So since all of the US election fallout I've been hearing a lot of people and purported experts espouse notions that implicit bias against women, minorities, etc. played a large role in the election result.
As someone with a B.Sci in Biology, and an interest in neurobiology and genetics, I'm curious as to the actual validity of these claims. IIRC, many of these purported researchers use tools like "Implicit Bias Tests" as evidence to support their conclusion, but I don't see how this can be considered valid. I myself have taken 2 of these tests; one for racial bias, and one for sexual orientation bias. Anecdotally, as someone who is politically independent, and a white bisexual male who has, thus far, only dated other non-black men (I'd chalk that up to chance, not bias), the test prompts concluded that I preferred blacks and straight people. How can these tests be measuring "bias" when they could simply be measuring my speed/accuracy in correcting recently memorized associations when assigning descriptors to categories? Does your handedness factor into it (the test directs you to use keys operated by each hand)? The claims that these tests actually measure implicit bias towards demographic groups seems more like political ideology than science. Could you not make people take associative bias tests for categories that are truly neutral (like shapes for example) and rationalize it with the same conclusion that I prefer squares to triangles?
The only other research I am aware is a study showing infants prefer adults of the same skin color, but even then, this strikes me as something that can't necessarily be extrapolated to adults.
bias studies are for lib-arts faggots who can't into real science so they default to 'muh statistics
Oliver Hill
Well memed user.
Well yeah, that's what I'm coming to realize. But I guess I'm wondering if there are any further neuroscience studies, maybe even brain scans, anything more empirical than simply this test? Are there any better studies out there? Because this is what I see presented most often as evidence for implicit bias by a lot of "social science academics" these days.
Jason Mitchell
The goal of those tests isn't to tell you "wow user you're a fucking racist," the goal is to see how easily you register connections between things (in this case, like/dislike and race or gender). The real implicit bias research comes from studies that then use these results and correlate them with other things. For instance, there was a study that found that higher white preferences were correlated with offering better loan deals and lower prices when selling cars to whites as compared with blacks. This is not the same thing as measuring open and conscious racism, but rather is designed to get at the underlying biases that people in a society which sends out certain messages have.
I can post a few papers on the research that comes after the test if you're interested, but I'm on my phone atm so it probably won't be for a while.
As to your second point, it turns out that race/gender/sexual preference don't have a huge influence on implicit bias. Over half of blacks display some degree of prowhite/antiblack implicit bias. One of the professors I studied with (a black man) even said he stopped listening to a lot of "black" music because of the fact that a lot of it devalues black lives.
You're right of course that it can't be 100% accurate but there are a number of studies that find similar correlations at similar strengths between IAT scores and experimental results.
Nolan Smith
>the goal is to see how easily you register connections between things How do you define "registering connections" though? Could it not be measuring reflexive, short term memory and your ability to accurately and rapidly forget and relearn new associations based on instructive prompts?
>This is not the same thing as measuring open and conscious racism, but rather is designed to get at the underlying biases that people in a society which sends out certain messages have. Well I understand that's what some academics and researchers are trying to say, but some of the same people are also pointing towards this type of research as "the underlying cause of a Trump presidency".
I'm still not seeing how these tests measure bias. Could you not argue, regarding the loans/prices study, that the conclusion of bias measured was actually caused by statistical averages regarding credit scores or loan defaults of those demographic groups? That is, is implicit bias causal, correlative, or non-causal and how do you know?
And would it not be true that, using the same test methodology used in IATs, you could show what would be argued as "implicit bias" between arbitrary or meaningless categories such as Triagles vs Squares, Squiggly lines vs Jagged lines, Kiki vs Booba (thanks Ramachandran), etc.? Would that somehow show that people have an inherent bias towards Squares over Triangles or vice versa?
I'll be eagerly waiting on those papers user.
Michael Rivera
That's the meme for physics you fag.
Juan Cooper
I believe it is fairly well known that the implicit association test in particular produces results that do not predict in any way discriminatory behavior. Meaning that the neat numbers do not actually measure anything relevant.
Christian Reyes
Fairly well known by who? Because all I hear in the news, pop culture, whatever, is the opposite.
Jordan Miller
Pop culture has just about stopped bothering with Freudian psychoanalysis, seventy years after being soundly rejected by actual scientists. Your point?
Adrian Powell
Phone user again: you're correct that showing relationships such as triangles vs. squares would not be particularly meaningful, but that is also a different category - whether or not you prefer or more easily associate squares or triangles with good vs bad is probably not going to be something that bleeds over into the real world as something like associating the way people look with positive or negative traits.
The point about whether implicit biases are correlative, causal, or confounded by some other factor is interesting and one that I have considered, but for the loan study at least (from my memory, still on the bus headed home) the applicants were all given identical credit scores and histories, so whether or not demographics play a part in the usual course of business, in this case applicants drew from a set pool of credit scores, jobs, etc. so I'm not sure if it would make a big difference.
I'll be honest I don't have enough knowledge to answer all of the questions I have, let alone the ones you have, but assuming the city of LA's public transit system is cooperative I'll be back home and searching my hard drive for the papers and refs I have on this subject in like 30 minutes. In my opinion, the underlying idea (people have unconscious beliefs or have feelings which may influence their conscious behavior) is sound and the research methods that I've seen are at least adequate, but of course I'm not a scientist by training and you may disagree.
Levi Cox
Irelevant
Jaxon Thomas
>meme evolution
Xavier Young
okay so here's a first pass at finding some relevant articles, my external where i have all of this is a mess so i'm sure i've missed some.
So i was wrong, it turns out this one is more a statistical analysis as opposed to an IAT style test, but the results are nonetheless related to the subject I think.
GNAT (related to IAT, both measure implicit attitudes according to their creators) is used, along with other methods, to explore the correlation between opposition to obamacare and racial biases.
gives some insight into a real world application - 133 judges took the IAT and then were given some hypothetical cases to decide. even judges have implicit biases according to these tests, and those were reflected in the initial decisions made in these hypothetical cases.
this isn't a particularly strong paper since it falls into some of the issues you talked about (is it really implicit bias or is it a third confounding factor, or something else?), but it does seem to indicate that there is a correlative relationship between implicit attitudes and other attitudes or actions.
Ryan Lewis
>you're correct that showing relationships such as triangles vs. squares would not be particularly meaningful, but that is also a different category I think my point is that I'm trying to demonstrate a potential fault in the methodology.
If you were to show that there are comparable or even identical frequencies of mistakes and response times with, say, racial groups as you would when answering questions regarding something as elemental as triangles vs squares, would it not be reasonable to conclude that those results being concluded as measuring bias between racial groups may actually be measuring something else more fundamental than bias in how our brains and nervous systems are wired?
That's why I bring up possible confounding variables such as handedness, reflex time and a person's ability to forget previous associations and then quickly operate within new ones. The tests seem (to me) to measure how quickly and accurately someone is able to switch from an initially memorized associative framework and then quickly switch to its inverse while that initial framework is still fresh in their minds. Yet it seems those rather basic results are being extrapolated to draw a very complex conclusion in which perhaps an uncountable number of confounding variables may come into play.
>in this case applicants drew from a set pool of credit scores, jobs, etc. so I'm not sure if it would make a big difference. I suppose you could argue that even if a black person did have an identical credit score to a white person, there is still that greater statistical possibility (on average) that the black person would default, perhaps due to limited reliability of credit scores as a measurement, or due to some variable not being reflected or captured in their credit scores. I'm sort of spit-balling with possibilities here though.
I look forward to some literature.
Levi Davis
Implicit race attitudes predict trustworthiness judgments and economic trust decisions, Damian A. Stanley et. al. 2011. pnas.org/content/108/19/7710.long
demonstrates the correlation between IAT and economic trustworthiness attitudes.
and now that i've spilled over into the second post i really need to go make some lunch or something. I'll be checking the thread if you want to keep discussing this topic, and I hope other anons will join in. Most of these papers are from a legal perspective because, well, I read them in law school, but the experimental methods are all drawn from psychology or economics.
inb4 >psych >econ >science, sure that's possibly a criticism of those fields but i don't really think that is a criticism of their experimental designs. again, i'm not an expert and i know it so i'd be happy if people who knew more than me could comment, but I have maybe 20-30 more papers on this subject that I haven't yet skimmed through so if you want some more cites i can just dump them, or maybe upload all the pdfs I have somewhere if that is more convenient.
Noah Barnes
I agree with you about the confounding variables, and I do think that there is a possiblity that the test is a better measurement of cognitive reordering than racial attitudes, but that is why i think the experimental data that shows not only that these results hold true across a variety of political/economic/racial/gender/etc categories, AND more importantly that there are actual real world instances where the measured association lines up with action is so important.
again though, i'm the first to admit that my specialty is not neuroscience or psychology.
Nathan Brooks
*and as an addendum, would it not be relevant to racial attitudes that it is easier for people to re-associate good with white and bad with black, which across all IATs seems to be the broad trend? even if the IAT is not actually measuring implicit racial attitudes, that trend still exists - maybe it is faulty design, maybe it is another factor, but some there must be some reason for it.
Kevin Cruz
Thanks for the dump. I'm a bit busy myself so it will take me some time to get through a lot of it. I'll keep checking back to see that the thread still lives.
Also, a while ago I think I saw that study regarding Judges. Maybe I'm confusing it with another study, but there was one that showed that defendants who were deemed by the jury or judge to be more ugly were also more likely to receive harsher punishments and/or convictions than those who were considered more attractive. Using an evolutionary perspective, this makes sense, considering physical appearance (ugliness or attractiveness) is in many ways, a means in which animals detect genetic fitness and compatibility among members of their own species. It could be argued that it's a way for a species to unconsciously seek to preserve or propagate genetic fitness within its own population. I'm not sure how this could apply to race or ethnicity though.
>AND more importantly that there are actual real world instances where the measured association lines up with action is so important. That's what I'm concerned about as well though. I think showing potential flaws in the methodology elucidates on the validity of the conclusions being espoused. It would at least show that the relationship is correlative, not causative.
John Ross
What if people were instructed to associate white people and black people with different sets of words that were not qualitative in the way "good" or "bad" are? Or if completely non-qualitative words were used? If the results came back similar or identical across people would that not also point to something more fundamental with regards to brain function than a complex emotional and social process of bias?
Xavier Moore
>Attracitveness
i've seen that one as well, and while i agree that there might be an evolutionary basis for it as to attractiveness, its difficult for me to think of a similar reason for race. additionally, even if it is a "natural" response to favor members of your own race, it is my opinion that the laws of a society shouldn't be organized so as to reinforce that. the reality of american society is that it is multi-racial, and for our society to function well the laws (and the people, but that's a whole different barrel of monkeys) should acknowledge that and try to make sure that all members of our society can at least rely on the legal mechanisms to not be biased.
at any rate, looking forward to your responses.
Jose Cooper
Well I certainly agree with that sentiment on principle. That's how our society and judicial system should operate, and I'd like to think that it does most of the time, but still, we're dealing with very ancient and primitive structures of the brain that, in many ways operate unconsciously on our decision making processes (if you could even argue that humans actually have free will, but that's a whole other can of worms). Education is probably the best way to suppress those sorts of emotionally driven "reptilian" parts of the brain concerned with social hierarchies and dominance among social animals.
Regardless, concerning the criminal justice system, it's probably the best system we have at the moment, since wide-scale brain rewiring and manipulation would be a monstrous and ethically fraught proposition. Maybe we would eventually be able to construct some completely impartial AI that would be capable of accurately discerning guilt without any inherent biases, but that solution, too, comes off to many as rather inhumane. With humans at least there is the capacity for mercy, whereas with something emotionless and impartial there is none.
As a side note, I think recently there was actually an AI that was able to predict criminal recidivism with greater accuracy than judges or juries (can't find the article at the moment).
Lucas Jenkins
I hope Trumpster will cut off all the funds for pseudoscience of this kind
Thomas Price
Unfortunately those cuts will probably include funding to climate research.
Eli Adams
Like it's a bad thing
Leo Wright
what makes it pseudoscience user? just because something isn't the result of an equation or a brain scan doesn't mean implicit association research isn't using scientific methodologies to formulate a predictive and descriptive account of at least part a of natural phenomenon, in this case human interactions.
or are you a memester who doesn't believe in studying human interactions?
Kevin Collins
I think if research bases on experiment which results can't be reproduced then it's not a real science
Jaxson Wilson
The argument has already been laid out in this thread how these IATs are not actually measuring bias or preference, so the conclusions being reached are erroneous due to faulty research methodologies and not fully understanding exactly what variables are being measured or you think you are measuring. Aside from that, how are these predictive in any way? They're descriptive, sure, but those descriptions rely on assumptions and preconceived notions that have yet to be validated. Using tests that probably don't measure what you think they're measuring as a means of reinforcing your assumptions isn't science.
