« Clive Crook | Main | Tilting at Windmills »

Candidate Attributes

15 Nov 2007 03:26 pm

Larry Bartels emails Ezra Klein some data about his research on the impact (in terms of statistical correlation) of perceptions of different candidate attributes on voting behavior: "The analysis was based on survey questions asking voters to rate presidential candidates on a variety of dimensions. Here are the estimated effects of those evaluations on voting behavior, averaged over elections from 1980 to 2000. (The numbers are not directly interpretable, but the relative magnitudes are.)" I turned the numbers into a handy chart, showing the average on the left and the 2000 result on the right:

caring.jpg

I'm not really sure what to make of this, though, as I sort of feel like people may tailor the characteristics they say they're looking for to suit the candidate they're going to vote for. Mainly, you want to be strong yet caring. Or caring yet strong.

Share This

Comments (17)

I'm not really sure what to make of this...

More blatant Paulophobia.

The chart clearly demonstrates that Dr. Congressman Ron Paul is the only man who can save America!

you want to be strong yet caring. Or caring yet strong.

And stupid.

Americans hate America. Their deep, abiding love of authoritarianism shows in that graph.

This would be appalling enough if it were what we wanted in our TV newsreaders.

That we prioritize any other fake attribute over any actual knowledge about the world for our president... well... that's just extremely disheartening. Hopefully I'm missing the larger picture about this data here somehow.

It would be interesting to see what seven years of Bush have done for the importance of "knowledgeable." One would hope that it's increased a bit.

Pretty graph. What did you use to make it?

Pretty graph. What did you use to make it?

Numbers -- part of Apple's iWork suite.

Yes, but what rap song does it represent?

What do the values of the y-axis mean? "Not directly interpretable" sounds like "meaningless" to me.

There has to be some underlying logic to them. Matt, can you supply it, or at least a link?

I checked EK's post, and it's not there either.

Why did "inspiring" drop off of the graph?

I would also want to have some more detailed data about the previous periods than is provided. Were they relatively stable around the averages given? I would expect this kind of data to have a large amount of completely random fluctuations. I don't think people really have that strong attachment to particular positive adjectives in their presidential candidates.

Not to mention the fact that most people just vote on party ID basis anyway. For it to (even potentially) mean anything at all for general elections, you would really only want to have this data for that subset of voters who use this kind of reasoning when casting their votes.

A 5.28 ratio of strength to knowledge. . . Hmm, there must be some kind of role-playing character, probably wielding a sword, that this matches. . .

Ah. I looked at Ezra's, and see that "inspiring" was, indeed, dropped from the polls. I should have read more carefully.

I guess 32 strength and 6 intelligence (5.33 ratio) might be reasonable for a high-level half-orc barbarian or something. I think you're right about the sword -- probably a two-handed greatsword or falchion.

Huh. I want my politicians to care about people who aren't like me. People like me are doing just fine, thanks.

the 2000 graph is flipping the bird, which seems about right.

Strong and caring...caring and strong...

Damn, this nation has some serious daddy issues.

How much does it suck that "knowledgeable" always comes in last and, apparently, no one even asks about "competent"?

I can't speak to the aesthetic beauty of the graph, but discontinuous horizontal lines won't exactly get you published in Science or Nature.

Also, why present time-sequenced data in the form of a bar chart? It looks like you're hiding a complexity.


Comments closed November 29, 2007.

Copyright © 2008 by The Atlantic Monthly Group. All rights reserved.