At the F&M Poll, a review of 2016 to sharpen future surveys

Any reporter covering a poll that asks questions about President Trump knows what to expect from readers who reject anything not fitting their partisan world-view.

"Lib-tard," we get called.

"Democratic lackey" is also likely.

And, of course, "fake news!"

As occupational hazards go, not so bad.  And, to be fair, some of this we inflicted upon ourselves.

We published on Nov. 1 the results of a Franklin and Marshall College Poll that showed Democratic nominee Hillary Clinton leading Trump by 11 percentage points among likely voters in Pennsylvania.

She lost the general election seven days later by 0.73 percent.

Politicians often say "polls are just snapshots in time" when they don't like the results.

For the final F&M Poll of 2016, that picture developed too soon.

F&M Poll director G. Terry Madonna now says that survey, which asked its last questions of respondents on Oct. 30, missed significant developments in the final week of the race.

Madonna, head of the college's Center for Politics and Public Affairs, said his colleagues at the college's Center for Opinion Research did post-election exit polling for more than 3,000 people who had been polled from July to October.

Madonna said they found significant shifts in undecided voters swinging toward Trump in the final week, along with voters who had been leaning toward Clinton changing their minds.

The final poll also did not capture much of the public furor that erupted in the closing days of the election when FBI Director James Comey sent congressional leaders an Oct. 28 letter saying he had revived an investigation into Clinton's use of a private email server.

And Madonna says the poll did not register the intensity of Trump support in the state's small towns and rural areas or slippage in Clinton support in urban areas such as Philadelphia.

In short, he owns it. And like any pro -- he's been polling in Pennsylvania for 25 years -- Madonna is trying to do better. Like any good pollster, he still has questions.

"Is this an atypical election?" he asked about 2016 last week after his latest poll came out. "Will this ever be repeated? How much do we have to change our methodology?"

This is certainly not the first election in which poll skeptics were heard loud and clear.

Former U.S. Sen. Rick Santorum threw a fit on Fox News Sunday in April 2012 after an F&M poll showed his lead had evaporated in the Pennsylvania Republican presidential primary.

Santorum accused Madonna of being a "Democratic hack" who "has probably singularly gotten more polls wrong than any other person I know in the history of the state."

Speaking of history, Santorum dropped out of the primary race 10 days later.

I checked on F&M's track record for polling in Santorum' previous races, for the U.S. House in 2004 and the U.S. Senate in 2000 and 2006. The poll accurately predicted Santorum's 2004 and 2000 victories and his 2006 defeat. The 2012 numbers were spot-on, too.

The was plenty of talk of "skewed polls" in the 2012 general election pitting then-President Barack Obama against former Mass. Gov. Mitt Romney.

So how did F&M do in measuring that race?

A week before the election, the F&M Poll said Obama held a 4 percentage point lead over Romney with 5 percent still undecided. Obama won Pennsylvania a week later with 5.4 percent margin.