With yesterday (9-2-10), being two months until the 2010 mid-term elections, here’s some food for thought-
The poll that scares the Dimocrats most
Posted by Moe Lane
Wednesday, September 1st at 2:30PM EDTIt’s this one, from the never-to-be-sufficiently-hated-by-the-Left Rasmussen: and on its face it’s innocuous enough. It’s the partisan identification poll, and it currently lists Democrats at 35%, Republicans at 33.8%, and Neither at 31.1%. Unsurprising, based on recent events, right? – Also, it’s a poll of adults, so this probably means a Republican advantage among likely voters, as that’s the usual rule of thumb for these things. So, nothing really unusual here, right?
Wrong. If this poll is accurate, it’s a harbinger of DOOM for the Democrats.
I don’t pretend to be a professional pollster, but I’ve been dealing with polls on a regular basis since 2003, so I at least know the basics. And I know that – once you get past the pure technical details about whether or not a poll has gotten a true random sample, or whether there’s deliberate bias in the questions – the two major questions that have to be addressed about an election poll both touch on how well it snapshots the actual electorate.
For example: experience shows that a poll that samples 1,000 adults will have a result that is significantly different than one that samples 1,000 likely voters*. The trick is determining what a ‘likely voter’ is, which is why many pollsters at least try to work with the more quantifiable ‘registered voters:’ it doesn’t give you as good results, but it at least screens out the people who can’t vote. It’s also why pollsters try to find out who is enthusiastic about voting, and who isn’t. But that’s only half of the problem; the other half is determining whether or not the current partisan mix of voters has shifted since the last benchmark. That benchmark is usually an election; it’s a truism that, generally, Republicans vote for Republicans and Democrats vote for Democrats. So pollsters look at reliable exit polls, and they look at election results, and every so often they do new partisan identification polls.
And that’s what makes this such a problematic poll of Rasmussen’s for the Democrats. As the pollster noted, historically speaking:
In August 2004, the Democrats had a 2.6 percentage point advantage. In August 2006, they enjoyed a 5.4 percentage point advantage. In August 2008, the gap was 5.7 percentage points. See the History of Party Trends from January 2004 to the present.
…and if you look at the results for those years, you’ll notice that they trended between August and November in all three years towards the party that ended up ‘winning’ those particular election cycles. Which implies that the breakdown is going to be even worse for the Democrats in November. It might even be close to equal.
Why this matters is that a perennial complaint this election cycle is that pollsters keep using partisan breakdowns that assume no major changes between the fundamental makeup of the 2008 electorate and today’s. Yes, pollsters will address the enthusiasm gap – but there is a difference between a politician being down five points because of one party not being motivated to get out the vote and a politician being down five points because there are less members of that party to draw votes from. If Rasmussen is right – and there are a lot of people out there in this business who have a vested professional interest in getting Rasmussen perceived as being wrong – then the problems for the Democratic party will not be addressed in better appeals to their base; they’ll be addressed by changing the policies that are apparently driving voters into the Republican camp**.
And if they don’t, they will simply not be prepared for the psychic shock of Election Night.
