Jerome Armstrong thinks that polling right now is even more favorable to McCain than it looks, because there are too many Democrats in the surveys. As evidence, he points out that McCain is well ahead among independents, and so the only way Obama can still be close is if there are more Democrats in the electorate than there have been in recent elections. The rather obvious rebuttal is that there might very well be more Democrats in the electorate than recent elections, especially since most polls indicate there will be more.
I used to bother looking at the partisan make-up of polls. I don't bother anymore. Four years ago, I learned my lesson, and since that time I have decided to stop trying to be more clever than scientific experiments of public opinion. There is only one scientifically valid way to determine the percentage of self-identified Republicans and self-identified Democrats in the electorate: conduct a public opinion survey measuring the number of self-identified Democrats and the number of self-identified Republicans in the electorate. The percentage of self-identified Republicans and self-identified Democrats in a poll cannot be determined by deductive reasoning before surveys. It has to be conducted through inductive data collection by surveys.
If we want accurate information from polls, we cannot impose our assumptions on top of the data. As such, in order to determine the state of the campaign, rather than looking at internals of individual polls, I now try to look at the broad trend across all polls. National polling trends like the one at Pollster.com will almost always be within 2-3% of the actual margin. That is really about as accurate as I feel the need to be, especially since being accurate to less than 2% on a regular basis might not even be possible.
McCain maintains his lead in the Gallup poll today, apparently stabilizing at a 5% lead for now. However, that is the best poll for McCain, with no other outlet showing him up by more than 2%. Three (New Diageo shows McCain +1%) Two polls show a tie, and two even show Obama ahead by 1% (ARG and ABC-WaPo RVs). Given that these polls are all trying to measure the same reality, right now it is likely that reality is somewhere between the two ends of the spectrum. Obama isn't ahead by 1%, but McCain isn't ahead by 5%. It is probably, as Pollster.com posits, McCain by about 1-2% right now. That is both the mean and the median of the nine national polls conducted entirely since the convention (note: I am not including the poll showing Obama up 5%, since it was conducted partially before Palin and McCain spoke). Given that those nine polls utilized about 14,000 interviews, I find it pretty hard to argue with that data. So I won't.
|
STATE BLOGS
Powered by: SoapBlox