Shooter Task study/game

[Ahriman238]

Try your hand at this, sociological experiment to see (if living in the US) how much your background effects implicit racial biases.

For 5-10 minutes (make sure you have the time and won't be interrupted) you're shown a series of photos before a person is inserted, you have about half a second to choose to shoot that person (j key) or put up your gun (f.) Some of these people will be armed with handguns, some with cell phones, wallets or soda cans. Many are randomly crouching or kneeling. All are male, black, white, some could be hispanic or mixed-race.

Shoot: 10 points for nailing an armed suspect, -20 for blasting a civilian
Don't Shoot: 5 points for sparing the innocent, -40 and "you're dead" message for not shooting a bad guy.
Ran Out the Clock: -5 if the person is harmless, -50 and "you're dead" if armed

I wound up in the negatives early and stayed there the whole game, mostly for hesitating. I let a bunch of people with guns go, and shot a dozen or so innocent bystanders. Yeah, if this task is any judge I'd make a terrible cop. I don't think there was any particular racial bias to my spread, after the first minute you kind of tune out everything but "Person! Okay, what's in his hand?"

Kind of long-running and repetitive, but that's because it really is an experiment not a game. Afterwards they'll have a 20 second questionaire; gender, age, race, and zip code. You don't get detailed results back.

I ended with -360. I think my high water mark after turning around initial setbacks was clawing my way up to -115. How will you do?

[Mr Bean]

I quit after six minutes as fatigued became a factor. My "high score" was 415 with zero civilians shot and two "to lates". Then I rapidly got knocked down to twice with two your deads, then a third before getting back up to 355 and just quitting because as I said six minutes is forever when your sitting there waiting for it to end.

My totals were being shot three times (All within thirty seconds) two "your lates" on civs for -5 each and no civ deaths. Maybe it's four years in the navy or twenty years in FPS but I'm very good at target identification.

[Brother-Captain Gaius]

This study is stupid. It appears to test reaction time and hand-eye coordination, not racial bias.

Whatever data they do get is going to be further skewed by gamers versus non-gamers.

[salm]

I ended with 280 after a couple of minutes but couldn´t finish it because it got absurdly boring.

[Zixinus]

Is the test as likely to show a black person armed as a white person armed? I'm asking because I'm curious.

Interesting test and I see how it tries to aim at not allowing time for conscious decision making but quick identification. To me it shows just how difficult spotting someone armed is.

[SCRawl]

I actually had a score over 500. I did horribly in the practice run, but I guess that that prepared me well for the real test. I was killed a couple of times, but I never shot anyone who was unarmed.

Of course, being Canadian my results are probably of no value to them.

[Ahriman238]

Is the test as likely to show a black person armed as a white person armed? I'm asking because I'm curious.

Interesting test and I see how it tries to aim at not allowing time for conscious decision making but quick identification. To me it shows just how difficult spotting someone armed is.

Yeah, over a ten minute period it got to be a bit much to keep track of, but there were a good spread of armed and unarmed men of both (all?) races.

[GrandMasterTerwynn]

This study is stupid. It appears to test reaction time and hand-eye coordination, not racial bias.

Whatever data they do get is going to be further skewed by gamers versus non-gamers.

I believe it absolutely does test for racial bias, and it's doing so at the subconscious "gut-instinct" level. For example, if you were raised in an environment steeped in racial bias, even if you don't exhibit any outward biases ... your subconscious conditioning may prompt you to make snap judgements (that you wouldn't make if you had time to take a careful look at a given picture) that could reveal what biases were ingrained into you by your environment.

[RogueIce]

I messed up five times. Twice I shot the wrong person ( ) and three times I ended up shot myself ( ). Both of the wrong people I shot were white; the three people who killed me were black.

Damn you internalized anti-white bias!

That said, I don't know if it really does or doesn't test for racial bias, except in assuming that it does because they say it does. But then my biggest problem was the stupid hand-eye coordination: with the time pressure and "snap judgement" I kept defaulting to clicking down with my right finger, likely due to my right hand dominance. Even when I was able to mentally ascertain that it was a wallet or cell phone - I even started verbalizing between "Safe" and "Shoot" - I still had an annoying tendency to click with the improper finger. And I would tend to overcorrect when I thought about it, picking F when I wanted J because I was thinking "damn it stop clicking J click F".

It wasn't until I said screw it and decided to eat the -5 from a No Decision on "good guys" that I managed any consistency, and thus the result of shooting two innocents and getting shot three times - though in my defense, I did click J the three times I "died" but apparently not fast enough to register. And when I did off those poor innocent fellows, it came after a string of "Good Shots" so I may have just fallen into a temporary pattern there.

Of course I have no idea if my experience was representative but I do question the methodology. I think if they modified it so that the only option was "Shoot" and inaction on your part still counted positively (you're not going to "squeeze a trigger" - represented here by pushing a button - on a non-threatening person) it might work better.

[Ziggy Stardust]

This study is stupid. It appears to test reaction time and hand-eye coordination, not racial bias.

On the surface, yes. But this is actually a very well known and well established study design. I have personally designed and conducted experiments that worked in a very similar fashion. It is a variant of studies designed to tease out "priming" effects. The idea of psychological priming is that exposure to one stimulus influences a response to another stimulus. There is a very long and relatively accurate Wikipedia article on the subject that I recommend reading if you are interested in the topic. But in essence, you "distract" subjects with one task (in this case the act of shooting and seeing what is in the hand of the person in the image) in order to investigate the implicit memory effect of racial discrimination. That is, if you believe very strongly that colored people are all criminals, this cultural mindset is "priming" the way you react in the target discrimination task. That is, your false alarm and true positive rates will be dramatically increased over the "non-white" stimulus set compared to the "white" one.

Whether or not this effect is strong enough to detect in this study, I have no idea. It is entirely possible they find nothing. Priming effects are very difficult to isolate, and often swamped by a multitude of other factors involved. Psychology is very complicated.

Whatever data they do get is going to be further skewed by gamers versus non-gamers.

Most likely, the reason they are asking for age and demographic information in the post-test survey is for the purposes of stratifying the data set into cohorts based on age groups (e.g. "25-35" or whatever) and along other demographic variables (race, gender, etc.). I would suspect that they believe any video game related confounding will be broken down along with age. For example, the empirical distribution of video game playing is strongly colinear with age (source. If they stratify this sample similarly, they can control for video game playing as a latent variable and account for it in their variance estimates for the model.

Further, this study is what is sometimes called a "Mechanical Turk." Mass distributed online surveys like this can pull in massive amounts of data. Typical psychology studies enroll at university campuses and such, and tend to only recruit a couple of hundred subjects, in an extremely biased sample. Online studies can get thousands upon thousands of participants across a much broadly spectrum of the population (although, I would argue, still biased in a different way). Large sample sizes go a long way towards normalizing your data sample and smoothing things out.

Again, I am not saying this study will ever produce viable or interesting results. I am simply pointing out that it is a sound study design.

Is the test as likely to show a black person armed as a white person armed? I'm asking because I'm curious.

Almost certainly. It would be a pretty poorly designed study if it weren't balanced. Of course, it doesn't necessarily need to be 50/50. They could split it evenly among all presented races (so equally likely to see white, black, hispanic, asian, etc.), or they may sample in such a way as to closely approximate current US demographic proportions, but that is unlikely and may be difficult to interpret.

[Purple]

Can I take the practice run without contributing any data? I am asking because I am not american and thus should not participate properly.

[Ahriman238]

Can I take the practice run without contributing any data? I am asking because I am not american and thus should not participate properly.

Sure, there's a five-minute dry run before the task, just to get you used to it.

Results.

-snip background on racial shootings-

So we spoke to Joshua Correll, a University of Chicago psychology professor who has for years studied bias against black men. His line of study specifically looks at bias when people are making decisions whether to pull a trigger on a bad guy.

Correll says the participants are universally more likely to fire at black men — whether the shooter is young, old, male, female or even black.

"Our working hypothesis is that in American culture, black is often associated with danger and crime and physical aggression," Correll says. "And it's really that association that leads us to see them as a greater threat and to pull the trigger. And everybody knows about those stereotypes. You don't need to be black or white or Latino or Asian to learn about those stereotypes. We all see them all the time, in news coverage, in movies, in a lot of music."

Correll, who is white, has been testing this hypothesis since 2002. Mostly, he's looked at bias when police officers decided to shoot, but he has also tested other members of the community. (Obviously none of these tests are directly connected to the Florida shooting.)

Here's how his test works: He gives the participant two buttons: Shoot or don't shoot. Then he presents pictures of black men and white men. Some are holding guns, others are holding harmless things like wallets and cellphones. The point is to shoot the guys with guns. What Correll has found is that no matter the race or the age of the shooter, he or she is more likely to fire at an unarmed black man. They're also less likely to shoot an armed white man.

"Everybody was faster to shoot a black target than a white target, and the magnitude of that bias was equivalent" regardless of race, said Correll.

-snip digression-

Correll points out another important point of his research: Age hasn't mattered in his research. So, participants who grew up in a more overtly racist time were just as likely to shoot the black target as younger participants who have grown up in a country that just elected its first black president.

It's a bleak landscape, Correll admits, and he hasn't found any evidence that cultural sensitivity training curbs people's biases.

"To the extent that we're all exposed to [these biases], it becomes a tricky situation," says Correll.

[Ziggy Stardust]

The link the in the OP is to a study out of California State University, not University of Chicago. The P.I. is Debbie Ma, who was previously one of Joshua Correll's PhD students. The OP study is still active, and as such they would not be analyzing or reporting any results yet. So that article doesn't actually reflect anything about the current study.

Now, obviously, the article is talking about previous research, and this is some sort of follow-up or verification study. The problem is that the results reported in that article don't seem to match up with what I can find searching Wiley Online Library for Joshua Correll's publications. His most recent paper that I can read, for example, talks about how when the study was conducted on police officers, no evidence of bias was discovered. And that the only strong evidence of bias was in a study conducted on college students. That's ... a bit more tempered than the grandiose proclamations of that article.