By Doug Heye, Thomas Jefferson Street blog
Seeing Andrew Breitbart tweet a Wall Street Journal opinion piece by two Democratic pollsters should be enough to pique anyone's interest. It certainly piqued mine, so I read (and recommend) Pat Caddell and Doug Schoen's piece "Don't Shoot the Pollster." Citing a complete dismissal by White House press secretary Robert Gibbs of a Gallup poll last month showing President Barack Obama's approval rating falling to 47 percent, Caddell and Schoen decry the recent trend of attacking the integrity of any pollster whose survey presents data one political side or the other disagrees with. They argue, as the piece's subhead says, "Attacks on Scott Rasmussen and Fox News show a disturbing attitude toward dissent."
Polling is both a science and an art, they write; a science because of the sampling techniques used, an art because of the construction of that sample. As they remind us, "The possibility of manipulation--or indeed, intimidation--is great."
The Democratic attacks on Scott Rasmussen's surveys cannot be because of qualms over accuracy. In the past three election cycles--2008, 2006, 2004--the Rasmussen Reports have hit the nail on the head. In two of those cycles, Republicans took it on the chin.
No, the Democratic attacks are because, Caddell and Schoen write, Rasmussen is "the leader in chronicling the decline in the public's support for President Obama." Rasmussen is certainly not alone. One would be hard pressed to find polling that does not show Obama's public support, especially among independents, in a free-fall, nor do the two Democratic pollsters suggest the decline is "perceived" or "alleged."
And because Scott Rasmussen regularly appears on the Fox News Channel, including The O'Reilly Factor and Hannity, his polling results give the far-left further heartburn.
Using only likely voters has helped Rasmussen stay ahead of the curve on Obama's dwindling support. Likely voter surveys tend to be more accurate (and also more Republican) in nature than those surveying registered voters or adults 18 and over. Polling in the Massachusetts Senate race shows this clearly; those polls with Democrat Martha Coakley holding a double-digit lead tend to be of registered voters, while polling that shows Coakley neck and neck with her opponent, Scott Brown, comes from surveys of likely voters—a much more reliable sample in a special election, where the turnout is historically low.
There's no question polling data should be scrutinized. Whether or not, for instance, a poll used robo-calls (the results of which news outlets generally would not touch until recently) or if a certain demographic was under-sampled or over-sampled is important and can be easily manipulated. But when poll after poll shows the same result, instead of shooting the messenger it would be wiser to improve the message.