Wisconsin-sized polling error could mask Romney sweep of battleground states
Mitt Romney would collect at least 72 of the 110 electoral votes available in eight battleground states if President Barack Obama’s current polling numbers, as reported by The Huffington Post, are overstated by a mere one percent.
Romney would win that electoral majority in Colorado, Iowa, Missouri, Ohio, Virginia, North Carolina, Florida and Wisconsin if there is a one percent undercount and if undecided voters there split evenly between Romney and Obama.
That would give the former Massachusetts governor at least 253 Electoral College votes — just a few votes shy of the 270 he will need to claim the White House.
But if the undecided voters break for Romney by two-thirds, Romney would win all those states’ 110 votes, pushing him well above the 270 margin and earning Obama a helicopter ride home to Illinois.
“It’s a good reminder that small shifts in votes can play a big role in electoral votes,” Trey Grayson, director of Harvard’s Institute of Politics, told The Daily Caller.
Grayson predicted that the battleground states are “likely to break together,” resulting in a strong showing for one candidate or the other. “A lot of these states have things in common,” he said, including Midwestern geography and higher than average populations of white voters.
Some Democratic-leaning organizations are concerned that the polls may overstate Obama’s support. “This is going to be a very tough year to poll,” MSNBC host Chris Matthews said Wednesday.
Tuesday’s recall vote in Wisconsin, in which incumbent Gov. Scott Walker prevailed by 6.8 percentage points, was a good indicator of how badly some pre-election polling can perform.
A June 3 poll by the Democratic-aligned Public Policy Polling firm underestimated Walker’s support by nearly 4 points. It showed Walker at 50 percent, only three points ahead of his Democratic challenger, Tom Barrett.
On May 30, Democratic pollster Celinda Lake predicted a dead heat at 49 percent to 49 percent, undercounting Walker’s eventual margin by 6.8 percentage points.
A polling tracker at the left-leaning Talking Points Memo also showed many Democratic-leaning pollsters predicted Walker would be stuck at 49 or 50 percent.
Some of the most accurate polls of Walker’s popularity were offered by We Ask America, a commercial survey firm, which put Walker at 52 percent on May 13, 54 percent on May 23, and 54.1 percent on June 3. But the firm repeatedly undercounted Barrett’s popularity, so the June 3 poll was flawed, showing Walker 12 points ahead of Barrett.
A May 30 Marquette University Law School poll predicted a seven-point win for Walker — 52 percent to 45 percent — but the same poll three weeks earlier showed Walker at 50 percent.
A May 9 poll by Rasmussen, which doesn’t tend to favor Democratic candidates, predicted a 5 point win for Walker.
The various polls do average out to a reasonably accurate margin prediction. For example, the 17 polls Talking Points Memo tracked predicted an average lead for Walker of 5.4 percent. But that’s still 1.4 percent less than Walker’s final margin of victory.
The largest exit poll taken in Wisconsin on June 5 didn’t predict the outcome any better. That poll, conducted by Edison Research for the Associated Press, indicated that Walker would win by a four-point margin.
One problem is that the exit poll could not account for absentee votes — an estimated 10-12 percent of those cast.
University of Virginia political analyst Larry Sabato told TheDC exit polls provide more room for error than just the absentee ballots.
“There’s nothing new about top-line exit poll problems,” Sabato said in an email, explaining that Republicans seem to disproportionately refuse to participate in them, possibly because of their dislike of mainstream media outlets that sponsor the surveys.
Sabato also believes comparing scientific polling to exit polls is like “comparing apples to oranges” because exit polls cannot be weighted to correct any over- or under-sampling of ethnic groups, age groups or either gender.
But he cautioned that “shifting a point to Romney for all swing states isn’t justified” if it’s done only in response to exit poll data.
Some Democratic strategists, however, have used the same Associated Press exit poll data to boast that Wisconsin voters supported Obama over Romney by a 51-44 percent margin.