Tipsheet

Shhh: Even Skewed WaPo Poll Shows Obama Trailing Romney

Alternate headline, via Jim Geraghty: "Unstoppable Incumbent Now Trails Romney Again."  Indeed, The One is losing to the Republican frontrunner in a Washington Post/ABC News poll published this morning, yet WaPo curiously elides this little factoid in its story about the survey.  Katie hit some of the poll's broad strokes earlier -- Obama's deep underwater on the economy (38/59), the budget deficit (32/63), and gas prices (26/65) -- but the closest the Post's write-up comes to addressing Obama's staggering general election erosion is a brief third-paragraph mention that Obama's is running "about evenly" with his possible opponents:
 

The potential political con­sequences are clear, with the ­rising public disapproval reversing some of the gains the president had made in hypothetical general-election matchups against possible Republican rivals for the White House. Former Massachusetts governor Mitt Romney and former senator Rick Santorum (Pa.) now both run about evenly with Obama. The findings come just five weeks after Obama appeared to be getting a boost from the improving economy.


One has to dig into the survey's data to discover that Obama is behind Romney 49 to 47 percent, and narrowly leads Santorum -- a confirmation of Rasmussen's weekend numbersWaPo's description of the race as "about even" isn't inaccurate; it just omits the hard numbers themselves (these must be sought out at a separate url).  Perhaps they're slightly embarrassed by the yawning canyon separating the results of this poll from their February survey.  One month ago, this same pollster declared that the incumbent had seized a "clear edge" over Romney, leading the former governor by 15 points 11 points (and six among registered voters).*  Today, with major economic indicators virtually unchanged, the race has supposedly swung by 12 points, with the leading Republican bursting into the lead.  As the Post suggests, some of this may be attributable to rising gas prices.  (Geraghty notices that Americans are significantly less inclined to directly blame Obama for this factor, compared to their views on President Bush's culpability in 2006.  Ahem, I wonder what might explain that disparity).  But the more salient cause of this poll-induced whiplash is dreadful methodology.  I blasted this news organization in February for the decision to censor its sample breakdown, which I argued rendered its data useless.  To its credit, WaPo's polling shop has pulled an about-face, and has begun releasing that information again.  Over to you, Ed Morrissey:
 

After hiding the partisan composition numbers for the last two months, today’s poll includes that data from the last three presidential polls.  Let’s take a look at the D/R/I series, keeping in mind that the D/R/I from the 2010 midterm elections was 35/35/30:

  • March: 31/27/36
  • February: 34/23/37
  • January: 32/25/36

The February poll from which the WaPo/ABC series derived its previous Obama approval numbers gave Democrats an 11-point edge and undersampled Republicans by 12 points in relation to their 2010 turnout.  


WaPo's February survey included a D +11 sample weighting (Obama leads by 11 among adults!), a ludicrious spread that has inexplicably contrated to +4 in March (Romney by 2).  As Ed mentions above, partisan voter turnout split evenly between Republicans and Democrats in 2010, at 35-35.  While it's reasonable to assume that Republicans don't have quite the same momentum they enjoyed in 2010, when they rode a red tsunami to seize control of the House, Gallup's recent numbers indicate that GOP-leaning voters still maintain an eight-point enthusiasm advantage heading into November (and that's despite the lackluster Republican primary field, about which the media reminds us at every turn).  So a case can be made that the Post's sample is still unfairly skewed in Democrats' favor, even though a four-point gap is at least tethered to reality.  Another point: Despite surveying national adults (as opposed to registered or likely voters, who generally skew slightly more Republican), WaPo finds Obama's overall job approval upside down, at 46/50 -- not to mention 50 percent "strong" disapproval on the economy.

Can you imagine what kind of results the Post could produce if they only polled likely voters and oversampled Republicans by 11 points?  Romney by 20!  Santorum by 15!  Palin tied!  Never gonna happen, though, because that would be unthinkable malpractice.  Hopefully the Washington Post has been reminded that the objective of polling is to measure public opinion, not influence or drive it.


UPDATE - A new CBS News/New York Times survey shows Rick Santorum holding a four-point national lead over Mitt Romney among Republicans.  This data point breaks from Gallup and Rasmussen, both of which peg Romney ahead by double digits.


UPDATE II - Fifty-nine percent of Americans tell Rasmussen that Obama is more liberal than they are, nearly matching an all-time high from a year ago.  Only one-fourth of the public believes Obama fairly reflects their ideology.  About one-tenth view Obama as too conservative.
 

*UPDATE III - The February Washington Post poll gave Obama an 11-point lead among all adults (March: Romney +1) and a six-point cushion among registered voters (March: Romney +2).  I misread the data and confused Obama's lead on economic issues (15 points in February's survey) with his overall general election lead.  The sample issue still resulted in a 12 point swing among adults. Interestingly, the Post did not ask the head-to-head Romney/Obama question on the economy in their new poll.  This post has been corrected to reflect these changes.