Two polling organizations asked Pennsylvania registered voters an identical question during a similar timeframe this June: “If the election for President were being held today, and the candidates were Hillary Clinton the Democrat and Donald Trump the Republican, for whom would you vote.”
That’s about the only thing they had in common.
Quinnipiac’s poll came back with 42 percent of respondents saying they’d vote for Clinton and 41 percent Trump. The other, from Evolving Strategies and Ballotpedia, had 49 percent saying they’d vote for Clinton and 35 percent Trump.
So depending on the poll, either Clinton has a comfortable lead or is barely ahead of Trump. Why the wide disparity? And should we be trusting one over the other?
For starters, here’s a breakdown of the two polls, which were both conducted via live interviews:
Quinnipiac
- Total respondents: 950
- Dates surveyed: June 8-19
- Race: 83 percent white, 8 percent black, 4 percent Hispanic, 5 percent other
- Gender: 54 percent women, 46 percent men
- Political affiliation: 33 percent Republican, 37 percent Democrat, 26 percent independent, 4 percent other
- College degree or better: 37 percent
- 18-to-34: 19 percent
- 35-to-49: 24 percent
- 50-to-64: 30 percent
- 65-plus: 22 percent
Ballotpedia
- Total respondents: 601
- Dates surveyed: June 10-22
- Race: 86 percent white, 8 percent black, 4 percent Hispanic, 2 percent other
- Gender: 53 percent women, 47 percent men
- Political affiliation: 37 percent Republican, 42 percent Democrat, 15 percent independent, 6 percent other
- College degree or better: 47 percent
- 18-to-24: 9 percent
- 25-to-34: 12 percent
- 35-to-44: 16 percent
- 45-to-54: 16 percent
- 55-to-64: 23 percent
- 65-and-up: 24 percent
Aside from number of respondents, seemingly the lone major difference between the two polls is the number of likely voters with college degrees. The Quinnipiac poll in which Trump did better featured 10 percent fewer people with at least that level of education and that likely favors Trump. In the primaries, Trump consistently performed well with voters who have lower education levels.
In the 2012 presidential election, according to exit polls, 48 percent of Pennsylvania’s electorate had a bachelor’s degree or higher. That number is almost identical to the respondents from the Ballotpedia poll that had Clinton in a large lead and significantly higher than the Quinnipiac poll. However, with the same ratio of non-college and college supporters for Clinton and Trump, 10 percent more college graduates and 10 percent fewer people with no college education in the poll would have given Clinton a 2 percent edge rather than 1 percent edge. Not much of a difference.
Quinnipiac is generally considered one of the country’s leading polling groups. Nate Silver’s FiveThirtyEight gave this particular Pennsylvania poll an A- rating. He did not assign a rating for the Ballotpedia/Evolving Strategies poll (FiveThirtyEight gives Clinton an 80 percent chance of winning Pennsylvania).
In April, for Quinnipiac’s last primary poll before the Republican primary, Quinnipiac had Trump in the lead with 39 percent, followed by Cruz, 30 percent, then Kasich, 24 percent. Trump went on to win with 57 percent of the vote, with Cruz defeating Kasich, at 22 percent to 19 percent. Ballotpedia didn’t have a poll, and another polling service, Franklin and Marshall, was also way off on Trump’s dominance and even had Kasich ahead of Cruz in its last poll before the Pennsylvania Primary (The New York Times used a March Franklin and Marshall poll as a good example of a bad poll for not weighting by race).
Nate Cohn, a national polling expert who works for the New York Times, tweeted that the Ballotpedia poll should be taken seriously.
In the last few months, with Trump surprising the GOP and Britons voting to leave the European Union, polls have largely been accurate. It’s just that in Pennsylvania, for now, we have polls showing two very different possibilities.
Clinton — or at least a Clinton PAC — appears to be taking Pennsylvania seriously. Priorities USA is spending $10.5 million TV ads in the state that will air from July 5 until the general election.
Trump? It looks like he’s finally starting to care. Local GOP leaders across Pennsylvania said they hadn’t even heard from his campaign. Trump responded by hiring two well-known Republican lobbyists to run his field operation.
