Obtaining a quantitative IAT score

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
Status
Not open for further replies.

ExploringOptions

Full Member
Joined
Jan 24, 2019
Messages
115
Reaction score
153
Hoping someone on here can help! I am working on my thesis proposal and it’s an extension of my current research. I want to include implicit bias testing that provides a quantitative number for comparison. Ideally, it will be a measure I can administer via computer/online. I know about Harvard’s IAT but it doesn’t give you a “number” result and research I have looked at does. Before I try to randomly email authors in the hope one will respond, I thought I’d check here. I want to be able to generate a score that isn’t seen by the participant. TIA for any help!

Members don't see this ad.
 
  • Like
Reactions: 1 user
Thank you! That’s helpful, but I guess I am also trying to figure out HOW you get the results yourself? I may not be phrasing things clearly, my apologies. Like, how do you get the response times etc? Do you pay for the testing and get the results sent to you some how? I’m really green as I’ve never used it at all for research, and neither has my advisor. Alternatively, is there a different measure you would recommend for testing implicit racial bias that would be better?
 
Members don't see this ad :)
Damn, just email the PI....
That wasn’t helpful, but thanks anyway. Like I said, I am also interested in other methods people have used. If you don’t want to assist, that’s fine, but no need to be snarky. Thanks.
 
  • Like
Reactions: 1 user
Thank you! That’s helpful, but I guess I am also trying to figure out HOW you get the results yourself? I may not be phrasing things clearly, my apologies. Like, how do you get the response times etc? Do you pay for the testing and get the results sent to you some how? I’m really green as I’ve never used it at all for research, and neither has my advisor. Alternatively, is there a different measure you would recommend for testing implicit racial bias that would be better?

Ahhh. Everyone I know has recreated it in either Matlab, E-prime or something along those lines but all were making slight paradigm adaptions. If you want to use that exact version and do so using their website, I agree, just email the PI and ask. They may or may not release them if you want to administer off their portal. Your IRB may or may not allow this or have certain requirements for it.

I generally expect to have to program tasks myself when using them and am happy when that isnt the case.
 
  • Like
Reactions: 1 users
Ahhh. Everyone I know has recreated it in either Matlab, E-prime or something along those lines but all were making slight paradigm adaptions. If you want to use that exact version and do so using their website, I agree, just email the PI and ask. They may or may not release them if you want to administer off their portal. Your IRB may or may not allow this or have certain requirements for it.

I generally expect to have to program tasks myself when using them and am happy when that isnt the case.
Thank you! I am not a whiz with Matlab but spouse is, so that may work! I appreciate your reply!
 
I used the website SocialSci for mine - it had built-in IAT, so I just received the results (and didn't have to figure out how to measure microsecond differences). It was free and worked well for me (IAT was one aspect of my study and I was able to set up the entire thing on their site). They do have a large pool of participants that you can pay to use, but I just used my own. My participants were given the opportunity to sign up for their large pool at the end of the survey - which is why they offered the system for free. I just tried to access them, however, and nothing came up, so I don't know if they're still around. Even if they're not, there might be something similar out there for you.
 
  • Like
Reactions: 1 user
I haven't checked in on the IAT in a while. Did they ever fix the numerous methodological issues to the extent to actually make it a useful measure?
I too am interested if the question about construct validity has been better addressed.
 
I haven't checked in on the IAT in a while. Did they ever fix the numerous methodological issues to the extent to actually make it a useful measure?
I dont do IAT so my exposure/knowledge is pretty cursory, but the methods groups I'm part of are still not fans. I remember it popping up specifically about race after a meta analysis showed no relationship to actual behavior.
 
  • Like
Reactions: 1 user
I’d argue it’s not a consensus because there are a lot of good researchers who use it, but I am not here to argue whether you like the measure or not.
 
Using a measure or a test is not indicative of it being good psychometrically. I can name dozens of tests in common usage that are garbage psychometrically, but that incompetent people still use. It's not about liking the measure or not, it's about whether or not it has a sound methodological background. Use if it you want, I don't care. It can join the rest of the garbage data out there.
 
I’d argue it’s not a consensus because there are a lot of good researchers who use it, but I am not here to argue whether you like the measure or not.
I think I need to start a dialogue with my social psychology colleagues about it. From my perspective, it has gotten a bad reputation outside of social psychology.
 
I haven't checked in on the IAT in a while. Did they ever fix the numerous methodological issues to the extent to actually make it a useful measure?

Yes and no. Early scoring was wonky, I think that has reasonably been settled. Still debate over other specific task components.

Validity somewhat depends what you are talking about (as its used for way more than just racial bias now). Generally weak. Then again, I'm not convinced our typical reference measures are necessarily very good.

I like lab stuff and am about 50% basic scientist at this point, so I think it can be worth getting but in a very "basic science maybe we'll get there one day" kind of way. Its a long ways from being diagnostic or practically useful in any way. Then again, the same could be said for about 98% of science. Including virtually the entire fields of cognitive/affective neuroscience (note: all my current NIH funding is largely within those domains so its not personal).
 
  • Like
Reactions: 2 users
Status
Not open for further replies.
Top