« Will's theorem | Main | The Power of Oprah to help Obama »
The nature of ability bias
Remember all those studies showing that people claim they are above-average drivers? Or above-average at other things they do? It may not just be self-deception. Here is the latest:
...we find it easier to consider the favourable evidence for a single person than we do for a whole group. Consistent with this is the finding that people tend to be biased when comparing any single individual, not just themselves, against a group of others.
There's also the possibility that we're biased towards the "target" in any comparison. The "target" is the entity that is being measured up against some benchmark. Following this logic, if I asked you how good all other drivers are compared with you (thus making other drivers the "target" of the comparison and you the benchmark), then this ought to reduce the bias you'd show towards yourself.
...A new study has tried to get to the bottom of what causes the "above average effect" by pitching these three explanations against each other. Zlatan Krizan and Jerry Suls Dozens asked dozens of undergraduates to list a group of friends or acquaintances, to take one member of that group and then compare that individual with the rest of the group on some attribute - say, generosity.
Of the three factors, our difficulty in seeing the quality of a group, relative to the quality of an individual, seems to be the primary source of bias in the ranking.
Fortunately, I am better at avoiding that bias than are my readers.
Posted by Tyler Cowen on August 5, 2008 at 05:49 AM in Science | Permalink
Comments
Of the three factors, our difficulty in seeing the quality of a group, relative to the quality of an individual, seems to be the primary source of bias in the ranking.
Fortunately, I am better at avoiding that bias than are my readers.
Why, yes, yes you are better at avoiding that bias than are your readers.
Posted by: chug at Aug 5, 2008 7:45:58 AM
"Fortunately, I am better at avoiding that bias than are my readers."
But not me or chug specifically, of course.
Posted by: odograph at Aug 5, 2008 7:58:37 AM
Isn't this what you'd expect from a classic Roy model? Self-selection on the basis of sector-specific skills can imply that all individuals are above average in their chosen field. Generalizing from sectors, there are many different kinds of driving--fast, careful, fuel efficient. In my mind, a good driver is a fast driver. I drive faster than average. Hence, I am above average. When someone else is asked, they think "I'm a careful driver, I drive more safely than others. Hence, I'm above average..."
Posted by: adam at Aug 5, 2008 9:11:24 AM
As an aside, I'm not sure there's a clear-cut definition of what constitutes a "good driver."
My rule of thumb is that a good driver does what all the other drivers are expecting him/her to do. By this definition, it's possible for everyone to be a good driver, given the assumption that other people's expectations are the same as yours.
Posted by: Rich at Aug 5, 2008 9:28:58 AM
The whole above average driver oddity is really meaningless when you get right down to it. "Rich" noted that people probably measure what it means to be a good driver differently than other people do.
Further, it's not like driving (legally on roads, not a race track) is very hard. I'll bet if you asked people whether or not they thought they were above average basketball players, you wouldn't find nearly the same bias since it's so easy to parse the good players from the poor. Two drivers could be very different in ability with no noticeable difference in actual driving.
Posted by: skimmer at Aug 5, 2008 10:07:15 AM
This is good.
Maybe part of why shorting is less effective than expected?
Posted by: Andrew at Aug 5, 2008 10:55:56 AM
The average driver has had more than zero at-fault accidents but I haven't had any so I have a legitimate claim at being above average.
It's possible for most people to be above the mean if the distribution is skewed, as driving skill probably is, with a long tail [in this case in the negative direction]. What's not possible is for most people to be above the MEDIAN.
-dk
Posted by: dick King at Aug 5, 2008 12:33:42 PM
As for the comments on driving, it is of course correct that different people have different ideas about what constitutes a good driver (same with good lover), but other people have thought of that before and altered the experiment accordingly. You can also show, for example, that most students in a course think they will get an above-average grade.
Posted by: LemmusLemmus at Aug 5, 2008 1:14:54 PM
I didn't look at all three factors, but the one I looked at seemed much better than the others.
Posted by: Colin Fraizer at Aug 5, 2008 1:41:30 PM
Except it doesn't explain the difference between men and women in rating their driving abilities.
Also, I'm totally better at avoiding bias than your average reader:
http://moaes.blogspot.com/2008/06/are-you-2.html
:P
Posted by: Robert Olson at Aug 5, 2008 9:32:50 PM
DK made my point, but I'll make it again in a different way. It is probably true that most people are above average drivers. There are a few really bad drivers that pull down the average. With driving it is possible to be really bad, but it isn't possible to be really good. I mean if you get places without crashing, skidding, or running off the road, you've probably done an average job. Is it possible to drive ten times better than that? Not really, but it is possible to drive hundreds of times worse, and sometimes people do.
Posted by: Andrew L at Aug 6, 2008 8:27:08 PM
"Fortunately, I am better at avoiding that bias than are my readers" was the best sentence I read today [followed closely by "But not me or chug specifically, of course"].
Posted by: Dave McDougall at Aug 6, 2008 8:41:55 PM
