I’m not going to shout down people who make observations about group behavior just because their observations haven’t been tested in a double blind trial yet. the data precludes us from making certain generalizations. it doesn’t stop the tentative creation of new ones.
if I made a generalization about people with fingers of certain length it wouldn’t generate nearly this much ire. we shouldn’t treat race any differently just because people made stupid generalizations in the past.
We don’t have enough data to make the case for OR against any racism (biological differences → behavioral differences)
we shouldn’t treat race any differently just because people made stupid generalizations in the past.
If a topic tends to historically collect relatively more stupid generalizations than other topics, isn’t it reasonable to keep a stronger default prior against such generalizations that aren’t backed by data?
We don’t have enough data to make the case for OR against any racism (biological differences → behavioral differences)
That’s an important point (that others’ generalizations on race are suspect by default). However, I’m perfectly happy to consider appearance (including race) as a conditioning variable in my own thinking. I guess it might be smart to not admit this, but I think it’s relatively uncontroversial amongst the mind-not-yet-killed.
There are some reasons to distrust one’s own baseline p(trait|appearance) estimates, other than the obvious (confounds, low sample size), say, particular personally experienced traumatic events, or exposure to explicit indoctrination on the matter. Most of us are not committing errors like: “in my experience, chinese people like chocolate ice cream more than strawberry, while everyone else prefers strawbeery; therefore chinese genetics code for chocolate-preference.”
The main problem is not that you can’t, or even shouldn’t, draw conclusions based on personal appearance. The problem is that obvious, superficial differences are very easily observed and remembered, and so seem to carry more weight than they deserve.
For instance, upon observing one woman and three men exhibiting Annoying Behavior X, many will immediately go for “it must be a guy thing” rather than looking for more powerful explanations, for instance all four people sharing the same profession, or being from the same geographic region, or any number of even more subtle things.
I’d like to test myself somehow, to find out how often I make mistakes along those lines, but nothing occurs to me right now. Yes: just because I have no reason to be especially biased toward making positive or negative associations toward e.g. pretty vs. ugly people (let’s not even consider race), it doesn’t follow that I’m free from a general tendency to form and cling to assocations from chance, or interpreted as causal without recognizing confounds.
Link to cool study/test, anyone? Such demonstrations on Overcoming Bias are the primary reason many of us are here today.
Unfortunately, I expect this sort of thing to be difficult to deliberately test an individual on, because if someone goes in knowing what’s being tested, or figures it out from the test, it’s going to alter the results beyond use. Self-testing may not be possible at all.
I recall having read about blind studies being done on related topics but, alas, I am terrible about keeping organized links to such things.
That I have some implicit association doesn’t actually tell me I make errors in thinking (but maybe if I’m distracted, my errors will tend in the direction of my implicit association?)
That’s a reasonable starting point, but I think his argument about the prevalence of stupid generalizations about “people of race X”, especially taken to wrongly prove “race X has more of a genetic predisposition for Y”, suggests a convincing case against your stance, unless you only mean to “listen” with great discrimination.
I’m not going to shout down people who make observations about group behavior just because their observations haven’t been tested in a double blind trial yet. the data precludes us from making certain generalizations. it doesn’t stop the tentative creation of new ones.
if I made a generalization about people with fingers of certain length it wouldn’t generate nearly this much ire. we shouldn’t treat race any differently just because people made stupid generalizations in the past.
We don’t have enough data to make the case for OR against any racism (biological differences → behavioral differences)
If a topic tends to historically collect relatively more stupid generalizations than other topics, isn’t it reasonable to keep a stronger default prior against such generalizations that aren’t backed by data?
So, lacking data, what is the null hypothesis?
That’s an important point (that others’ generalizations on race are suspect by default). However, I’m perfectly happy to consider appearance (including race) as a conditioning variable in my own thinking. I guess it might be smart to not admit this, but I think it’s relatively uncontroversial amongst the mind-not-yet-killed.
There are some reasons to distrust one’s own baseline p(trait|appearance) estimates, other than the obvious (confounds, low sample size), say, particular personally experienced traumatic events, or exposure to explicit indoctrination on the matter. Most of us are not committing errors like: “in my experience, chinese people like chocolate ice cream more than strawberry, while everyone else prefers strawbeery; therefore chinese genetics code for chocolate-preference.”
The main problem is not that you can’t, or even shouldn’t, draw conclusions based on personal appearance. The problem is that obvious, superficial differences are very easily observed and remembered, and so seem to carry more weight than they deserve.
For instance, upon observing one woman and three men exhibiting Annoying Behavior X, many will immediately go for “it must be a guy thing” rather than looking for more powerful explanations, for instance all four people sharing the same profession, or being from the same geographic region, or any number of even more subtle things.
Example: xkcd
I’d like to test myself somehow, to find out how often I make mistakes along those lines, but nothing occurs to me right now. Yes: just because I have no reason to be especially biased toward making positive or negative associations toward e.g. pretty vs. ugly people (let’s not even consider race), it doesn’t follow that I’m free from a general tendency to form and cling to assocations from chance, or interpreted as causal without recognizing confounds.
Link to cool study/test, anyone? Such demonstrations on Overcoming Bias are the primary reason many of us are here today.
Unfortunately, I expect this sort of thing to be difficult to deliberately test an individual on, because if someone goes in knowing what’s being tested, or figures it out from the test, it’s going to alter the results beyond use. Self-testing may not be possible at all.
I recall having read about blind studies being done on related topics but, alas, I am terrible about keeping organized links to such things.
The link you are looking for.
I’ve seen that, but you’re right, it’s related.
That I have some implicit association doesn’t actually tell me I make errors in thinking (but maybe if I’m distracted, my errors will tend in the direction of my implicit association?)
the null hypothesis for me is that i’ll listen to a wide variety of other people’s hypothesis.
I’m not sure you’re clear on what a null hypothesis is. Your statement sounds to me more like the wrong kind of humility.
That’s a reasonable starting point, but I think his argument about the prevalence of stupid generalizations about “people of race X”, especially taken to wrongly prove “race X has more of a genetic predisposition for Y”, suggests a convincing case against your stance, unless you only mean to “listen” with great discrimination.