r960-b4a4856b27b17d7dc1d6f6da1a7c919e.jpg


The failure to recognize Donald Trump’s chances of being elected president wasn’t the fault of bad polling data, but of bad analysis.

A review of national and battleground polls heading into Trump’s stunning victory over Hillary Clinton on Tuesday revealed red flags for the Democratic nominee, and signs that her Republican opponent was positioned to win.

In Florida, the final RealClearPolitics average showed Clinton and Trump in a statistical dead heat. In North Carolina, the final RCP average showed Trump ahead by 1 percentage point; and in Pennsylvania Clinton only led by 1.9 points.

Yet the consensus analysis was that Clinton would finish on top in at least two out of the three, either because she led nationally or because her voter turnout operation was judged (and probably was) superior to Trump’s.

Trump ultimately won all three states on his way to a solid Electoral College victory. Meanwhile, Clinton won the national popular vote — yet another example of the polls calling the race correctly.

Sean Trende, a polling analyst for RealClearPolitics, said during an interview with “Examining Politics,” the weekly podcast from the Washington Examiner, that Pennsylvania is a good example of misinterpreting what the polls were actually saying.

“We had a two point race, which I think most people intuit is a race that can go either way, but people convinced themselves — no, no, a 2 point lead in the polls is for some reason solid,” Trende explained.

“I really do believe in my heart of hearts that a lot of this is just wish-casting by people,” he added. “It’s easy to blame the polls … There were forecasters who were saying this was a 99 percent done deal for Clinton and I don’t think that was a realistic reading of the polls.”

In the immediate aftermath of Trump’s stunning win, political analysts and some professional strategists lauded the president-elect’s intuitive approach that shunned polling and consultant driven campaigning.

McConnell sees Pence as White House liaison

Also from the Washington Examiner

Senate Majority Leader Mitch McConnell is throwing out a lifeline to Vice President-elect Mike Pence.

11/10/16 12:01 AM

Trump’s success, they said, was evidence that survey research and data analytics forecasting used by the Clinton campaign and the Republican National Committee failed miserably and should be reconsidered in future elections.

Yet even where Trump’s wins were most surprising — in Michigan and Wisconsin, for instance, the problem might have been a lack of polling data to gain an accurate picture of the state of the race.

Perhaps had more polls been conducted in those states, Clinton’s leads in the final RealClearPolitics average would have been closer than 3.4 percentage points (Michigan) and 6.5 percentage points (Wisconsin.)

The dearth of polling in individual states was something that several media elections analysts complained about.

Robert Cahaly of the Trafalgar Group, a GOP pollster out of Atlanta who called the race correctly, said the problem is not paying enough attention to the margin of error present in each survey. “We never thought he was going to lose Florida,” he said. Cahaly also picked Trump to win Pennsylvania.

Conservative clean energy could be Trump's climate plan

Also from the Washington Examiner

Jay Faison said he is turning his focus to the Trump team to help it figure out where it stands on clean energy.

11/10/16 12:01 AM

Data analytics, the process of compiling up to hundreds of granular pieces of information on individual voters for the purpose of determining if they would vote, and if so for whom and for which reasons, did not fare as well in the election.

The Clinton campaign modeling of the composition of the electorate, generally a more detailed and reliable predictor of voter behavior and the outcome of the race, clearly fell short in states like Michigan, Minnesota and Wisconsin.

Trump won Michigan and Wisconsin and almost won Minnesota. These three states had been reliably blue for decades. The Clinton campaign didn’t pay much attention to them, although they did invest resources in Michigan and Wisconsin in the final weeks of the campaign.

Even the RNC, which handled data analytics and field operations for the Trump campaign, didn’t forecast a Trump victory. Patrick Ruffini, a Republican pollster and analytics consultant said the problem is that the Clinton and Republican turnout models misfired.

“What happens when we have a situation where the electorate is redrawn somehow, where entire groups of people are less motivated to vote than they were in 2012?” Ruffini told “Examining Politics.” Data analytics “has been successful to the extent that [the electorate] looks like electorates in the past.”

Strategists cautioned against over-generalizing the problem and assuming that big data, as it’s referred to, have been over-hyped.

Indeed, data analytics was quite successful Tuesday at modeling the electorate and forecasting the results in several down-ballot Congressional races.

David Winston, a Republican pollster who uses data analytics in his practice, described it as an evolving technology that is dependent on proper tuning and continual refinement.

Winston also said the Clinton campaign and RNC analytics programs might have been stressed by the unique nature of the two nominees.

Clinton and Trump were historically and unusually unpopular for two politicians in their positions.

Their unfavorable ratings were above 50 percent. That factor could have added instability into the model and contributed to the breakdown in the forecast.

“Modeling is like any other process where you’re trying to turn something into a mathematical equation,” Winston said. “You are going to work through that process and sometimes it’s going to work well and sometimes you’re going to learn things.”

Abortion foes dream big with Trump White House victory

Top Story

At the top of pro-life groups’ to-do list is defunding Planned Parenthood.

11/09/16 11:00 PM



Source link

About the Author:

Leave a Reply