U.S. Presidential race - Democrat Discussion

[ptwonline]

Uh...maybe I'm missing something in your logic ptwonline -- but if they went out and asked 100 voters who they will vote for and they came up with 60% dems and 40% reps....and did this over and over and over and over with the same results, wouldn't that be the fair results?

Throwing out some of the dems to make it even would just get an even poll and not actually reflect the voting population.

It isn't oversampling democrats if there are just flat out more democrats.

Not saying that's the case, but artificially forcing the polls to have equal dems and reps regardless of the actual numbers seems just as faulty.

If they kept coming up with 60/40 Dems but historical actual voting results are about 37/35 identified as Dems (balance being independents), then that would indicate that there could be something seriously wrong with your polling methodology that is not accounting for the differences between poll respondents and actual voter behavior.

[Lenonis]

If they kept coming up with 60/40 Dems but historical actual voting results are about 37/35 identified as Dems (balance being independents), then that would indicate that there could be something seriously wrong with your polling methodology that is not accounting for the differences between poll respondents and actual voter behavior.

Aren't these polls supposed to give information on their polling methods? At least somewhere?

I'm not entirely sure how you would consistently get one side or the other unless you are targeting specific areas (like big cities, or blue states) consistently. I think a lot of polls that are done over the phone tend to bias republican because the skew of people who have land lines being older, which tends to also skew republican, so it is interesting how they are getting more dems.

Still -- the electoral map is an interesting story -- even on Rasmussen's site, which is absurdly republican leaning in just about every poll I see, has Obama with a healthy lead over Romney in electoral votes.

[ptwonline]

Aren't these polls supposed to give information on their polling methods? At least somewhere?

I'm not entirely sure how you would consistently get one side or the other unless you are targeting specific areas (like big cities, or blue states) consistently. I think a lot of polls that are done over the phone tend to bias republican because the skew of people who have land lines being older, which tends to also skew republican, so it is interesting how they are getting more dems.

Still -- the electoral map is an interesting story -- even on Rasmussen's site, which is absurdly republican leaning in just about every poll I see, has Obama with a healthy lead over Romney in electoral votes.

They do give some info, but it is also sometimes quite murky since they don't want to give their competition all their secrets. In particular, "Likely Voters" seems to be a mystery of how it gets decided.

And Rasmussen seems Republican leaning but they are one of the only ones not to have a strong Democratic advantage in party identification. He seems to try to keep it close to historical weighting which will tend to bias polls closer together, but based on historical results could actually be prudent. Since most of the other polls aren't and their polling comes up big +Dems, they show Obama up, and Ras shows it even or Romney up.

[belfpala]

I guess I haven't researched it much...

I'm voting for Obama this year. But if asked about affiliation, I'd say Independent. My state, thankfully, doesn't require me to declare a party.

Might that be a factor?

[Meleti]

They do give some info, but it is also sometimes quite murky since they don't want to give their competition all their secrets. In particular, "Likely Voters" seems to be a mystery of how it gets decided.

Speaking really, really informally here. Campaigns creaate LVs by looking at people who are in a voter database and who have voted in the past and/or are demographically likely to vote (which is a combination of age, gender, income, etc.) and then choosing a sample of the LVs to poll. I have no idea how polling agencies do it, because I don't think they have the information that campaigns have.

[Fulci]

I guess I haven't researched it much...

I'm voting for Obama this year. But if asked about affiliation, I'd say Independent. My state, thankfully, doesn't require me to declare a party.

Might that be a factor?

No, Just make sure you're registered.

[Meleti]

Depending on where you live you might need party affiliation to vote in primaries, though.

[smelltheglove]

I guess I haven't researched it much...

I'm voting for Obama this year. But if asked about affiliation, I'd say Independent. My state, thankfully, doesn't require me to declare a party.

Might that be a factor?

only if you want to vote for someone different in a particular party's primary

[belfpala]

only if you want to vote for someone different in a particular party's primary

Both parties have open primaries here. I could technically vote in both.

I was referring to the "oversampling" of democrats.

[Spectral]

DI've been looking at some polls lately and hearing conservatives complain that the polls are biased towards Obama because they oversample Democrats. So naturally the first thing I do is to see if that claim is actually true. And what do you know: the major polls seem to use--on average--about 6-8% more Democrats. So naturally they would show Obama higher in the polls, right? Especially since historically, turnout between "Democrats" and "Republicans" (various ways counted: self-identified, registered, etc) is only slightly higher for Dems.

But here's the thing: as far as I can tell they don't choose Dems or Reps, they simply poll and party identification is one of the things they ask. They don't throw out parts of the poll or re-weight it to make the Dem and Rep numbers more even, since that could be pretty much pre-determining the poll results. But then why are Dem numbers so much higher?

So is it possible that the polling is faulty and that Romney and Obama could actually be roughly tied? Er, tied in polls that is.

No. Polls that oversample a party relative to their prevalence in the general population are corrected in their final results for this sampling error. If that's not done, it's a faulty polling method, but it's done in nearly every poll.

[Wells]

No. Polls that oversample a party relative to their prevalence in the general population are corrected in their final results for this sampling error. If that's not done, it's a faulty polling method, but it's done in nearly every poll.

Yeah in any official poll release at the bottom of the paper where it talks about the sample make up it gives the weighting they applied.

[Spectral]

To be fair, there was a poll that oversampled Dems and didn't correct, back in April or so. I think it was done by ABC if I recall. The result was Obama leading by like 11 points, which led to irrational elation from Dems and irrational fury from Republicans. I can't recall what the explanation was for the mistake.

[KingHorse]

You didn't really indicate in your response whether or not you are for the idea (I'm inclined to say you aren't) however, I would say that this statement of yours is a good reason to HAVE fact checkers -- force them to come up with a platform that is defensible without gross misrepresentations.

Not sure why you would be inclined to say that. Still assuming I'm a Republican parrot? I'm still not.

I would be in favor of it so long as it didn't stop the flow of the debates too badly. Something with a structure like:

Fact checking is ongoing, but results are only displayed/discussed during breaks. Both candidates get access to the results from both sides during the break, and can discuss that information (in their allotted time) when the debate resumes.

[Spectral]

Honestly, I think fact checking during debates would just lead to even more fact-free content than what we already have. I do not desire this.

[Naxere]

Glad to see Obama has time for Letterman appearances, to meet with Jay-Z and Beyonce, and for Talk like a Pirate Day, but not time to meet with Israel to discuss Iran.


http://news.yahoo.com/blogs/ticket/o...-election.html

[Nadev]

Who even watches Letterman anymore? What constituency is there for Obama?

[obdigore]

Who even watches Letterman anymore? What constituency is there for Obama?

The Elderly I assume?

[Nadev]

The Elderly I assume?

I imagine he's too crass for the average geezer.

[obdigore]

I imagine he's too crass for the average geezer.

I've never watched Letterman, but from what I understand except for Conan, you mostly have the baby boomers and older that watch the late night talking heads.

I don't know anyone (I'm around 30) my age that watches any of them except a few that watch Conan.

[FusedMass]

Obama is the President, are we to fault him for appearing on a comedy show. No. He is reaching out to his voters. Like he bravely did with Reddit. You think Romeny would have went online. Not in a billion years. Obama is the first President to be re-elected coming this Nov, to do this.