Ross Kaminsky - Gallup prefaced the results of their recent Obama vs. Generic Republican poll results thus: "A generic Republican presidential candidate has a slight but not statistically meaningful edge over Barack Obama, 44% to 39%, in registered voters' preferences for the 2012 presidential election. Historical data from June in the year before an election have not been very predictive of the outcome."
In my view, Gallup's language understates just how bad this news is for President Obama.
Let me stipulate that Gallup's assertion is true that this poll, based on past history, is not highly predictive. (One then has to wonder why they bother with it.)
But to argue that it's not statistically significant seems an excuse too far.
First, President Obama trails the generic Republican by 5 percent while the notes to the poll give it "95% confidence that the maximum margin of sampling error is ±4 percentage points."
Second, the last time the same poll was taken -- a month ago -- Obama held a three point lead over a Republican-to-be-named-later. Thus, the swing is a large 8 percentage points, an amount which can't possibly be statistically insignificant. (And again, if it is, then this poll question should never be asked in this way.)
Finally, it's one thing for an incumbent to trail in polls against a particular opponent. But it's another thing entirely to trail a generic, nameless, faceless candidate. In the two prior election cycles where there was no front-runner at this point in the election cycle, Gallup polled the question of the incumbent versus a generic member of the other party. In both cases, the incumbent had a huge lead -- although George H.W. Bush ended up losing to Bill Clinton. For Barack Obama to trail a generic Republican right now shows far more weakness than Gallup would have us believe.
In my view, Gallup's language understates just how bad this news is for President Obama.
Let me stipulate that Gallup's assertion is true that this poll, based on past history, is not highly predictive. (One then has to wonder why they bother with it.)
But to argue that it's not statistically significant seems an excuse too far.
First, President Obama trails the generic Republican by 5 percent while the notes to the poll give it "95% confidence that the maximum margin of sampling error is ±4 percentage points."
Second, the last time the same poll was taken -- a month ago -- Obama held a three point lead over a Republican-to-be-named-later. Thus, the swing is a large 8 percentage points, an amount which can't possibly be statistically insignificant. (And again, if it is, then this poll question should never be asked in this way.)
Finally, it's one thing for an incumbent to trail in polls against a particular opponent. But it's another thing entirely to trail a generic, nameless, faceless candidate. In the two prior election cycles where there was no front-runner at this point in the election cycle, Gallup polled the question of the incumbent versus a generic member of the other party. In both cases, the incumbent had a huge lead -- although George H.W. Bush ended up losing to Bill Clinton. For Barack Obama to trail a generic Republican right now shows far more weakness than Gallup would have us believe.
No comments:
Post a Comment