We’ve reached the point in the election where the press decides to mostly report on how the election is being perceived rather than on any particular events, and since the president is doing well in the polls this results in a lot of “desperate Republicans do foolish things” stories. The flavor of the week seems to be the media’s discovery that somewhere out there in the right-leaning internet, there are people who have made a hobby of “re-weighting” polls in order to reflect what the re-weighters think is a more likely partisan composition of the electorate come election day.
There is, yes, a certain sad desperation about this. Now that election reporting is often more about “the race” than about issues or events, being behind in the race is crippling and so people come up with way to try to explain it away. Those with long memories (eight years counts as long in our modern age) may recall that when Bush was so rude as to be ahead of Kerry in the 2004 race, Michael Moore and those like-minded rolled out a theory that all the polls were wrong because an army of voters who only used cell phones and not land lines (and thus couldn’t be polled) were out there ready to vote against Bush.
However, just as everyone’s getting ready to announce that Republicans, in their constant flight from the “reality based community” have decided they don’t believe in polling, we find out that the left has its own reality problem: They’re convinced that the economy has been getting better over the last couple months, despite the fact there’s little reason to believe this. Gallup and the Pew Research Center both have data out showing that Democrats’ opinions of the economy and the job market have suddenly started improving, despite almost universally bad news over the last several months.
As you can see, partisan affiliation wasn’t much of a dividing factor in assessments of the economy a year ago, but now that a bad economy might mean President Obama not being re-elected, Democrats obediently come to the conclusion that the economy really isn’t that bad. According to Pew, the same divide now exists on the job market, consumer prices, the financial market, real estate, and even gas prices. You would think that at least people could agree on what the level of gas prices is, but no, apparently not, though the gap is narrower there than elsewhere: 89% of Republicans say they hear mostly bad news about gas prices while 65% of Democrats do.
The trope goes that you are entitled to your own opinions, but not your own facts. However, as the political divide has become wider and more entrenched opposite sides increasingly do have their own facts, as reality become filtered through a partisan lens.
The polls have tightened now that we are in October. In one day the poll average has fallen from 4.0 advantage Obama to 3.2 advantage Obama on the Real Clear Politics Average:
http://www.realclearpolitics.com/epolls/2012/president/us/general_election_romney_vs_obama-1171.html
This happens almost every presidential election cycle. A cynic might observe that the closer we get to election day the more that pollsters want their polls to be accurate.
As for some of the wilder polls that we saw in September showing Democrat turnout in states highers than 2008, the best election for Democrats since 1964, Jay Cost explains why they were bogus:
http://www.weeklystandard.com/blogs/morning-jay-are-polls-tilted-toward-obama_653067.html?nopager=1
As for the economy, it takes a special type of mindset to view it and not regard it as a disaster. I can barely understand someone thinking that Obama has done a bad job but Romney would do worse so they are sticking with Obama, but to deny the evidence before their eyes, what we have all been living through the past four years, is simply delusional.
Here are two answers to skewed polls: “Caller ID” and “voice mail.”
About 91% (see Instapundit post) of us that have caller ID and see a number we don’t know let the call go to voice mail; the lying liberal poll organization hangs up; and we delete it. I let it happen about six times a day.
Certainly, I hope that Rasmussen has been more accurate overall than a lot of the one-off polls, and I can believe that polling is tricky because it’s likely to be a low turn out election where victory relies primarily on who shows up to vote — but I think the efforts to do amateur poll re-weighting based on party affiliation are, while well intentioned, kind of silly. Party identification is one of the things you seek to measure in the poll, not one of the things you should weight it by. If you get way too many people of one party, that may indicate your sample is bad. But polls should only be re-weighted to fit non-changing demographics (sex, income, age, race) not changing demographics like party identification.
That said, I have a certain sympathy, at least, with the desire to fight the polls. Polling is necessarily imprecise and hard to understand, and it’s used far too often as a way to shape the vote. Insisting that the economy took a sudden turn for the better during the late summer, on the other hand, is fairly crazy.
Regarding the polls, I agree that Rasmussen is probably the best but even he is using a D+3 model which is why Obama is coming out on top in his calculations. Do people really believe that the democrats will have that much of an advantage on election day? Sorry – I’m not buying it. If Romney holds on to the independents, and the republican/conservative/libertarian/tea party people come out in big numbers and draw even with the democrats, which I think is very likely, than Romney wins. Better yet, if they match the 2010 election which was a R+1, then Romney wins comfortably. Even if you bring it down to a D+1, Romney still wins. Why don’t they start presenting polls that reflect this possibility? I would like the media to say: People, this is what we think will happen if the turnout is D+3, D+2, D+1, even, R+1, etc., etc. Some honesty from them would be refreshing.
I think what you neglect in all the controversy about polls is some problems with the sampling frames not as severe in previous years:
1. Low response rates generally.
2. Difficulties in contacting people who lack landlines.
3. Variable methods among pollsters which produce divergent results (manifest now in a way it was not thirty years ago).
4. Odd and novel biases in propensity to respond (manifested in exit polls eight years ago).
—
And, yes, the curious partisan balance in some published polls is an indication there could be problems with the sampling method used. We are not going to find out how serious the problems have been for another month.
I wonder how much the local job market colors these responses, especially with various industries doing better/worse. Prior to leaving Seattle, things seemed ok (not great, not horrible) with regard to employment (aerospace was ramping up). Here in Waco, things seem better. Heck, we are having a hard time filling new positions at my current employer. On top of that, the local politics are very different too.
[…] Last week, before the debate, I noted that Democrats were mocking Republicans for trying to explain … The debate came and Romney routed Obama on the debate stage in a way that exceeded my wildest hopes. Now we see an unprecedented post-debate surge for Romney in the polls, with Gallup and Rasmussen both showing Romney in a tie with Obama and a post-debate Pew poll showing Romney beating Obama by 4% among likely voters, a twelve point swing from Pew polling a month before in which Romney trailed Obama by 8%. […]