Thursday, June 19, 2008

Let Not Your Heart be Troubled – Or Fooled By Polls

I saw an article by The Anchoress (who apparently did not read my discussion of the Quinnipiac poll) regarding the apparent dominance by Democrats in polling. Her theory is that because voting by Democrats was so heavy in the Spring, this means that Democrats will dominate voting in the fall.

Mmmmmmmmmm ... no, that's not really a good description of the situation. Polls are not quite the magic ball they are hyped to be.

For example, take a look back in 2004. Kerry got a lot of excitement among Democrats because he had strong primary support, and a lot of pundits were saying this meant trouble for the fall. Then 62 million voters decided they preferred President Bush for another term, defying the media’s proclaimed scenario. And while I will grant that 2008 is different from 2004, human nature is notoriously hard to change. For this election, that means the following points will continue to matter:

[] Barack Obama, like John Kerry, built a lot of his support through young voters in the primaries. However, even young voters who vote in primaries are relatively unlikely to vote in the general election. Candidates have been trying to change that condition for decades without significant success.

[] John McCain’s core constituency appears to be senior voters, pro-military voters, and moderate Republicans. The first two of these groups have established a strong representation in votes.

[] Hillary Clinton’s 18 million supporters are not completely enthused by Obama’s campaign. While most of them will support Obama, fewer will actively work for his election (as in recruiting new voters and pursuing grass roots operations), and there have already been reports of Hillary supporters working to get McCain elected.

[] Voters’ opinions and mood can change quickly, but their core beliefs are slow to change, and will change only when confronted with strong evidence. The essential differences between Barack Obama and John McCain will resonate in the election results.

[] Every presidential election, there are states which appear early on to be ripe for a candidate to ‘steal’ from the other side’s roster, but in actual fact few states change philosophies, and the general election is seldom a great surprise to anyone familiar with History.

[] The essential strategic difference in mood between Senators Obama and McCain, is that Obama started with strong positive support and almost no negative reaction, but his attractiveness has steadily declined as the election season progressed, while in McCain’s case he began with strong opposition and little support, but his competence and consistency have strengthened his campaign as he progresses. The result of these trends will depend on the speed of their progress relative to the election date, and whether these trends are halted prior to the election.

[] Opinion polls are the product of polling groups, which are often reported as news, but that representation distinctly is not correct. An opinion poll is the result of a collection of interviews using a standardized set of questions and methodology. That poll is weighted to match a desired demographic profile, and any error in demographic assumptions, in party identification, or methodology as relevant to the real population pool will be reflected in erroneous conclusions. It is a salient fact that most opinion polls which attempt to reflect the specific voter results more than one day ahead of the election will be off by more than the statistical standard deviation, and therefore no opinion poll should ever be taken as a predictive indicator of voter intentions.

All in all, it’s June and that makes it the noise of spin, nothing more.

1 comment:

rightwingprof said...

Quinnipac in 04 and 06 was more often than not wildly different from the other polls, for whatever reason, so much so that I don't pay much attention to their polls. I do wish realclearpolitics made all of their old poll compilations accessible so I could document it, though.