A pair of independent polls released Wednesday offer conflicting pictures of the presidential race in Pennsylvania.
An NBC/Wall Street Journal/Marist poll shows Democrat Hillary Clinton with a 9-point lead in Pennsylvania over Republican Donald Trump, 45 to 36 percent.
Yet a Quinnipiac University poll shows Trump surging in battleground states and holding a 2-point lead in Pennsylvania, 43 to 41 percent, within the margin of error.
The Quinnipiac survey shows voters think Clinton is better prepared for the White House, but when asked which candidate is more honest and trustworthy, voters picked Trump by a 15-point margin, 49 to 34 percent.
“You have to figure the cumulative effect of all these investigations about her emails can’t help,” said Quinnipiac assistant poll director Tim Malloy. “There’s a chance that Bill Clinton’s visit to the attorney general, which just looked bad, whether it was bad or not, couldn’t have helped.”
When Libertarian Gary Johnson and Green Party candidate Jill Stein were included in the surveys of Pennsylvania voters, Clinton held an 8-point margin in the NBC/Marist poll. In the Quinnipiac survey, Trump’s lead surged to 6 points.
How do we account for the difference between the polls? The sample dates are a little different, and there is a margin of error for both.
But another big factor is the sample of voters who are surveyed.
The Quinnipiac poll, which showed Trump ahead, was based on a sample that’s 35 percent Democrat, 34 percent Republican.
A million more Democrats are registered in Pennsylvania than Republicans, but the question is always who actually shows up to vote.
CNN exit polls from the 2012 presidential election found 45 percent of Pennsylvania voters said they were Democrats, 35 percent Republicans. If those proportions show up on Election Day this year, the result will likely be a lot better for Clinton than the Quinnipiac poll.
I asked Quinnipiac poll director Douglas Schwartz about his sample, and he said they didn’t adjust their results to reflect a particular partisan mix in the sample.
They interviewed 982 voters, and the Democratic and Republican proportions were what they were.
As for the partisan turnout in the 2012 race, he said you can’t use exit polls from one race to predict turnout in another.
Different pollsters use different methods for this stuff.
The available information on the NBC/Marist/WSJ poll didn’t reveal a partisan breakdown.