It's even worse than that since the experiment is based on appearance, but claims it's about race, and race is a social construct. The labels "black" and "white" in the western world are practically meaningless today. "Mixed-race" people are increasingly common, or at least recognized, and the phenotypes associated with "white" and "black" have always been present in people around the world without necessarily having much if any European or African ancestry.
EDIT: and similar to what another comment said, this ignores all other notions of "race" besides "black" and "not black"
> When subsetting to non-Black pedestrians, who make up 93% of our sample, our results do not meaningfully change.
Huh. Not... quite what I think most people would expect to find in the paper, from the headline and abstract. Am I reading that right? Black pedestrians behaved similarly to non-black pedestrians, as far as staying farther from a young black man standing on the sidewalk, than from a young white man doing the same? Or have I misunderstood?
Look at figure S21, "results don't meaningfully change" can be a result of an effect not existing, or a result of the authors failing to have sampled enough black people for results to meaningfully change.
While sample size is low, it sure looks to me like the black population stayed closer to the black confederates than the rest of the population.
Ah, yeah, wondered if that being a smaller sample size might be the cause. If the effects are actually significant, seems worth re-running with different pedestrian demographics. Could be interesting.
[EDIT] Interesting, I mean, because there are a lot of dynamics that might be sussed out by this. Might find that e.g. being different from most other pedestrians enhances this effect even for pedestrians of the same race as the confederate, or any number of other quirky effects (or the opposite, or none of that! Who knows?). Or that under certain circumstances, the effect vanishes. Lots to play with, here. They call out a bunch of other studies in the paper, but it'd be cool to see the same study run several times in very-different locations.
Regardless of cause, it does seem that Black men in Manhattan of a certain age are avoided more by pedestrians as measured by distance than White men in Manhattan of a certain age. That sucks.
Still a few issues:
First, they clearly push the "Racial avoidance hypothesis" but Figure S21 seemed to bolster the "Pedestrian outgroup salience hypothesis" over the "Racial avoidance hypothesis" since black people didn't seem to avoid other black people. The author just seemed to brush off this hypothesis as plausible but not having enough evidence to support it while pushing the favoured racial avoidance hypothesis. Yet how can he know if his hypothesis is correct without ruling out the pedestrian outgroup hypothesis or knowing if perhaps both hypothesis are correct and contributing to the effect seen?
Second, this entire quote:
>Pedestrians on our selected Manhattan street corners are likely to be wealthier, better educated, and more politically Left-leaning than the average American. These traits may, on the whole, make subjects less prone to racial avoidance.
This is directly contradicted by figure S19, which found that people gave a wider berth to black people in the wealthier upper east-side neighbourhood than the poorer midtown neighbourhood. I also should point out the bias of the author, a wealthy educated man, talking shit about how poorer less educated populations (say - black people) may be more racially avoidant.
Third, this entire quote:
>One alternative interpretation of our findings is that pedestrians are “giving people space” rather than avoiding them. We reject this interpretation, and point to the literature on proxemics, which focuses on how people use space during social interactions, and demonstrates physical avoidance of stigmatized groups in a variety of contexts, linking patterns of avoidance to implicit measures of prejudice
Anyway this is based on doi: 10.1007/7854_2015_431. It mentions a study where it's specifically referring to response latency as the "implicit measure of prejudice" correlated to less eye contact, more blinking, more distancing from interviewers, more distancing from virtual moroccans, and more shooting of virtual black people.
I'm confused as to on what grounds this hypothesis was rejected. Even if you establish that the people who were distancing themselves from black's were indeed prejudiced against blacks, that doesn't prove that they're not motivated by giving them more space. Differential treatment based on race is inherently prejudicial to begin with, so this entire mode of analysis makes no sense. On top of how nonsensical this entire analysis is, it's very questionable what the linked study actually links patterns of avoidance to valid "implicit measures of prejudice", most research on implicit prejudice is very controversial.
Fourth, S5.5.2 found that females both stayed further from black confederates AND confederates in general. However since the study DIDN'T examine how far men would stay away from women confederates, we can't say if it's because they're men or because they're a different gender. Now we didn't establish this, so we shouldn't jump to any conclusions, but imagine if we just followed the lead of the authors here and jumped to the conclusion this was gender bias and said:
>Avoidance experienced by MALE individuals, day in and day out, likely imposes a psychological toll... ...eroding mental and physical health. The very public, widespread, and chronic occurrence of pedestrian MALE avoidance may have spillover effects that influence others’ behavior in subtle but destructive ways. For example, law enforcement – even if trained to recognize their own implicit biases – may implicitly or explicitly detect bystander behavior and overestimate the level of threat posed by MALE Americans in ambiguous situations.
Oddly the authors didn't opt to draw this conclusion nor give any explanation as to why we shouldn't! They found a racial bias and a gender bias, piled on empathy for the former, and just totally omitted the latter, which itself reflects bias. This is beyond the prejudicial attitude that poor uneducated conservatives might be worse (but not better) than Manhattanites.
This is a really stupid paper. Their effect sizes are tiny (a 3-4" deviation??). I don't care how accurately you estimated it.
Also, they don't talk about the things that are obviously driving this. It's not race per se, it's things that correlate with race. If you live in an environment where all black and non-black people have the same distributions of behavior, you won't learn to treat people differently on the basis of being black vs. non-black. It's the fact that black Americans are ~7-10x as violent[0] as non-black Americans that is driving the behavior the paper observes. Obviously. The fact that they don't touch this obvious driver makes them, in my view, cowardly, bad scientists out for a cheap thrill.
According to Wikipedia, Staten Island has the lowest percentage of black residents, at 9.7%. NYC overall is 26.6% black. You’d have to be either pretty deliberate or pretty careless to work with a sample population of only 7% black folks.
Ghastly. How can this possibly be published by Nature? Flawed from the start since reliably detecting "race" is more complicated than just physical appearance, yet the data is based on surveillance cameras. Also, culture may be just as if not more significant. Maybe New York just has a lot of assholes and low individualism. Also, how can a mere correlated difference of reaction be considered "racist"?
They placed the people the avoidance of whom was being measured (the authors' "confederates", as the paper calls them) so they knew for a fact what those guys looked like. It seems like they also didn't include data points if they didn't think the footage was clear enough.
> Maybe New York just has a lot of assholes and low individualism.
But they measured a difference between behavior around white and black "confederates".
> Also, how can a mere correlated difference of reaction be considered "racist"?
It's not exactly concrete proof per se, but I think a well-constructed study of this sort (I am not prepared to vouch for this one) could at least reach a finding that points strongly in that direction.
If they do so judge, however, what are they basing it on? Apparent race seems like a totally valid variable if behavior around people of apparently-different races is what you're trying to measure.
Despite the cleverness of your response, I don't belive the paper has anything to say about the kind of judgements people are making. They observed stuff, they didn't read minds.
The paper's title says "racial avoidance" and the paper itself claims to be about race.
EDIT: unable to reply to your reply since it seems we have reached the maximum depth of this thread, but have you considered that judging race by appearance is... racist (and unreliable)?
If your concern is that the researchers failed at picking confederates whose race, judged by appearance, would be interpreted the way the researchers expected it to be, close enough to 100% of the time to not have affected the outcome... I very much doubt they failed in that way.
If that's not your concern, then I've lost track of what your complaint is.
[EDIT] Reply to your edit: Yes, that's... the point? [EDIT AGAIN] Wait, are you saying that it's racist for the researchers to categorize pedestrians' races based on appearance? If so, I'd regard that objection as specious. That's not racist. Having-to-do-with-race isn't what racism is. It could be inaccurate (or not!) but it's not racist. In any event, it seems they didn't actually measure a difference in reaction of pedestrians based on their own race (!) so it may not be relevant anyway.
> If your concern is that the researchers failed at picking confederates whose race, judged by appearance, would be interpreted the way the researchers expected it to be, close enough to 100% of the time to not have affected the outcome… I very much doubt they failed in that way.
That’s a legitimate concern since the methodology on how “racial phenotypes” of “Black” and “non-Hispanic White” were identified for the confederate pairs is not identified. We know their assessment of the race of the subjects whose behavior is studied is measurably lower than 100%, because they actually have reliability data on that by comparison with a second rater.
I wonder if they used the traffic camera data to record the number of violent incidents to see if there was any correlation between the number of violent incidents and the amount of avoidance for each racial category.
I'm curious how they isolated the effect of race (but I can't read the paper, so, I dunno) among other signs. If they even did—the abstract doesn't make it clear to me that they're claiming race is the causal factor, but may just be observing that distance correlates with the race of the subjects (for whatever reason).
[EDIT] cpitman's link elsewhere in the thread helped: they placed the person the avoidance of whom they were measuring, so they could at least partially control for affect and clothing and such.
It's easy to picture all sorts of specific bad things that might really be represented somewhere among a broader group we disagree with on some kind of ideological grounds, or, in your case, something they might do in the future! It's not useful, though, to describe these pictures every time a topic arises that just happens to be about that group (the group in this case being academics studying something that has to do with racism).
"If I can abstractly construct an ulterior motive for the thing - despite having absolutely no evidence - that means the thing is false. Checkmate, science!"
If folks where actually skeptical of the paper, you would think they would provide some evidence to back up their thought process, based on experiences from academia, or a study of crowd movement in relation to race that shows contrary evidence to the paper itself, instead of saying they can imagine the possibility of bias.
I mean I can picture an evil cabal of racists who purposefully show up in discussions like this to purposefully sow fear, uncertainty and doubt. Is that the level of HN discourse?
The ability to imagine something isn't evidence of it's existence, nor is it healthy skepticism.
So essentially, his assertion that academia is systemically bias, can only be proven through:
>Experience in academia
>Papers, which you can only get money to work on full time through academia
Seems convenient.
Looking at the paper itself it found evidence of both a gender bias (either ingroup/outgroup or against men) and a racial bias (either ingroup/outgroup or against blacks). It notably, did not censor any facts about this gender bias, and instead published them in full. In the discussion however, it discussed these two facts with overwhelming bias, treating the observed racial bias with the utmost seriousness while not discussing the observed gender bias at all.
I think there's grounds to be sceptical of anything academia says on politically charged topics.
The point of the parent's example was to criticize the picturing of random technically possible but bad things about people you disagree with. They are specifically saying that it would be silly to make the exact claim that you are accusing them of making.
As a scientist, if you find "Black people good, white people bad" --> publication + promotion
You find "White people good, black people bad" --> you lose your job
Completely orthogonal to the morality of it, it creates something called publication bias (or the file drawer problem) in the social sciences. It's a huge issue.
>"Black people good, white people bad" --> publication + promotion
>"White people good, black people bad" --> you lose your job
It is totally believable that the bias problem you're referring to exists, but your description is provocative, dramatized, oversimplified, and carries a smug tone, and I wish it wasn't any of those things.
To back up the parents claim: The authors also submitted different test studies to different peer-review boards. The methodology was identical, and the variable was that the purported findings either went for, or against, the liberal worldview (for example, one found evidence of discrimination against minority groups, and another found evidence of "reverse discrimination" against straight white males). Despite equal methodological strengths, the studies that went against the liberal worldview were criticized and rejected, and those that went with it were not. - https://theweek.com/articles/441474/how-academias-liberal-bi...
And of course, ignore Asians at all costs. Or if the study is about overrepresentation, invent a new, "Asian + white" category, and hide disaggregated numbers.
