[links]

Oct. 20th, 2005 05:12 pm
wispfox: (Default)
[personal profile] wispfox
Fascinating link from [livejournal.com profile] dancingwolfgrrl, Implicit Association Test, where it attempts to sneakily figure out if you have implicit associations about various things.

Date: 2005-10-20 09:37 pm (UTC)
From: [identity profile] leiacat.livejournal.com
Having played with that quite a bit a while ago, I have for the most part discovered that I have a "preference" for the first combination of patterns that I'm given in each test, and it took me extra time to overwrite the first association-set. I am not inclined to take it too seriously.

Date: 2005-10-21 01:16 am (UTC)
From: [identity profile] wispfox.livejournal.com
"preference" for the first combination of patterns that I'm given in each test, and it took me extra time to overwrite the first association-set.

Yep, I started noticing that, myself.

Date: 2005-10-21 01:54 am (UTC)
From: [identity profile] metahacker.livejournal.com
They do attempt to counterbalance that by counterchanging the order of things; everyone will have a slight preference for the association they've just learned (the first one they are presented); they account for this to a certain extent.

It's an interesting, and reasonably-well documented result, that you can get cognitive dissonance (and therefore slightly slower reaction times) from grouping things you find dissimilar. Stroop tests -- like "read the *colors* of this text: " -- are an extreme example of this. Extrapolating "you find it hard to associate Rich and Black" from this is a bit more dodgy, however.

I need a brain icon.

Date: 2005-10-21 03:24 am (UTC)
From: [identity profile] leiacat.livejournal.com
I am certainly not arguing that the associations exist; I am, however, wholly unconvinced that this testset does an adequate job of compensating for other factors. What little I've seen of it, the results seem fairly bogus to me.

Date: 2005-10-21 01:18 pm (UTC)
From: [identity profile] dancingwolfgrrl.livejournal.com
Yeah, they use some algorithm to compensate to some extent. Interesting anyway, even if not a totally valid measure :)

Date: 2005-10-21 03:27 pm (UTC)
From: [identity profile] wispfox.livejournal.com
Brains!

(ok, not quite what you meant. ;)

Date: 2005-10-20 10:41 pm (UTC)
From: [identity profile] dragontdc.livejournal.com
I also found it more a test of ability to learn new reaction patterns than any sort of real preference. The learning curve weights the test heavily in favor of the first subject presented because when testing for the second, one must unlearn the first reaction and learn the new one.

The test is worthless.

Date: 2005-10-21 01:17 am (UTC)
From: [identity profile] wispfox.livejournal.com
Yep, I noticed that, in my brief explorations.

So, the results are likely worthless, but it's a fascinating example of what areas people think there is likely to be hidden bias. ;)

November 2024

S M T W T F S
     12
3 456789
10111213141516
17181920212223
24252627282930

Most Popular Tags

Active Entries

Style Credit

Expand Cut Tags

No cut tags
Page generated Jul. 2nd, 2025 12:52 am
Powered by Dreamwidth Studios