1/n Data from a survey experiment shows that white liberals significantly overestimate (M=31) the extent of discrimination against job applicants with black-sounding names. White conservatives, in contrast, were most accurate in their estimates (i.e. off by only 3 callbacks). 
cc @PsychRabble
Correction: both the red line and the bars refer to # of sent resumes, not callbacks received
Thus the average white liberal estimated that blacks had to submit over 30 resumes to receive 1 callback
(And White conservatives off by 3 resumes*, not callbacks)
Here’s the paper the author wrote with the data papers.ssrn.com/sol3/papers.cf…
In a subsequent part of the survey, respondents were asked to estimate the acceptance rate (41%) of Airbnb reservations under black-sounding names. All white respondents were way off, but white liberals significantly if modestly more so than moderates and conservatives.
White liberals (27%) vs. white moderates (32%); p=0.002
White liberals (27%) vs. white conservatives (31%); p=0.03
Standard deviation of white estimates=17%.
White liberals (27%) vs. white conservatives (31%); p=0.03
Standard deviation of white estimates=17%.
Question wording for the above
For those wondering how the black and black liberal resume-callback estimates fared those of white liberals, the former were also significantly more accurate.
@MattGrossmann I was mentioning on your podcast how difficult it'd be to empirically assess whether white liberals overestimate the extent of discrimination. Here's some suggestive data I found on TESS (and the corresponding paper papers.ssrn.com/sol3/papers.cf…)
