participants were also prevented from changing prior answers to whatever Q1-Q4 question they were assigned.
Generalizing Whites as intelligent had 2.640491 times the odds (p < 0.0000001) of being described as “supremacist” while assertions of Asian intelligence had 1.677493 times the odds (p = 0.0021842737) of being rated as inadvertently harmful.
Among self-described anti-racists, generalizations of White intelligence had 3.130256 times the odds (p < 0.0000001, SE = 0.2115705) while for everybody else, there was an insignificant effect in the same direction (OR = 1.441176, SE = 0.3912552, p > 0.1). The difference in log odds (0.7756551) had a standard error of 0.4447951, thereby yielding a z score of 1.743848, which integrates to give a p-value of 0.0405927570.
Are Racists Evil?
Prior research [2] has established that “racists” are dehumanized in contemporary society. As such, at the beginning of the survey I asked participants “Is racism a strong synonym for evil?” Among all participants, a slim majority (547 respondents) said “No” while among self-described antiracists, 407/784 said “Yes.” Self-described “antiracists” had 3.219787 times the odds (p < 0.00000000001) of agreeing that ‘racism is a strong synonym for evil.’
Summary:
“Anti-racists” think you’re evil if you aren’t Anti-White.
References:
Angleson, C. (2024). Supplementary Material for “Anti-Racist” = Anti-White. half-baked thoughts. Retrieved from not-equal.org/content/publicationSupplements/Angleson2024.supp.csv
Heiphetz, L. A., & Craig, M. A. (2020). Dehumanization and perceptions of immoral intergroup behavior. Oxford University Press. Retrieved from https://academiccommons.columbia.edu/doi/10.7916/w19k-se64/download
Write it up as a paper. You can always leave it as a preprint if you are too lazy to go thru review.