ANSWER ME NOW HARDSCI! Or, are you just a softsci? [Boarding a cross country flight as an excuse to not respond back, ever.]
-
-
Replying to @BrianNosek
Sanjay Srivastava Retweeted nick davis
Ok: zooming out a sec. Respondent just got a score, wants to know what it means. How good of an understanding do they come away with? As noted here that's a tough yardstick - but in this context it is the correct onehttps://twitter.com/ntdPhD/status/986679703787524096 …
Sanjay Srivastava added,
1 reply 0 retweets 0 likes -
Replying to @hardsci @BrianNosek
Telling a person that your (individual) score means you (individually) have a slight/ moderate/strong preference; it is "unusual" to get very different feedback on retaking; or "It is well-established that implicit preferences can predict behavior" all convey a lot of certainty
1 reply 0 retweets 0 likes -
Replying to @hardsci @BrianNosek
Re the "can predict behavior" one, what will a regular person come away with? Probably that the test I just took says something of practical importance about how I'm likely to act in relevant domains
1 reply 0 retweets 0 likes -
Replying to @hardsci @BrianNosek
Not that scientists can reject the null that r(IAT, behavior) == 0.000000, nor that we can pick one IAT-behavior relationship that is pretty substantial
1 reply 0 retweets 0 likes -
Replying to @hardsci @BrianNosek
That FAQ gives very little sense of any uncertainty in score interpretation, very little sense of the variability in how well or poorly different IATs predict the kinds of behaviors a layperson would think it would, very little sense of limitations overall
2 replies 0 retweets 3 likes -
Replying to @hardsci
How does this not convey uncertainty? It says--variation is expected (uncertainty), if somewhat different on multiple tests-average them (uncertainty), and if wildly different consider inconclusive (uncertainty). I am always interested to improve clarity, but this seems off markpic.twitter.com/YY8NT1TsnS
1 reply 0 retweets 0 likes -
Replying to @BrianNosek
I was talking about uncertainty broadly. But my reading of this graf is that it still sounds pretty certain. e.g. "No test is perfectly accurate" sounds lip-servicey to me. You (reasonably and in good faith) disagree 1/
2 replies 0 retweets 0 likes -
Replying to @hardsci @BrianNosek
The larger problem is that implicit bias tests depend on the assumption that any perceived differences between racial groups are false. Even a small real difference in personality or behaviour would be enough to alter reaction times of judgments.
1 reply 0 retweets 1 like -
Replying to @JamesPsychol @hardsci
That's not correct. Implicit bias tests do not presume that perceived differences are false. In fact, much of the literature on attitude formation with implicit bias measures assesses their sensitivity to actual contingencies observed in the environment.
2 replies 0 retweets 0 likes
Why is it called 'bias' then? My guess is that stereotype accuracy will correlate with these response bias effects. Quicker processing of typical cases seems reasonable, odd cases require extra scrutiny because the prior is lower.
-
-
Bias in psychological use doesn’t necessarily imply immoral or unwanted see eg SDT. Also, Banaji, Nosek, & Greenwald 2004 and others. Theory emphasizes ordinariness of bias and its moral assessment is often independent of the cognitive processes.
1 reply 0 retweets 1 like -
"Bias" does seem to mean that. But it does not follow because response time differences aren't shown to be a bias until it is shown that they are larger than what is reasonable. Analogous to test bias, can't prove bias by showing a group gap. https://en.wikipedia.org/wiki/Bias pic.twitter.com/jFv6lgqWAt
0 replies 1 retweet 3 likes
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.