It seems that (almost) every race/ethnicity/national-origin -- including all subsets of European -- give a negative bias? How can this be? Is the zero-level (baseline) itself somehow skewed? Does this overall fact hint that the underlying studies -- as a group -- just 'can't' be legit? That each one found what it set out to find? Publication bias may be behind it?
With the race/ethnicity/nationality categories, they're always comparing callbacks for minority groups relative to majority groups. So, for instance, for the Northern and Western European groups, they might look at discrimination against English people in Finland or French people in Germany. Likewise, for the Eastern European groups, they might look at Russians in Finland or Poles in Sweden. (See Footnote 22 in the paper.) The key finding, IMO, is that some minority ethnicities face greater discrimination than others.
The zero baseline isn't skewed, because the studies always compare callback rates for the target group against a relevant comparator group - e.g., minority ethnicity vs. majority, female vs. male, disabled vs. non-disabled. If the callback rates were equal, the point estimate would be zero.
At first glance, it seems a bit dodgy that all groups other than females fall to the left of the zero point. But in each case, discrimination against the group shown also represents discrimination in favor of the comparator group. So, discrimination against disabled individuals is discrimination in favor of the non-disabled, discrimination against older individuals is discrimination in favor of other age groups, etc.
Also, for many categories - the ones that aren't in blue - the findings aren't significantly different from zero. That means they didn't just find what they expected to find. Ditto for the (small) bias in favor of females.
There may be publication bias, but we can't infer it just from the pattern of results shown in the graph.
It's probably legit to conclude from this that -- virtually without fail -- a minority group will probably be 'discriminated' against in this way by the majority. But I don't think I can agree with what you've said was the 'key finding': that some minority ethnicities (necessarily) get this worse than others.
It's true that the anti-European discrimination is pretty minor -- but we're observing predictably-minor INTRA-European sentiments (Poles in Sweden or Brits in Finland). Aren't we comparing that to how Arabs or blacks get treated in America or Europe or Japan? This is neither 'mirror image' nor is it 'apples to apples.'
Imagine that we'd compiled a bunch of studies on how Nigerians were treated in Sierra Leone; Ghanans in Benin; or Saudis in Jordan. We'd probably find and conclude that discrimination against blacks and Arabs appeared to be quite minor. And it'd seem silly to compare that to studies on how Germans were treated in China, or how Frenchmen fare in Pakistan: we'd conclude that people of European descent face much greater discrimination than Arabs or Africans.
Does nobody study how European/White-American job applicants would fare in central Africa simply because we already know the answer wouldn't be good?
Hi Jeff. Yep, I agree. I did consider adding that my "key finding" applies only to the regions where the studies have been done, but didn't want to overburden the reply with too many qualifications. It's an important point, though. Aside from the fact that the rank order of biases will vary from place to place, conducting the studies only in the West could give people the impression that the biases are a uniquely Western pathology, as opposed to reflecting panhuman ingroup-outgroup bias.
It would be interesting to see if this pattern also applies to the final employment decision.
Do you use a particular tool or app to stay updated on research topics that interest you? If so, I’d love to hear your recommendations.
I’ve noticed that many platforms I’ve used, like ResearchGate and Academia, often have a lot of noise due to their lack of effective personalization and filtering options for the newsfeed.
Yep, that would be very interesting. From what I've read, the more individuating information that employers have about a candidate, the less likely it is that demographic stereotypes will influence their decision. That's good news - but of course, you've got to get past the hurdle of the callback first. A reason to do blind evaluation of candidates at the early stage!
I use multiple tools to find articles. I get table-of-contents alerts from lots of journals I like; I use Twitter, Facebook, and Reddit; and I keep an eye out for interesting studies mentioned in books, papers, and podcasts. Like you, I've never found ResearchGate or Academia.edu particularly useful.
It seems that (almost) every race/ethnicity/national-origin -- including all subsets of European -- give a negative bias? How can this be? Is the zero-level (baseline) itself somehow skewed? Does this overall fact hint that the underlying studies -- as a group -- just 'can't' be legit? That each one found what it set out to find? Publication bias may be behind it?
Hi Jeff, good questions.
With the race/ethnicity/nationality categories, they're always comparing callbacks for minority groups relative to majority groups. So, for instance, for the Northern and Western European groups, they might look at discrimination against English people in Finland or French people in Germany. Likewise, for the Eastern European groups, they might look at Russians in Finland or Poles in Sweden. (See Footnote 22 in the paper.) The key finding, IMO, is that some minority ethnicities face greater discrimination than others.
The zero baseline isn't skewed, because the studies always compare callback rates for the target group against a relevant comparator group - e.g., minority ethnicity vs. majority, female vs. male, disabled vs. non-disabled. If the callback rates were equal, the point estimate would be zero.
At first glance, it seems a bit dodgy that all groups other than females fall to the left of the zero point. But in each case, discrimination against the group shown also represents discrimination in favor of the comparator group. So, discrimination against disabled individuals is discrimination in favor of the non-disabled, discrimination against older individuals is discrimination in favor of other age groups, etc.
Also, for many categories - the ones that aren't in blue - the findings aren't significantly different from zero. That means they didn't just find what they expected to find. Ditto for the (small) bias in favor of females.
There may be publication bias, but we can't infer it just from the pattern of results shown in the graph.
Thanks for the detailed response!
It's probably legit to conclude from this that -- virtually without fail -- a minority group will probably be 'discriminated' against in this way by the majority. But I don't think I can agree with what you've said was the 'key finding': that some minority ethnicities (necessarily) get this worse than others.
It's true that the anti-European discrimination is pretty minor -- but we're observing predictably-minor INTRA-European sentiments (Poles in Sweden or Brits in Finland). Aren't we comparing that to how Arabs or blacks get treated in America or Europe or Japan? This is neither 'mirror image' nor is it 'apples to apples.'
Imagine that we'd compiled a bunch of studies on how Nigerians were treated in Sierra Leone; Ghanans in Benin; or Saudis in Jordan. We'd probably find and conclude that discrimination against blacks and Arabs appeared to be quite minor. And it'd seem silly to compare that to studies on how Germans were treated in China, or how Frenchmen fare in Pakistan: we'd conclude that people of European descent face much greater discrimination than Arabs or Africans.
Does nobody study how European/White-American job applicants would fare in central Africa simply because we already know the answer wouldn't be good?
Hi Jeff. Yep, I agree. I did consider adding that my "key finding" applies only to the regions where the studies have been done, but didn't want to overburden the reply with too many qualifications. It's an important point, though. Aside from the fact that the rank order of biases will vary from place to place, conducting the studies only in the West could give people the impression that the biases are a uniquely Western pathology, as opposed to reflecting panhuman ingroup-outgroup bias.
It would be interesting to see if this pattern also applies to the final employment decision.
Do you use a particular tool or app to stay updated on research topics that interest you? If so, I’d love to hear your recommendations.
I’ve noticed that many platforms I’ve used, like ResearchGate and Academia, often have a lot of noise due to their lack of effective personalization and filtering options for the newsfeed.
Yep, that would be very interesting. From what I've read, the more individuating information that employers have about a candidate, the less likely it is that demographic stereotypes will influence their decision. That's good news - but of course, you've got to get past the hurdle of the callback first. A reason to do blind evaluation of candidates at the early stage!
I use multiple tools to find articles. I get table-of-contents alerts from lots of journals I like; I use Twitter, Facebook, and Reddit; and I keep an eye out for interesting studies mentioned in books, papers, and podcasts. Like you, I've never found ResearchGate or Academia.edu particularly useful.