Biases
Posted on January 24, 2005 Posted by John Scalzi 25 Comments
The Washington Post Magazine did a story this week on the Implicit Association Test, which purports to show whether people have an implicit bias for one group over another; for example, for white people over black people, or for fat over thin people. One of the things that it shows, or so the story reports, is that people have rather more biases than they may be consciously aware of, or that they would like to admit — in the opening paragraphs, a gay man and a lesbian take the test and discover their implicit biases are toward straight people (the two, who had agreed to have their names published in the story, withdrew their names for attribution after their results came in).
I was curious to see whether the test would register a bias in me, so I went to the Project Implicit website and took the test for race — specifically the one for a preference for black Americans over white Americans, or vice versa. I had my own suspicions on how the test would turn out, but you never know until you take it. The test description states that the test “indicates that most Americans have an automatic preference for white over black” — and according to the magazine article, that preference includes nearly half of black Americans. Am I any different? Apparently not: “Your data suggest a moderate automatic preference for White American relative to Black American,” the test told me once I was done.
I’m not terribly surprised. I am white and raised predominately among white people. The high school I went had a high percentage of minorities but they were predominately Asian and Indian/Middle Eastern with very few black students: There were none in my graduating class, for example. My college was also racially mixed but again fairly few black students. Work life? Same set-up. And now I live in a small Ohio town with almost no minorities of any sort, and I write science fiction. I was at a science fiction convention this weekend, and out of 900 or so people, you could have counted the number of black participants and not run out of fingers.
Do I feel bad I have this bias? Well, I wish I didn’t, of course. But this bias is not news to me; I’m self-reflective enough to know where many of my biases lie. I would feel bad if I let this bias go unchallenged in myself, so I try not to do that. If one looks at the actions of my life, it’s fairly clear I don’t let this particular bias rule how I live, where I stand politically or how I make my friends. A stupid man would take a bias as an excuse for behavior; I, hopefully smarter, see bias as something to question.
I also take (small and possibly not appropriate) comfort in knowing whatever automatic white/black bias I have is subsumed by a much more active automatic bias I have, which is arrayed along economic/educational lines rather than racial ones. I can pretty much guarantee you that if I were to take an implicit association test which featured white-collar black people working in an office and white folk in John Deere caps coming off a hunting trip, I’d be skewing toward the black Americans at least moderately. Educated, reasonably affluent people are “my” people, because I’ve always lived among the educated and reasonably affluent (even if as a child I was poor myself), and my attitudes are molded therein.
This is the “Target vs. Wal-Mart” class bias; basically, people whose biases slice by way economics and education are okay popping into Target for cheap products from China, but would be mortified if anyone ever saw them walking out of a Wal-Mart with the exact same products in tow. One of my favorite stories is the time I was in a small Illinois town with two good friends with whom I went to high school, one a lawyer and the other in the film industry. My friends needed to get a DVD player and the only place to get one in this little town was a Wal-Mart. Pretty much all Wal-Marts have their departments in the same place, so I steered my friends to the electronics section. They were appalled at the fact I knew the Wal-Mart layout at all. They, of course, had never been in one.
(Later, all three of us went to the Subway in the strip mall and were arguing about the merits of the then-current Adam Sandler film Punch-Drunk Love; my friends loved it and I didn’t hate it, but then the guy behind the counter said that Adam Sandler’s other films were better because they were funnier, and everyone in the store who was listening in was nodding their head in agreement. Same “Target vs. Wal-Mart” attitude, different exhibition of it.)
This “Target vs. Wal-Mart” bias is, I should note, is no more fair. And it’s with no small irony that I’ve found myself living in a small rural town where the vast majority of residents have no more than a high school education and work as truckers, farmers or in other intensely blue-collar fields. Aside from the color of my skin, there is very little I have in common with most people in my little town. And yet, for all my automatic biases against the lower-income and lesser-educated, I will tell you that I have truly excellent neighbors, who go out of their way to help us when we need help, and for whom we do the same. I’m glad to live where I do.
Is this the end to my biases? Goodness, no. I’ve got a bunch, and aside from the two mentioned I won’t bore you with them. Basically, I know what my biases are. I also know that my biases are wrong. My biases are what they are; I work to change them and keep them from making me approach individual people unfairly.
One of the things in the Washington Post article that I think is interesting is that the people who developed the implicit association test don’t think that one’s automatic biases are destiny, and that we can make a conscious decision to work against our own biases. Naturally, I agree, and I find it encouraging that they believe so. It’s a reminder that psychologically speaking, human beings are more than a mere bundle of their basest fears and desires. Or can be, in any event.
Amazing how fast this stuff perpetuates through the blogosphere — I blogged on the same thing this morning. I think I like your take better though.
I like to think that I don’t treat black Americans any differently than I treat white Americans, but given our current society, if that’s true, it’s because I spend a lot of conscious effort making it so.
However, I am an classist. I definitely come down on the Target side of the Target/Walmart divide, though part of that is political – Walmart has a policy of actively encouraging manufacturers to outsource, and I disagree with that. (That’s not entirely true – Walmart requires that all of their suppliers drop their prices by 5% every year. The end result is the same. They also price bulk items more expensive per unit, instead of cheaper per unit, than smaller sizes.) However, I am firmly outside of my comfort zone when I’m around those of a different socio-economic group than I am. (This does work both ways – I’m uncomfortable around both the wealthy and the poor.)
But all articles and studies like this tell us is that we’re most comfortable around those like us. It’s neither ground-breaker, nor something to be ashamed of. I am most comfortable around white males between 25 and 35 years old who are used to an upper-middle-class lifestyle. Because that’s who I am, and that’s who most of my friends are. That doesn’t mean that I dislike those outside of that group, or that I won’t choose to associate with others. But I’m definitely outside of my comfort zone.
For the record, the white/black IAT listed me as “little or no automatic preference.” I know that’s not true. But I am good at pushing buttons fast.
K
Sorry for the double-post.
One last thought. Even if you have an “automatic preference” (as the IAT puts it) for one group over another, that doesn’t mean that you belong to that group, or that you would choose to associate with them, it merely means that you are more comfortable there. Take the gay/straight example from the first paragraph above. Did the gay man and lesbian show hidden biases because they secretly (wish they weren’t gay, like straight people, are ashamed of themselves, whatever), or because more than 90% of people and 99% of media representations are of straight people? Same goes for race. It’s just what you’re used to.
K
My God. The “Target vs. Walmart” bias sounds like the most idiotic thing I’ve ever heard. I’m not saying it doesn’t happen, but it proves to me America is doomed to implode from its own self-absorption.
(*Note: Use of variations on the word “absorb” does not constitute a fondness for Spongebob Squarepants, ‘cuz we all know what that would mean.)
I wouldn’t put too much stock in these tests, they seem completely nonsensical to me. Let’s leave race aside for a minute since that can turn into a hot topic, I took instead the young vs old people test. They claim the majority tends towards a slight bias in favor of the young, maybe that’s true, maybe not, either way unimportant. I know for a FACT that I have a bias, it’s not unconscious, specifically I have a bias against people who do idiotic things, and since it’s been my experience (mainly in the UK, not so much in the US), that a high percentage of people act with extreme idiocy (binge drinking every day after work is a good example) well into their 30s, I have acquired a more positive automatic response to people who appear to be over 40. Now getting back to the test, first of all the test does NOT define young and old, since these are subjective terms that’s a mistake right there, from the pictures involved it seems they mean something like young = 25 and below and old = 50 and above, apparently everyone in between doesn’t exist… Whatever, let’s ignore THAT lack of clarity as well, the main problem with the test is that it simply isn’t valid, your error rate and speed aren’t governed by any subconscious effects, they’re governed by keyboard confusion and pattern setting, the short of it is that if they make you hit 1 key often enough in a row, you’ll simply fail to hit the other when the time comes, you’re being asked to work fast and a reflex response has set in, once you’ve then gotten annoyed with yourself over the mistake you’ll slow down to control your fingers better and pause to read the words longer, this will happen in ALL the tests to some extent, and your overall rate will be governed by how often the keys switch around, not by what they signify. Proof? I’m used to twitch typing and am less prone to pattern setting and confusion, I therefore performed nearly exactly the same on all tests and resulted in an inconclusive result, no bias at all, which is a direct contradiction to what I know to be the truth.
My implicit biases
It turns out, according to the implicit bias test, that I have a “moderate association between male and career.” So, what does this mean? I don’t know. My wife has always had a more “serious” career than me. She makes more money than I do. I do mo…
John —
Thanks for bringing the Washington Post article to my attention. I am a social psychologist and I teach about this stuff every year to undergraduates whose grasp on their own biases is very tenuous, and their grasp on this test even less so. It was refreshing to read a well-articulated post on the subject.
The idea that biases are may not be conscious is not one that sits well with many people. As a scientist myself, I have been skeptical of this research for a long time. I still am skeptical, because as yet I don’t think we understand the relationship between any sort of unconscious bias and our actual conscious thoughts and behaviors, and that’s where a lot of the action is, isn’t it? It’s certainly where many of the problems lie.
But as researchers have investigated the idea of nonconscious biases using many different approaches, I have come to believe that there is *something* there. What exactly that something is, I still do not know.
Hey, that’s often the fun of research!
Kevin, the bias is ‘nothing to be ashamed of’ when you’re talking about something with no real consequences to others, such as hanging out with your buddies on a Saturday night. A bias for one’s own kind is very much a problem when it translates into, say, the workplace. My spouse, who was in IT for years, noticed this over and over again. There was never any overt bias–if you’d said “Hey, let’s not hire any women or anyone who isn’t white,” his co-workers would have been genuinely appalled. But when it comes down to the “4 a.m. test” (who do you want to be slaving away over work with at 4 a.m.?), again and again, they’d pick people just like them….25-35 year old middle-class white guys. And they professed bafflement that the workplace was so homogenous: it’s not like they were bigots, after all!
It’s also a problem in social organizations, as John’s example of the SF con demonstrates.
I’ve recently found that when *forced* to deal with people outside our comfort zones, we have a very human tendency to try to find some commonality–and we can find that so much more often than we think.
I think self knowledge is always a good thing, but that doesn’t mean I have to agree with Scientology that my body is covered with Thetan space cooties.
Let’s play evil scientist for a minute. Assuming that people have a bias towards that which is most familiar, and the media has a big role in defining what is ‘familiar,’ if you controlled the media what would you want to make ‘familiar’ for the good of society?
“Let’s play evil scientist for a minute. Assuming that people have a bias towards that which is most familiar, and the media has a big role in defining what is ‘familiar,’ if you controlled the media what would you want to make ‘familiar’ for the good of society?”
Thetan space cooties.
mythago, you’re right, unexamined biases are not a good thing. But I still maintain that you shouldn’t be ashamed of it. You should be aware, and you should counteract it, but shame only leads to denial, and that helps nobody.
Jim, about the Target/Walmart divide: I can’t speak for every place, but I can say that in my area, Targets are clean and brightly lit, while Walmarts are dirty, darker, and have jumbled shelves. Prices are possibly lower at Walmart, but I feel dirty while I’m there.
Tripp, I would make everybody “familiar” with whatever I was selling. Which brings up a good point – I wonder if people have an innate preference to, say, McDonalds over Burger King, or Coke over Pepsi. What would name-brands do to the test?
K
I just took the test and got a “no particular bias” result. I’m black, but most of my friends and co-workers are white (and it’s been ever thus, all the way through school), so I thought I might have some unconscious bias that way. I definitely have a class bias; I’m pretty well off, and despite a long period of relative poverty in my adolescence, I’ve always felt socialized that way. (I like Target, buy a lot of stuff for the house there. Went to a Wal-Mart for the first time over the holidays and was disgusted.) While I was taking the test, it felt like more of a button-pushing and attention exercise than a bias test, but maybe I’m really not that biased on a purely racial basis. Regarding Jesse Jackson’s lament, I’d actually feel much more comfortable seeing neatly dressed black kids behind me on a dark street than white ruffians. Race alone doesn’t do it for me–maybe because I don’t conform to many racial stereotypes myself.
I’m with Guy Matthew on this. I took three tests and I think they are poorly designed; the way the groups/labels are matched together is directly connected to how you answered the question: How do I feel about X group. The implicit test should be given randomly, with the explicit questions be asked afterwards.
I took the race test several week ago. I got little to no bias. My girlfriend took the test and shaded toward whites. I’m white and she’s black. She was mad and I was laughing. But we both love going to WalMart and Target. Uniting America through big box retailers.
Guy Matthews wrote:
“your error rate and speed … [are] governed by keyboard confusion and pattern setting,”
When I took some of the tests a few weeks ago it looked like they were designed to correct for such things.
Guy Matthews wrote:
“… inconclusive result, no bias at all”
An inconclusive result is not at all the same thing as a result of no/low bias. “Inconclusive” doesn’t say anything about your bias at all.
I see two problems with the test. I don’t see how they can correct for the effects of learning (unless learning follows the same pattern for everyone). The bigger problem I see is the gap between the test results and the *interpretation* of those results.
Kevin Q wrote:
“shame only leads to denial”
Oh, I don’t know about that. I am ashamed that I am not unbiased regarding race in my interactions with others. And I’m not in denial. Despite being abused by some sects I do think that shame and guilt do have some utility.
That being said, I’m not sure there’s much point in being ashamed.
Scalzi wrote:
“A stupid man would take a bias as an excuse for behavior; I, hopefully smarter, see bias as something to question.”
Amen to that. I don’t doubt that unconscious biases cause a great deal of harm in American and Canadian society, and it’s not clear to me the best way to deal with them.[1] But the deliberate form is still around and venomous, and what to do in that case is more obvious. Let’s not let racists/racialists claim they represent us, they do _not_. Let’s cross out or erase racist graffiti. Let’s not stay silent in the presence of racist ideology. Such reactions are so obviously right that I don’t worry that my motivation may be less than pure.
[1]… Although I do take it as a given that greater awareness of such biases is a starting point.
“You should be aware, and you should counteract it, but shame only leads to denial, and that helps nobody.”
This is silly. Shame does not “only” lead to denial. Shame also leads to repentance. Nor is the only reaction to realizing bias is wrong “shame.”
If I don’t believe that there’s anything at all wrong with behavior X, why on earth would I change it? If I instead decide that behavior X is bad, isn’t it normal for me to regret that behavior and not do it anymore?
I’ve been staying out of this for a while, in an effort to correctly frame my remarks. Now that others have commented, it’s a little easier:
1) Pattern learning
That was my first reaction as well. I found it interesting that in every example given, the bias was always against the *last* thing people were asked to do (e.g., black people & negative words). I wonder what would happen if the same person did each section of the test on different days, or if they did multiple iterations of each section & took an average, etc.
2) What is bias? Does it matter?
Well discussed above. I would just add that our reactions to images of strangers on a video screen could be very different than how we treat friends, colleagues, family members, or even strangers. The test seems closest in the visual experience to watching television or surfing the web. I’m sure there are many studies out there about how often people of different races are associated with positive/negative words in the media (rightfully or wrongfully). My point here is that some of the “bias” could be external conditioning based on the test’s medium.
3) Contrary evidence
I’ve always had this beef with academic research: evidence in favor of the hypothesis is seen as “telling,” while evidence against it is seen as “surprising” or “counter-intuitive.” So, the fact that people with “white-bias” are more likely to give a ficticious black person a longer sentence in a theoretical jury trial corroborates the bias test (is the simulated jury test valid? Don’t know…they didn’t say). But the fact that a person who has *chosen* to spend his/her life defending the rights of minorities registers with a “white-bias” means he/she is “lying to him/herself,” not that the test could be less than 100% accurate. Why aren’t these kinds of results “telling” as well?
“I can’t speak for every place, but I can say that in my area, Targets are clean and brightly lit, while Walmarts are dirty, darker, and have jumbled shelves. Prices are possibly lower at Walmart, but I feel dirty while I’m there.”
The Target/WalMart divide exists at least in part because Target wants it to be there. A few years ago I did some research on this and found that a very conscious part of their strategy is to be the anti-WalMart: Clean stores, no flourescent lighting, employees wear clean red shirts instead of blue vests, no greeters, that sort of thing. Target figured — rightly, it turns out — that there were lots of people who want similar prices to WalMart’s but would play slightly more for a less grody shopping experience.
Brian Greenberg:
“Well discussed above. I would just add that our reactions to images of strangers on a video screen could be very different than how we treat friends, colleagues, family members, or even strangers.”
Well sure. Hence the additional studies of how the results correlate with the treatment of strangers in various situations.
Brian Greenberg:
“But the fact that a person who has *chosen* to spend his/her life defending the rights of minorities registers with a “white-bias” means he/she is “lying to him/herself,” not that the test could be less than 100% accurate.”
Point. But people do lie to themselves all the time. People are rife with biases in perception and behaviour that they are quite unaware of. And I don’t see that large a leap from unconcious biases in perception to unconcious biases in behaviour.
Andrew Wade:
“When I took some of the tests a few weeks ago it looked like they were designed to correct for such things.”
On the contrary Andrew, if anything the test is designed in such a manner as to INDUCE error-related stalling. If the test is out to measure your response times relative to the word and image pattern distribution it shouldn’t tell you at all if you’d made an error, but rather take silent note of the fact and let you move on unimpeded. Instead it stops dead, requiring you to reverse your choice, a significant artificial stall in your response rate, and worse, it gives you a big glaring red X to indicate your error, the psychological impact at both conscious and subconscious levels of that kind of response is HUGE, significantly boosting your sense of self-chastisement, which in turn affects your rate further in a manner not related to bias as discussed in my previous post.
Andrew Wade:
“Well sure. Hence the additional studies of how the results correlate with the treatment of strangers in various situations.”
Actually, no. The additional studies show how the results correlate with *theoretical* treatment of strangers in the other studies.
This was precisely my point. A person who makes gay rights advocacy, for example, their livelihood has demonstrated by their actions over a long period of time where their biases are. If the test then shows them to have an “anti-gay” bias, I would interpret those results to be a false positive in the test, not a secret bias that they’ve by lying to themselves about all this time.
On the other hand, if we take the position that the test is correct and this unconscious bias *does* exist, then I would submit that the way the person lives his/her life proves pretty conclusively that the bias is of no consequence whatsoever. Following from that, we can conclude that *finding* this bias is a useless excercise…
Guy Matthews wrote:
“On the contrary Andrew, if anything the test is designed in such a manner as to INDUCE error-related stalling.”
Well sure. Which means the speed of completing the task is probably going to be related to the difficulty of the task. The effects of “pattern setting” should be similar in all trials, are more than that the noise that such effects introduce will be measurable. Now the effects of “self-chastisement” probably won’t be the same for all trials.
Brian Greenberg wrote:
“Actually, no. The additional studies show how the results correlate with *theoretical* treatment of strangers in the other studies.”
Say what? Some of the treatment is quite real. Now whether the behaviour continues when people don’t know they are being studied is another matter…
Brian Greenberg wrote:
“On the other hand, if we take the position that the test is correct and this unconscious bias *does* exist, then I would submit that the way the person lives his/her life proves pretty conclusively that the bias is of no consequence whatsoever.”
Well no. Humans are messy beings; there are many hypothetical ways the gay activist could be without unconcious bias, and no way to know what they are never mind pick the “right” one to compare against. Defining the consequences is an exercise fraught with difficulty.
With another example or two, we could probably conclude that the test is not 100% accurate in predicting behaviour. But that’s no surprise.
Even with random answers, it is possible to conclude interesting things.
Try to answer randomly to this test. http://intelligence.sergi5.com The computer will predict your answers.
The only thing the implicit project can do, from what I see, is use there knownledge base of what people declared to be, biased or not, and what they answered to the quick questions, to predict to somebody new in wich way he thinks to be biased…
John Scalzi,
I’m not sure I understand your point. Are you saying that it’s ok that you’re biased against blacks, because you think you may be even more biased against poor people?
You also claim to be more comfortable around the affluent, regardless of race, and yet by your own admission you choose to live in a less affluent area with no minorities, rather than a more affluent area with many minorities.
I’m glad to see that you would rather not let this bias go “unchallanged” in yourself, but since you live in a white bread world, I don’t see that the “challange” will be all that difficult.
Greg writes:
“Since you live in a white bread world, I don’t see that the ‘challange’ will be all that difficult.”
My world is not as white-bread as you seem to think, as my family here in Ohio has people in it with African, Hispanic and Native American ancestries. The town where I live is indeed very white, but the family I spend time with on a day-to-day basis is rather quite *not.*
As I say in the column, I’m not happy about the biases, and I work to change them and get past them. But I don’t see much value in suggesting they *don’t* exist, or not acknowledging that some biases that I have are stronger than others.