How do you define ‘racism’? Survey results

I made a google forms survey asking people a bunch of demographic information, and what they meant by the word ‘racism.’ I kept it simple and only gave three options – “Discrimination on the basis of race”, “Prejudice + power”, and “other.”

I posted the survey all over, but got respondents mainly from 4 locations – fetlife (1050), my blog (234), reddit (5045), and twitter (963). Total N (after cleaning up a few fake-looking responses) was 7,381.

“How do you define racism?” total responses

I anticipate reddit and fetlife as being ‘less biased’ than my blog and twitter, because the vast majority of people who follow me on those platforms don’t know anything about my politics, polls, personality, etc.. If anything I’d assume fetlife as a source leans more liberal than average, as it’s an extremely sex-positive community source.

Here’s what people said, broken down by all the demographic stuff.

Smallest bin was 63+, with 102 responses

Correlation of age and answering “discrimination on the basis of race” was r=-0.06, very small, but definitely significant at a likelihood ratio of 3 million to one.
Likelihood ratios mean basically how much more likely the observed correlation is than a correlation of 0.

Smallest bin was “didn’t finish high school” at 189, second smallest was doctorate’s at 314.

This one is much less surprising to me – the more educated you are, the more likely you are to report “prejudice + power” as how you define racism. Probably confounded by age, I didn’t check for this. Education was weakly anti correlated with answering discrimination on the basis of race’, r=0.07, LR 1/8 million.

Bin size: Asian=394, black=350, hispanic/latino=747, native american=79, white=5131

I’m not sure how to interpret the ‘native’ results here; maybe selecting both this ethnicity and ‘other’ were troll responses? I did clear out troll responses I noticed so I think this is a bit less likely, but who knows.

Obviously the other outlier here is “prejudice + power’s” black/african american’s 29.3% (as compared to e.g. white’s 17.4%). This seems pretty in line with the discourse. It’s interesting that the discourse doesn’t seem to have touched the Asian or Hispanic populations I sampled!

I didn’t include ‘pacific islander’ (few responses) or ‘other’ (???) category for this graph.

Bin sizes: amab=174, afab=46, female=351, ftm=29, male=6709, mtf=69

Keeping in mind some of the bins here are small (afab and ftm especially), these results also aren’t super surprising. I suspect some degree of this is confounder, though I’m not sure in which direction – a higher percentage of female responses came from fetlife, which also registered as a more “prejudice+power” source (26%, as opposed to twitter’s 10%)

Over 100 responses in each bin here; I combined a few nearby low-response locations and didn’t include the few remaining locations with low response rates, such as Africa.

I’m surprised to see ‘middle’ outranking ‘east’ for USA’s “prejudice + power” responses; although it’s only a 3-percentage-point difference, the bins were big; 1192 for ‘east’ and 637 for ‘middle.’

But besides this, I’m unsurprised to see the US leading the world in the “prejudice+power” definition’; it also seems related to discourse which is so far a mostly US-centric phenomenon.

I also asked respondents to pick which statement they agreed with more:
Definitions are: Descriptive: We know the correct definition of a word based on observing the way it’s used – we describe” its use”
Definitions are: Prescriptive: We know the correct definition of a word based on what we’ve decided it should be – we ‘prescribe’ its use

10% more people who said ‘discrimination on the basis of race’ saw definitions as descriptive, compared to those who said definitions were more prescriptive (r=0.09, likelihood ratio 2e+12). A pretty weak but real effect.

There weren’t many interesting other correlations (including education!) with this question besides the ‘how do you define racism’ one , though there were two even weaker ones with with age and twitter as a source (r=0.04, tend to say ‘discrimination on basis of race’ more)

If you’d like to doublecheck my work or play with the data yourself, here’s the raw responses. They should be fully anonymous, given the girth of this sample size, but I binned the age data anyway to be safe. If I’ve made any mistakes please let me know and I’ll update this post!

Leave a Reply