October 19, 2004

1:55 PM

Every day, we are"treated" to new poll numbers showing Kerry and Bush at various percentages (Bush is leading, and seems to be widening that lead, according to most, except on that Wacky electoral-vote.com thing, but whatever...)

What I don't understand is how these percentages correlate with (or do they describe?) Electoral Votes. Does anyone understand how these correlate/connect? Can someone explain?
posted by ParisParamus to (29 comments total)

National poll numbers have nothing to do with the electoral vote (or really the presidential election). It only indicates who the possible winner of the popular vote will be, which, as you know, is worth beans.

Electoral-vote.com and other predictions of the results of electoral college voting take polls at the state level, not nation-wide polls, average them and then assign the state's electoral votes to either Bush or Kerry depending on who is polling higher in the given state.

Is that your question?
posted by loquax at 2:04 PM on October 19, 2004

The short answer is, they don't.

National Polls are just that- they're numbers across the nation. Electoral College votes line up roughly, but not exactly, with population, and are only specific to individual states. In addition, the "winner take all" approach of most states w/r/t their electoral votes means that it doesn't matter how wide the margin is. In other words:

1000 for Kerry, 1000 for Bush = 50/50 split.
State X has 1500 people and 15 votes.
State Y has 500 people and 5 votes.

If 500 Bush supporters live in State Y and 500 in State X, and all 1000 Kerry supporters live in State Y, then Kerry still gets all 15 electoral votes from State X, and Bush gets 5 electoral votes from State Y. So, although the popular vote is evenly split across the two states, Kerry still beats Bush in electoral votes, 15-5.
posted by mkultra at 2:05 PM on October 19, 2004

There has also been an awful lot of discussion on the web and in the mass media about how much relevancy these polls have.

If they are polling "likely voters" there is a good chance they are not talking to any of the folks that have registered this year (although we can debate what % of those folks will vote). There is also a growing concern that people being polled, even in a random sample, are becoming more and more self-selected -- e.g. I never answer my phone if I don't recognize the number.

Additionally, I second what loquax said. The only polls that do matter, if they matter at all, are the state-level polls which may give an indication as to how each state is leaning.
posted by szg8 at 2:11 PM on October 19, 2004

Thanks. Well, then, I guess the poll numbers are only relevant in terms of trends. Which isn't made obvious by the media. How very lame of them. Now, if we could only figure out if tele-anonymity skews left or right. My guess is right, i.e., the poor, and those of modest means are less likely to have cell phones and use caller id, so they may be oversampled in surveys.

(on a lighter note, I hate the Evil Empire Yankees, but I don't want ANYONE from Massachusetts winning this fall. So, for the first time in my life, I want the Yankees to defeat Boston. And then to lose to that Texas team!)
posted by ParisParamus at 2:20 PM on October 19, 2004

What prevents the electorate from voting against the popular vote in their state (i.e. an electorate from NY voting for Bush or a third party candidate)? As I understand it, there aren't any legal ramifications.
posted by Raze2k at 2:26 PM on October 19, 2004

My guess is right, i.e., the poor, and those of modest means are less likely to have cell phones and use caller id, so they may be oversampled in surveys.

Well, it also skews older, and seniors are supposedly some of the most likely voters, so that may add a little weight. Not sure.
posted by PinkStainlessTail at 2:29 PM on October 19, 2004

An interesting discussion of the trouble with polling this time around.

And an interesting discussion about faithless electors.
posted by loquax at 2:37 PM on October 19, 2004

Pay zero attention to the big media polls. They mean nothing, because they don't factor in the electoral breakdown, and no one in the big media will dare to say anything much but "it's a close race" at this point. Election night coverage will be interesting this time around, given the Florida debacle last year. Expect a big show of autheticity and caution in reporting the returns.

I think you should really sit down with "that Wacky electoral-vote.com thing" and figure out what it's about if you want to see how depressed or happy you should be. Kerry has been tracking well for a few days according to it, but Bush has been reported widely ahead there for weeks at a time. The guy who runs it is openly liberal but seems dedicated to making that transparent, and not letting it skew his methods.

I thought you were taking an election season sabbatical, Paris? I wasn't begruding you your interim use of AskMe, but this question makes it seem like you just can't stay away.
posted by scarabic at 3:11 PM on October 19, 2004

Pollsters were quite off in the recent Canadian election, I think changing demographics and the prevelence of cell phones might have something to do with this. Wouldn't surprise me if a lot pollsters end up scratching there heads after the U.S. election.

I've heard that in Canada's case, the pollsters got pretty close with the percentages but when they tried to extrapolate that to figure out how many seats each party would win (sorta like the electoral) they messed up.
posted by bobo123 at 3:20 PM on October 19, 2004

Well, you know what President Dewey always said about the reliability of polls...
posted by Sidhedevil at 4:02 PM on October 19, 2004

National "horserace" polls are meaningless--more to the point, quite a few voters could change allegiances in California and Texas, enough to change the "horserace" polls by a point or two, and it wouldn't make diddly squat of a difference in the Electoral College.

electoral-vote.com has some nice graphs that let you pull out the numbers reported by single polling organizations state-by-state (like, a graph of only Rasmussen polls in Ohio versus a graph of only Strategic Vision polls in Ohio).

race2004.net has nice state-by-state tables of polling trends for each state.
posted by gimonca at 4:27 PM on October 19, 2004

This is an interesting look at what the polling numbers really mean, showing how the incumbent's horserace number (not the spread) and his approval numbers are more important than the margin between the two.

LA Times

And here's lots more polling goodness.
posted by CunningLinguist at 5:03 PM on October 19, 2004

My husband says the place to go to is 2.004.
posted by evening at 5:03 PM on October 19, 2004

ParisParamus - who says the trending is toward Bush? You trying to pull a Rove and convince the undecideds to go with the "winner"?

My source thinks the trend is toward Kerry.
posted by coolgeek at 7:35 PM on October 19, 2004

I listened to a certain radio show early this morning. True, he's officially right wing, but he recited nationaL polls that showed Kerry falling behind.

I don't know any more.

Scarabic: your question has merit, but I distinguish between AskMetafilter, and Metafilter/Metatalk. This post, admitted is in a gray area, but I honestly wanted an answer to this question.

I'm already scared to death that Kerry may win..

Please don't hate me.
posted by ParisParamus at 7:43 PM on October 19, 2004

Don't obsess over the polls (and this goes for supporters of any candidate). We'll know how it turns out on the evening of Nov. 2. (Or, at the very least, know about one or a few states where the election was very close and upon which the results hang.) Reading every new poll between then and now isn't going to change the outcome. And frankly, you're unlikely to get any reassurance from any polls that come out between then and now that you don't already have.

If you could figure out how the election was going to turn out from the polls, we wouldn't need to have an election. You won't know how it turns out until Nov. 2, at the earliest, and no amount of poll-watching will change that.
posted by DevilsAdvocate at 9:29 PM on October 19, 2004

Hate you? I should thank you. It's a pleasure to watch you squirm a little after months of your blustery "when Bush wins" talk.

Believe me, I feel your pain (in reverse) but I don't see why you're letting it get to you. If you understand what polls really are, then you understand that all they're telling us right now is "we don't know yet." There's no reason to jump for joy or drop fudge right now.

But... inasmuch as the national polls are what you're looking to for reassurance right now, they're even less meaningful than the electoral breakdowns. If you're going to look at anything, look at this.
posted by scarabic at 8:46 AM on October 20, 2004

I'm already scared to death that Kerry may win

Paris, maybe the fact that your editorializing has already been edited out of this thread at least once should be a clue to you to stop doing it? We all know who you're going to vote for. Give it a rest.
posted by ook at 8:48 AM on October 20, 2004

I don't hate you for your political views. I hate you for posting this piece of shit on AskMetafilter. This is fucking unforgivable.
posted by Eamon at 9:05 AM on October 20, 2004

Raze, electors in a number of states are now legally bound to vote in favor of the candidate their constituents vote in favor of.
posted by Yelling At Nothing at 9:38 AM on October 20, 2004

Eamon: the question was entirely genuine, and any tangent I offered was for levity.
posted by ParisParamus at 10:08 AM on October 20, 2004

you really can't let go, can you?

please start another MetaTalk thread about you disliking the politics of this place and taking a "sabbatical", ParisBunnyfire, go ahead
posted by matteo at 11:24 AM on October 20, 2004

I don't hate you for your political views. I hate you for posting this piece of shit on AskMetafilter. This is fucking unforgivable.

You've really got a bit of an attitude problem Eamon.
posted by loquax at 11:33 AM on October 20, 2004

You've really got a bit of an attitude problem Eamon.

I know that it's only a matter of time before AskMe suffers the fate of MeFi, but that doesn't mean I'm going to be quiet about it.
posted by Eamon at 1:32 PM on October 20, 2004

Maybe try doing it without the "hatred" next time?
posted by loquax at 2:07 PM on October 20, 2004

This is the matrix I was sort of refering to, originally:
http://www.realclearpolitics.com/polls.html
posted by ParisParamus at 5:53 PM on October 20, 2004

I'll refrain from editorializing about Paris and say only this: It really is a shame that "the media" is doing such a shoddy job of reminding the unwashed masses about how the system "works."

It's been quite some time since I took a stats class, so I wouldn't hazard a guess as to exactly how, but there's no question in my mind that the constant, yet effectively irrelevant polls seriously impact the results of the actual election (especially given how many of them insist on declaring a "winner" in a statistical tie).

You can argue that it's the people's obligation to know this and that the media itself isn't obliged to remind them (or you can even trot out that tired old "there is no one 'media" argument) but personally, I think this is a sterling example of the industry-wide failure Jon Stewart has been talking so loudly about.
posted by Sinner at 6:20 PM on October 20, 2004

PP, your concept of "levity" needs a bit of work.

I'm kind of embarrassed for you right now, actually.
posted by Sidhedevil at 6:27 PM on October 20, 2004

You go Eamonn!

in girly voice: He'll save us all!
posted by scarabic at 10:17 PM on October 20, 2004