Grading the Pollsters: June 8 edition

Today we have high profile races in South Carolina, Nevada, California, and Arkansas to look back on and see how well the pollsters’ numbers matched the actual results.

As I’m just one person who lacks the luxury of doing this site full time, I’m not ashamed to work off of Real Clear Politics to make these assessments.

First up, the Arkansas Senate runoff between Blanche Lincoln and Bill Halter. Last time around, both Mason Dixon and Research 2000/Daily Kos widely overestimated Lincoln’s margin of victory. But for the runoff M-D didn’t show, and R2k was the only pollster in.

Sadly for them, they projected Halter the winner by 4 points, 49-45, or subtracting the undecideds, 52-48 for Halter. The actual result was the opposite: Lincoln wins 52-48. Off by an 8 point swing and projecting the wrong winner is about as wrong as you can be.

Moving on, we look in at California. Three polls came out election week for the California Republican Senate primary. We ignore the Democrats because Barbara Boxer and Jerry Brown were effectively unopposed. The actual Senate results? Fiorina 56, Campbell 22, DeVore 19.

Field Poll was good and bad, getting Campbell and DeVore exactly right, but blowing all of that by undershooting Fiorina by 19 points. It’s hard to score that as a clear win for Field.

Magellan Strategies I’d say gets a win. At Fiorina 54-Campbell 19-DeVore 16, the margins are almost exactly correct, as are the proportions.

SurveyUSA suffered a milder version of the Field Poll syndrome: Campbell and DeVore close, Fiorina way off. Again, hard to say they got it right when they missed the winner’s margin by a mile.

Now we shift to the Republican race for Governor in California. The final tally had Meg Whitman over Steve Poizner 64-27.

Again Field misses the winner’s margin substantially, understating Whitman’s result by 13 points while only missing Poizner by 2. SurveyUSA had an outright shift toward Poizner: 59-30. Magellan again got closest: Nailing Whitman’s total down but understating Poizner by 5.

Harry Reid is unopposed to my knowledge, so we turn only to the Republican side of the Nevada Senate Primary. actual results: Sharron Angle 40, Sue Lowden 26, Danny Tarkanian 23.

We begin to see a trend. Suffolk University, one of the big winners in the Pennsylvania primary, underestimates the winner’s momentum with Angle 33, Tarkanian 26, Lowden 25. It’s hard to say that’s really close to the actual result. And since both Research 2000 and Mason-Dixon had similar findings, I’ll say everyone lost in this race.

Republicans also had a race for Governor in Nevada, which was won by Brian Sandoval over the incumbent Governor Jim Gibbons 56-27. And again, both Suffolk and Mason-Dixon underestimated the winner by 9 points.

And finally we have the South Carolina Governor’s race. PPP skipped the Democrats in the last poll, so we only have their June look at the Republican side to grade. The actual result was Haley 49, Barrett 22, McMaster 17, Bauer 12. And PPP had Haley 43, Barrett 23, McMaster 16, Bauer 12. That’s pretty impressive I’d say, despite yesterday’s trend of the winners overshooting the polling estimates.

To sum up: Research 2000 botched another primary for the Democrats. Magellan Strategies impressed in California. Public Policy Polling and SurveyUSA were solid. Field missed a golden opportunity. Suffolk and Mason-Dixon faded. Rasmussen Reports didn’t show up.

Are primaries really that hard?