They use this report by Rasmussen to prove that more R’s than D’s exist.
Sure enough, most polling firms do sample more Dems. So it begs the question - are there more Republicans or Democrats going into the November election?
The “grievance” isn’t that there are (depending on what source you use) more registered Democrats than Republicans, but rather that the vast majority of polls are simply oversampling the number of Democrats who are going to turn out in relation to Republicans and Independents in order to make it seem Obama is ahead by more than he is or winning when he is not. Case and point, this wonderful PPP poll of Florida, which has Obama leading 50 - 46 in a 44D/36R/20I sample. In case you’re wondering, in the great Democratic wave of 2008, turnout was 37D/34R/29I (and since then both Republicans and Independents have seen their numbers swell, the latter more than the former).
If you believe that such a split will occur when-- at least according to that poll-- Obama has lost a significant amount of support from most groups from 2008, then I would like to sell you some tickets to Super Bowl XV.
Those tickets probably have some value to a collector, but before I can make an offer, I have to check with an expert. And if I buy them, Chumlee will probably use them to light his bong, and the Old Man will blow a gasket.
It’s possible there is oversampling going on, but most polling companies don’t adjust for party affiliation. Maybe Democrats are just that enthused, even more so than in 2008. I think it’s unlikely, but when have the pollsters been consistently wrong? Not since 1948?
The only thing that makes me think they might be is Gallup. Either the polls converge as we get closer to the election, or someone is going to be right and someone is going to be wrong.
I’ll bet that sounded more profound in your head.
Most every polling company does adjust for likely turnout based on history and the D/R/I split among voters in each state. If they didn’t, they’d have Democrats winning most every election as (again based on what source you use) there are more registered Democrats than Republicans, though it differs from state to state. Now ever since I saw that D+13 poll from Pew a few months back (among others using similarly eye popping splits), I’ve pretty much stopped looking at every poll except for Gallup, Rasmussen and ABC. Dating back to 1984, the highest split was D+7 in 2008, and that was a combination of low Republican enthusiasm and high Democrat enthusiasm. Don’t expect something like that this year.
/conservativewhining
That’s what I’d like to think, that we can’t possibly have more than D+7 this year.
Another thing that makes me wonder is the sudden increase in Obama’s approval rating and optimism in the economy. That definitely seems like a symptom of oversampling Democrats. Even if more people were getting behind Obama I can’t think they’d suddenly start thinking everything is getting better all of a sudden.
Linden Arden and OMG,
Please see this thread on party identification. Rasmussen and Gallup fail to see the decrease in GOP ID that the other polling houses all see. Other firms see more voters switching from calling themselves Republican to calling themselves Independent. The issue is why the divergence between the universe of reputable polling firms and these two.
Party affiliation is determined before applying a LV screen. It is applied to the random sample after that sample has been adjusted to meet whatever demographic standard is chosen. One does not explicitly correct for party ID, it is a number that emerges after the dataset is massaged to meet the that chosen demographic mix standard.
Rassmussen robocalls only. Right off that misses all cell phone only voters who are identified to skew Democratic. Rasmussen also has a long history of a GOP leaning house effect and a poor track record of reliability (see link provided in that thread).
Gallup is not so easily dismissed but it is (again see link and summary in that thread) apparent that the methods they use to identify the racial composition of their sample and the methods they then use to adjust that data into a demographically valid sample result in a different racial breakdown than the other firms obtain. Perhaps they have it right and everyone else has it wrong. You can hope. But the discrepancy seems to emerge from those methodologies.
Likely voter screen is another issue altogether. That tends to revert the RV polls towards the GOP on average by 2ish points.
The exodus from GOP identification should not be surprising to a party applying strict guidelines over who is extreme enough to be a member and disparaging of anyone who has any tendency to moderation on any particular issue as a RINO or worse. Those made to feel they do not belong are not suddenly likely to become Democrats, but ceasing membership in the GOP club? Sure. Add in a leader/candidate who articulates no real vision and the results in party ID are a bit predictable.
It goes something like this:
“Everybody I know hates Obama. The economy is still in the tank. No way Obama wins … heck, Romney is going to win in a [thread=665431]landslide[/thread]!”
Well, the polls aren’t showing that. They show a very close horse race, with Obama actually holding a small but consistent lead.
“That can’t be right! Nobody I know is voting for him. Nobody with any sense could possibly vote for him. Let me see those polls … A-ha! Oversampling Democrats! Your polling is wrong! Wrong, I tell you! It’s 400 electoral votes for Romney!”
Polling science has come a long way since Dewey “defeated” Truman in 1948. Of course they are not perfect, but they all publish margins of error. To assume that ALL the polls except the ones that lean Republican are wrong, and there’s a secret, untapped wave of Romney voters getting ready to roar to the polls in November, is little more than wishful thinking. If anything, the untapped wave may very well be voters without landlines, who are more likely to be young and more likely to vote - gasp - Democratic.
Now, the margin in most polls is very close. I am not brave enough to call the election today. But I am willing to accept the polling numbers are a fairly close approximation of how the electorate is feeling right now. Looking for errors in the polls because they don’t agree with your “gut” is a fool’s game.
Honestly, the only polls that really matter for Romney at the moment are Ohio, Virginia and Florida. If he doesn’t win all three … heck, if he doesn’t win Florida … I think he’s toast.
And yet Pauline Kael gets smugly trotted out any time Democrats get frustrated when Republicans take the lead in the polls.
But remember, “better” is a necessarily relative judgment, a comparison. Besides, everyone’s circumstances and standards of “better” are different, national averages notwithstanding. So I can see how it could happen.
Yeah, but why would people see it getting better only just now? Democrats have been more optimistic because Obama’s in office, so you’d expect optimism to rise if you add more Democrats to the samples. But if people are flocking to Obama because they just don’t like MItt Romney you wouldn’t see people start to like the economy better.
Because when it’s not election season they’ve got shit to do.
I happen to disagree entirely with that analysis. I’m more inclined to believe Gallup/Rasmussen’s numbers as they seem to be more inline with observable swing state trends. A quick glance at the voter registration numbers in a couple of swing states (that I could find the data for) show an increase in the number of Independents at the expense of Democrats, while the number of people identifying as Republicans showed minimal change (and in many cases, an uptick).
First, Colorado.
This is the partisan breakdown of Colorado as of 8/01/2012 (active).
709,496 Democrats
819,398 Republicans
695,921 Indepents
2,246,362 total registered voters
Broken down as percentages:
31.58% Democrat
36.48% Republican
30.98% Independent
This was the breakdown on 12/30/2008 (active).
842,484 Democrats
857,753 Republicans
725,971 Independents
2,440,011 total registered voters
Broken down as percentages:
34.53% Democrat
35.15% Republican
29.75% Independent
Do you see a great migration towards the Democrats? I don’t.
Next, Florida.
As of 2012, this is the partisan breakdown of registered voters in the state:
4,627,929 Democrats
4,173,177 Republicans
2,782,261 Independents
11,583,367 total registered voters
Broken down as percentages:
39.95% Democrat
36.03% Republican
24.02% Independent
This was the breakdown in 2008:
4,800,890 Democrats
4,106,743 Republicans
2,504,290 Independents
11,411,923 total registered voters
Broken down as percentages:
42.07% Democrat
35.99% Repulican
21.94% Independent
Do you see a great migration towards the Democrats? I don’t.
Next, North Carolina.
This was the breakdown on 08/04/2012.
2,753,838 Democrats
1,992,465 Republicans
1,609,644 Independents
6,370,984 total registered voters
Broken down as percentages:
43.22% Democrat
31.27% Republican
25.27% Independent
This was the breakdown on 12/27/2008.
2,870,500 Democrats
2,005,482 Republicans
1,402,471 Independents
6,282,575 total registered voters
Broken down as percentages:
45.69% Democrat
31.92% Republican
22.32% Independent
Do you see a great migration towards the Democrats? I don’t.
Next, Iowa.
This is the partisan breakdown of Iowa as of 8/02/2012 (active).
598,995 Democrats
620,584 Republicans
659,838 Independents
1,881,145 total registered voters
Broken down as percentages:
31.84% Democrat
32.99% Republican
35.08% Independent
This is the partisan breakdown of Iowa as of 12/01/2008 (active).
708,996 Democrats
598,580 Republicans
727,979 Independents
2,036,912 total registered voters
Broken down as percentages:
34.81% Democrat
29.39% Republican
35.74% Independent
Do you see a great migration towards the Democrats? I don’t.
And finally you can go here for Pennsylvania. There are a bunch of .xls files you can download for ease of access. You can clearly see a not-so-good trend for the Democrats in that state (though I don’t expect Romney to win it).
I’d be willing to bet that if there is any significant move towards the Democrats, it’s in states which are overwhelmingly Democratic in nature as is. Simply looking at the data above from 2008 to 2012 tells me that there isn’t this “mass exodus” away from the Republican party that you’re portraying (at least not in the above looked at states). In fact, those trends most closely mirror the trends in Rasmussen’s/Gallup’s analysis than it does the polls in HuffPo’s analysis. Now outside of the Republican base being REALLY deflated this election cycle-- you simply aren’t going to see anywhere near the same splits we saw in the 2008 election cycle. And surely nothing exceeding it in the Democrats’ favor. Hence my initial post. I’m pretty confident in saying that most pollsters are oversampling Democrats and assuming some ridiculous turnout in order to make it seem as if Obama is leading when he’s not or is leading by a greater percentage than he actually is.
BTW> Every criticism of Rasmussen invariably seems to come back to Nate Silver. I find that a bit odd, as Nate’s analysis of Rasmussen is the only one I’ve seen which rates his polls poorly. Generally, whenever I look it up, his polls are rated highly. I’m pretty sure Nate simply penalizes Rasmussen because they put out more polls than other polling firms in less known races (as well as earlier), which means their polls are more susceptible to errors.
That’s good analysis. The only way Democrats could be legitimately increasing in number is if registration drives are having unprecedented success.
OMG,
What part of the analysis do you disagree with? The question is why is there such a huge divide between the universe of pollsters assessment that there have been fewer RVs willing to identify as Republican and these two. The analysis is that there are methodological differences, as basic as Rasmussen’s robocall landline only approach, and several different factors for Gallup, that result in a different demographic balance.
Let me phrase the question slightly differently so you can find it palatable: what is everyone else doing wrong that they are systematically all finding fewer RVs willing to self-identify as GOP? Or do you believe there is a systemic effort to sample more (or to “correct” the samples to create the appearance of more) who identify as Democratic?
Btw the polling question about party identification has nothing to do with which party someone is registered with. Being registered with a party is a pretty sticky item (changing it requires effort and it is often only done at time of registration or in anticipation of primaries in states that require registration before voting in a party’s primary process); voter identification as polled is a fluid entity that can vary month by month. The drop in that polling identification as GOP is one that was only seen over the half a year (see this graph) and is not something that party registration would reflect.
That definitely sounds like wishful thinking.
Yes, it’s wishful thinking. No one has ever won money betting that the polls are wrong in a Presidential race, at least not since 1948.
But I’ve been watching polls since 1992 and I’ve never seen anything like this. I guess you can’t call Obama a “comeback kid” since he was always leading or tied, but his bounce is across the board. Obama’s up, his approval’s up, economic optimism is up, congressional generic ballot is up for Democrats, and individual Senate races are up. As in, all of them.
Only one of these things can be true:
-
Everything is going right for Democrats. No local issues dragging individual candidates down.
-
Pollsters are oversampling Democrats and started doing so right after the convention.
As unlikely as #2 seems at first glance, I’m telling you, never seen anything like this before. Usually improvement in poll standing is driven by independents moving from one candidate to another. I’ve never seen polls just show more Democrats appearing out of thin air.
Candidates have ‘coat-tails’.
So, OMG and Adaher, if the election were held today, what do you think the EV results would tally, after you correct for the (apparent) errors in polling?