Comments on: Polls Are Smarter Than You

By: milieu

milieu — Thu, 23 Oct 2008 12:50:41 +0000

do you have a cite for this?
No, just what seems a plausible speculation on my part. So not going ahead with that to add to the noise :)

By: razib

razib — Thu, 23 Oct 2008 12:19:46 +0000

Many undecided might get influenced by the poll figures and vote for the leading or trailing candidate based on their psychology.

do you have a cite for this? this is a common perception, and i believe one of karl rove's theories. but one thing that would to work against this is that undecideds are low information voters, who probably aren't the types following pollster.com closely ;-)

By: razib

razib — Thu, 23 Oct 2008 12:14:19 +0000

Beats me why Drummond should get hit with a diatribe of GNXP invective for doing exactly what good bloggers are supposed to do. Give the guy some respect!

i'm appalled you found that drivel plausible, and that comment certainly reworks my priors in terms of taking you seriously. polls are imprecise; that does not entail that an elaboration which illustrates ignorance is worthwhile. not at all.

By: ben g

ben g — Thu, 23 Oct 2008 11:14:51 +0000

self-reference is fun and all, but why would we want to try and poll an alternate universe where polls didn’t exist?

By: milieu

milieu — Thu, 23 Oct 2008 08:22:21 +0000

Probably its not a good metric to measure a poll reliability by the final outcome. Since, the poll can itself have an influence on the final results. Many undecided might get influenced by the poll figures and vote for the leading or trailing candidate based on their psychology.
Maybe thats why the pres. campaigns have their own independent polls.

By: ben g

ben g — Thu, 23 Oct 2008 08:17:17 +0000

not gandhi,

Gore was underpolled by 3 points in 2000, compared to Bush being underpolled by 1 in 2004.

Looking through past polls I see you're right about Dole being underpolled in the last few days. Possible explanations could be last day movement toward Dole, or perhaps issues with Clinton's soft support not coming to the polls. (I'm too young to have been following that election in-depth, so I don't know what it could be.) Worth looking into, but I think it's safe to say that it's methodological and not a result of pollster political bias.

By: Tod

Tod — Thu, 23 Oct 2008 06:03:28 +0000

Pollsters wrongly predicted conservative defeat in the UK a while ago (might have been based on an exit poll), the explaination they came up with was that conservative voters are more likely to mislead them. As there is a suspicion attached to saying you won't vote for Obama there might be something in the idea that the polls overstate his lead. It is not obvious to me this would be good for him. A respected poller claimed the media were understating the lead he had over Hilary.
"fun to see him arguing"
Being tested for prostrate (or breast) cancer is not always a straightforward decision for informed people. The testing may well turn up certain relatively innocuous cancers that could be better left alone but oncologists will not do that.

By: Not gandhi

Not gandhi — Thu, 23 Oct 2008 05:46:40 +0000

I think you are exactly right and his arguments are wrong. Nonetheless there is the problem that empirically the polls seem to have overestimated Democrat support most of the time relative to the actual vote count in the last half dozen presidential elections or more. The last election seems to have been the most accurate but even then you had the problem with the early exit polls. Even in the Dole drubbing, Dole’s actual vote count was higher than the poll averages a few days before. So I wonder if it’s a problem in the sampling methodology?

By: Matt McIntosh

Matt McIntosh — Thu, 23 Oct 2008 04:53:54 +0000

When I see the word “fisking”, MEGO. It never, ever heralds a valuable use of anyone’s time. Ever.

By: bgc

bgc — Thu, 23 Oct 2008 01:41:56 +0000

I don't like the ad hominem tone of Razib's critique of what was IMHO a very interesting and informative post by DJ Drummond.

He (if it is a male) laid his political-interest cards on the table very honestly and provided a lot of relevant data and background.

The point is surely that polls near to elections are not the same as elections - they differ systematically in _many_ ways. People poll and vote using different criteria and methods - and it is interesting to explore ways in which polls and elections differ. And polls weeks or months remote from elections are even less like elections than just-before polls.

The real craziness is the majority (?) of people including pundits who apparently assume that polls done weeks/ months before the election will predict the election result.

Beats me why Drummond should get hit with a diatribe of GNXP invective for doing exactly what good bloggers are supposed to do. Give the guy some respect!

I say - Thanks for your post DJ Drummond. I'll be bearing in mind your arguments over the next couple of weeks.

By: razib

razib — Thu, 23 Oct 2008 00:21:45 +0000

page down to the list of polling data by date:
2008 vs. 2004. bush was leading in 2004, but obama is leading a lot more. i think this explains the greater magnitude of right-wing craziness in concocting theories. the left-wing had a greater probability of beating the odds, so ruy teixeira sounds a lot less crazy in his rationalizations because he had a few polls to hold on to 'till the end.

By: razib

razib — Thu, 23 Oct 2008 00:18:04 +0000

What I mean is that it's not rational to assume that the "most likely" outcome is the one that will happen.

depends on the shape of the distribution, right? there's a mode. and then there's a mode....

By: David Boxenhorn

David Boxenhorn — Thu, 23 Oct 2008 00:10:29 +0000

i don't get the "it's not rational to assume for the worst."

What I mean is that it's not rational to assume that the "most likely" outcome is the one that will happen.

By: David Boxenhorn

David Boxenhorn — Thu, 23 Oct 2008 00:05:07 +0000

that has nothing to do with this thread

I suppose so, neither does "there's a right way to deal with a loss", as the loss hasn't happened (yet).

In any case, as Ben says, the pollsters themselves are worried about systematic bias - which happens all the time in the real world. To my way of thinking about GNXP, that is the relevant topic, not how stupid some blogger might be.

By: razib

razib — Wed, 22 Oct 2008 23:49:01 +0000

The real world is full of surprises. It's not rational to assume the worst, even when the "odds" are against you. You should know about that (believe me, I do).

that has nothing to do with this thread. the wizbang-retard is in denial of the most likely outcome and is making stupid stuff up (or, is preening his ignorance, as detailed in ben's post).

now, as NH primaries shows us polling is not always right, so it is not the inevitable outcome. that being said, the reasons wizbang-retard gives are pretty mostly false or specious. the polling companies supposedly studied what went wrong in NH, but for whatever reason they haven't gotten around the releasing the results. it seems possible that they actually don't know what happened, but don't want to admit that. sometimes that's all you can say with any confidence. the dozen polls that are tracking the race right now might be off for a host reasons. if you want the outcome that isn't being predicted, just hope for it.

concocting bizarre conspiracy theories, taking the outlier poll as a sign of "hope," or highlighting that the *nickelodeon* poll shows the race tied just makes one look foolish. that being said, perhaps that's the sort of hope people want to have, they need something concrete. even if that concrete thing is a poll of elementary school age kids.

there's no excuse for not distinguishing between the possible and the probable, and the probable and the inevitable.

p.s. i don't get the "it's not rational to assume for the worst." i mean, i'm not trained as an engineer, but it seems that it is generally rational to assume for the worst in many situations so you can take the appropriate mitigating measures. or, if you run a small business, you shouldn't always assume best-case scenarios in terms of revenue so you don't go overboard on capital inputs.

By: David Boxenhorn

David Boxenhorn — Wed, 22 Oct 2008 22:53:59 +0000

“Look, there’s a right way to deal with a loss, and it isn’t retreating into a fantasy world”

The real world is full of surprises. It’s not rational to assume the worst, even when the “odds” are against you. You should know about that (believe me, I do).

(I put “odds” in quotes because we’re not talking about real chance, but uncertainty due to lack of information. There is no way to calculate this kind of “odds”.)

By: razib

razib — Wed, 22 Oct 2008 17:53:54 +0000

Look, there's a right way to deal with a loss, and it isn't retreating into a fantasy world:

for retards life is one big fantasy. a postmodern cornucopia if you will.

By: ben g

ben g — Wed, 22 Oct 2008 16:57:29 +0000

Several pollsters actually include cell phones, so I don't think it has a big effect. I mainly brought it up because it's comparable to the enthusiasm gap between McCain and Obama in slightly biasing the overall polling by perhaps a couple of points in either direction.

By: TGGP

TGGP — Wed, 22 Oct 2008 16:31:21 +0000

people who either have no home phone or use cell phones are typically not being sampled. These people are disproportionately young and/or poor (both demographics favor Obama).
I remember when all the LewRockwellites were denouncing polls that showed low support for Ron Paul, claiming that internet polls had greater reliability due to the cell phone issue. Ron Paul ended up getting just what the pollsters said he would.

Some other yahoo wrote a long post about how The Left has completely distorted our perception of who the likely winner will be in a post Dave Weigel called "a great big wedding cake of stupid".

By: gcochran

gcochran — Wed, 22 Oct 2008 15:25:56 +0000

It’d be fun to listen to this guy arguing with his oncologist. Look, there’s a right way to deal with a loss, and it isn’t retreating into a fantasy world:

“Thought shall be the harder, heart the keener,
courage the more, as our might lessens. “