On March 17, NPR published a poll under the innocuous headline "More Voters Think U.S. Is On Right Track." That poll had a rather surprising result on the question of generic Congressional ballot preference, with Democrats and Republicans tied at 42, and 15% undecided. Whoa. Is that true?
Well if you check the pollster trendlines, this seems entirely plausible:
Being a general believer in the utility and statistical veracity of polling, I can't claim to know for a fact that these polls are in error, but I can point to other evidence which makes even this aggregate group of polls unlikely in the extreme.
Absence of Corroborating Evidence
The biggest problem with this result is that it suggests that voters are losing faith in Democrats to the point of being willing to reconsider and re-elect Republicans. But the other polling results both in that NPR poll and in many others don't back that claim. First, as NPR's own title noted, voters are increasingly viewing the country as being on the right track:
5/08
3/09
Response
16%
31%
RIGHT DIRECTION
80%
63%
WRONG TRACK
Now, the right/wrong tracks are still overwhelmingly negative, but the 80% of voters who said "wrong track" in 2008 didn't boot the Democrats out, is it likely that the 63% saying "wrong track" today are planning to do so in 2010? But the plot thickens further:
Not only are voters increasingly viewing the country as on the right track, but they are increasingly approving of their federal legislative branch. 36A/56D isn't anything to be excited about, but the trend is clearly increasing. So again, not conclusive, but definitely suggestive that voters are not really gearing up to vote Republican. Let's get a little more partisan now:
Here's where I think the plausiblity of the NPR and Pollster aggregate is most in question. While the Democrats in Congress are not exceedingly popular, hovering around and below a net of 0, the Republicans are somewhere below "abysmal" and actually sinking. Voters are ambivalent about the Democrats in Congress, but they loathe and despise the Republicans. Collectively (and I could also point to Obama's strong approval and popularity numbers, a very important metric for the performance of a party in a mid-term election) these make a pretty strong case that the Generic Ballot polling is wrong.
So What Happened? - NPR
Aside from being a smaller sample (~800), the big problem I have with the NPR poll is demographics.
Question D4 on page 15 of the poll asks about ideology, and provides all the historical, an excerpt of which I've extracted:
5/03
1/08
5/08
3/09
Response
43%
40%
35%
45%
TOTAL CONSERVATIVE
16%
19%
21%
19%
TOTAL LIBERAL
45% is the highest proportion answering "conservative" in any iteration of this poll listed. Even going back to the heady days of the post-invasion period of apparent success in Iraq of May 2003, only 43% said they were conservative. Is it really believable that between May 2008 and March 2009, with the landslide election of a Democratic tri-fecta in November, the country became 10% more conservative?
What About Pollster's Trend?
The problem here, in a word is Rasmussen. If you look at the actual polls that make up this aggregate, it is overwhelmingly Rasmussen comprising this trend. 16 of the 20 polls since the 2008 election are Ras. I don't have access to their internals to pick at their demographics, but in an election where Democrats won the national House popular vote by 53 to 44, Rasmussen's first post-election poll which started the day after has it only 43D/41R. From a real world PV victory of +9, Ras finds Dems only +2 literally the day after. Call me a little suspicious of their likely voter screen.
Someone else needs to poll this question for greater certainty, but for now this looks like an outlier from NPR combined with Rasmussen's poor screen.
