Public polling is supposed to predict and reflect public opinion, not drive it. But by the looks of two consecutive national surveys, it seems as though certain media organizations are far more interested in achieving the latter end than the former. Yesterday, the Washington Post and ABC News published a poll purporting to show the presidential race tied at 47 percent. As Dan mentioned, the poll 's partisan sample was a D+9, with a D/R/I of 33/24/36. This is preposterous. That would mean that this fall's electorate will be two points more Democratic than the Democrat wave year of 2008. As a point of reference, the 2010 midterms showed Democrats and Republicans represented exactly evenly. In spite of this terrible sample (for which WaPo polls are becoming infamous), the race is all knotted up. One crucial note from the internals: Romney is beating Obama among independents by 14 points. Let's be frank -- if Mitt Romney wins indies by anything close to 14 points in November, Barack Obama will be a one term president. It's that simple. But one risible poll wasn't quite enough for this week apparently; Reuters has gotten in on the action as well. Their samples have been notoriously bad all cycle, too, and this latest survey is no exception. Like its WaPo counterpart, Reuters' polling outfit concocted am identical, ludicrous partisan sample of D+9 (among adults, D+5 among registered voters). The new poll's findings?
(1) Barack Obama "leads" Mitt Romney by 6 points overall, with independents woefully under-represented.
(2) Obama's job approval rating is 48/47, a +1 result in a D+9 sample. That's bad news for the president.
(3) Worse news for The One: Of the few independents sampled, only 41 percent approve of his job performance. As Jay Cost notes, that's worse than Dukakis territory.
(4) The president is 10 points underwater on the economy -- again, with a sample skew that should really goose his numbers on every question.
(5) Almost inexplicably, Republicans hold a four-point advantage on the survey's generic Congressional ballot. I repeat, D+9 sample.
On item (3), I'll direct you back to my point about the WaPo survey: If Obama drops that group by a somewhat healthy margin -- let alone 41-59 -- he loses. Period. Incidentally, I interviewed Cost on Hugh Hewitt's radio show on Monday, and he pointed out that historically, incumbent presidents' final job approval ratings very closely portend the percentage of the popular vote they attract on election day. He says that especially in the summer months prior to a general election (when many eventual voters still aren't paying attention), watching that statistic is more useful than obsessing over head-to-head tracking polls. To that end, Gallup's presidential approval poll showed Obama dipping back to 44 approve / 47 disapprove yesterday. For what it's worth, his five-point bounce from last week has been completely erased, with O and Romney tied at 46-46.