Presidential Election: The Science of Polls

Hard as it may be to believe, public opinion polling is actually grounded in science. Sure, vagaries of statistical analysis create meaningful — and exploitable — gaps. But let’s not let demagogues feed our paranoia.

Today is the first day of the next four years of American political life.

It’s been a messy election.

And though we’ll probably have a “winner” by late tonight or early Wednesday morning, what’s now a perpetual partisan contest will continue.

The first battle is likely to be over the vote count in crucial “swing states” — like Florida, for example, where we’ve seen this movie before, and Ohio, where the 2004 election was decided.

Donald Trump has already set the foundation for post-election challenges to the legitimacy of a potential Hillary Clinton win with a series of tweets citing voter fraud in specific locations and questioning the overall legitimacy of a process “fixed” by the mainstream media.

It’s a message that’s resonating with his coalition.

As for “large-scale,” “widespread” electoral fraud, it simply isn’t happening.

As Dale Ho of the American Civil Liberties Union writes in a November 3, 2016, Op-Ed published by The New York Times, “A comprehensive study by Justin Levitt, a senior Justice Department official, found only 31 credible allegations of in-person voter impersonation from 2000–2014, during which over one billion ballots were cast.”

That’s about 0.0000031%. And it ain’t enough to swing a presidential election.

The logistics alone — it’s not a single, national election, but 50 separate state elections — are impossible for a would-be fixer to navigate.

One area of scientific inquiry that hasn’t benefitted from the era of ubiquitous computing is public opinion polling.

In short, the proliferation of mobile phones has made it harder for pollsters to reach enough people to generate a statistically significant sample size.

Eight years ago, more than 80% of U.S. households used landlines. Today, less than 50% do.

In the late 20th century, pollsters could get response rates approaching 40%. Now, they’re connecting less than 10% of the time.

One area of scientific inquiry that hasn’t benefitted from the era of ubiquitous computing is public opinion polling.

Ramin Skibba recently broke down the elements of polling for the journal Nature:

The ingredients of an accurate poll are fairly simple, but they can be hard to find, and everyone uses a different recipe to pull them all together. Start by recruiting a large group of people — preferably more than 1,000. The sample should be split evenly between women and men. And it should reflect the population’s mix in terms of race, education, income, and geographical distribution, to represent these groups’ different views and voting behaviors. Once the data are in hand, pollsters analyze the gaps in their sample and weight the results to account for groups that are underrepresented.

This concept of “sampling” and “weighting” to account for variability of group sizes has already generated some heat from right-leaning observers/s**t-stirrers keen to support Trump’s position in pursuit of anarchic aims/fomenting chaos.

Zero Hedge reported it this way: “New Podesta Email Exposes Playbook for Rigging Polls Through ‘Oversamples.'”

Meanwhile, as the “establishment” Pew Research Center explained it: “Oversampling Is Used to Study Small Groups, Not Bias Poll Results.”

According to Pew:

Oversampling is the practice of selecting respondents so that some groups make up a larger share of the survey sample than they do in the population. Oversampling small groups can be difficult and costly, but it allows polls to shed light on groups that would otherwise be too small to report on.

This might sound like it would make the survey unrepresentative, but pollsters correct this through weighting. With weighting, groups that were oversampled are brought back in line with their actual share of the population — removing the potential for bias.

Zero Hedge quotes extensively from a series of “Podesta” emails from January 2008, including a summary of “Polling & Media Recommendations” compiled by an outfit called The Atlas Project, to conclude this is “how you manufacture a 12-point lead for your chosen candidate and effectively chill the vote of your opposition.”

And it is, indeed, a compelling and convincing case.

It works because it plays on one of the most difficult aspects of the polling process: gathering data from a representative sample that reflects the electorate.

You’re basically trying to predict who’s going to show up and vote. And you’re relying on the word of your respondents when you calculate “oversampling” and “weighting” criteria.

There’s plenty of room for manipulation, too, particularly in service of whatever your “media bias” may be.

At the same time, these folks guard their proprietary approaches closely because that’s how they differentiate themselves and generate new business. It is actually in their best interests to generate accurate results.

But it’s a difficult problem.

Failures of sampling and weighting led to what’s already an infamous episode of polling error: the June 2016 “Brexit” vote.

The 2015 general election in the U.K. was also characterized by widespread polling errors, due in large part to “under-sampling” of Tory voters.

Trump and his coalition point to the Brexit outcome as a sure sign of his continuing viability.

His voters are enthusiastic about overthrowing the establishment, though they may also be a little reticent to identify themselves, like Conservative voters in the United Kingdom were in 2015 and 2016.

This “Tory effect” is compared to the “Bradley effect” from the 1982 California gubernatorial campaign, where the African-American mayor of Los Angeles led his opponent, Republican nominee George Deukmejian, by a couple of points all the way up to Election Day but lost by a narrow margin.

Research later revealed that survey respondents were hesitant to register opposition to Bradley when questioned by a real, live pollster. In the privacy of the voting booth, they could act without regard to social reaction.

We’ll only know late tonight/early Wednesday morning.

Trump will push the issue as long as he can. And should he prevail, you can be sure that Democrats will fight back against “voter suppression” efforts in Florida, Ohio, and other jurisdictions.

Either way, it looks like Election Day is just another battle — a major one, indeed, but still just another set piece in an endless scrum.

We haven’t even touched on the s**t-show to come should Hillary take the oath of office on January 20, 2017.

Ed Kilgore, writing for New York magazine’s online Daily Intelligencer, cites right-wing bellwether Rush Limbaugh’s October 5 commentary:

Hillary Clinton is the most prepared to be impeached in advance of any presidential candidate this country has ever had! Hillary Clinton will be elected to be impeached… By the way, I’m not being glib. If elected, Hillary Clinton could be impeached based on what we already know, and there’s plenty more yet to be discovered. Emails and other documents that could be used in impeachment proceedings are waiting to be found like Easter eggs laid out for 3-year-olds.

It’s not just a talk radio hot topic.

In a recent interview with The Washington Post, Rep. Trent Franks, a Republican from Arizona who heads the Constitution and Civil Justice Subcommittee, said he thought that “all options,” including impeachment, “are definitely on the table.”

Sen. Ron Johnson, facing a tough re-election campaign in Wisconsin, has also said that should Hillary be elected, she should be impeached.

One way that number crunchers mitigate potential errors built into a single pollster’s methodology is to aggregate polls — creating, in effect, polls of polls.

All the major aggregators — including RealClearPolitics, FiveThirtyEight, Princeton Election Consortium, PredictWise, and HuffPost Pollster — have Hillary winning today.

Either way, it looks like Election Day is just another battle — a major one, indeed, but still just another set piece in an endless scrum.

It will mark both an end and another beginning.

This Week In…

Donald Trump is not going to “make American great again.”

Neither is Hillary Clinton the “plus sign” that will make us “stronger together.”

But there is hope. And studies suggest it comes down to us — me and you, black and white, liberal and conservative — getting to know one another a little bit better.

New Scientist weighs in on “how the U.S. can heal its political rift”:

Thankfully, there is a way forward. Studies suggest that when people live with others who share their opinions, their decisions tend to become more extreme, and they tend to become more hostile to outside groups. But reminding people of basic, shared human qualities may help bridge the gap. One experiment reduced participants’ anti-Arab prejudices by making them look at pictures of diverse families doing everyday things, and asking them to consider their own childhood memories versus foreigners’ experiences (Journal of Experimental Social Psychology).

In another study, white and Hispanic students signed up for several weeks of “friendship meetings” with a stranger from the opposite background. Afterwards, students who had previously scored highly for implicit prejudice showed lower levels of the stress hormone cortisol, and were more likely to report seeking contact with people of different ethnic backgrounds in their free time (Journal of Personality and Social Psychology).

Organizations like The Village Square are built on the “contact hypothesis” — the idea that positive interpersonal contact between disagreeing groups is vital to reducing prejudice. A similar body, American Public Square, now operates in Kansas City, Missouri. Its dinners feature live fact-checking of attendees’ claims by staff from a university library, and people are given “civility bells” to ring if they think the conversation is turning too hostile. It’s slow going, but “the process is the product,” says founder Allan Katz, who also co-founded The Village Square. “You’re hoping to raise the consciousness of a community, which also puts pressure on your elected officials and lawmakers to behave civilly.”

In Cambridge, Massachusetts, Jessica Weaver and colleagues at a group called Essential Partners quietly held meetings between pro-choice and anti-abortion community representatives for several years, following shootings at a local abortion clinic. Borrowing from the strategies of family therapists, a major part of their process is to prepare people for this dialogue.

Facilitators interview those involved beforehand, to get a better grasp of their perspectives and goals, and the potential pitfalls to avoid. During the meetings, they try to make participants reflect more carefully on what they say: by scheduling pauses to encourage people to listen and collect their thoughts, for example, or by encouraging questions centered on their feelings and experiences, rather than on their positions. Suggested questions, for example, include “What life experiences may have shaped your current views about abortion?” and “Have you ever felt stereotyped by those who hold different views on this issue?”

Since then, the group has applied the same methods to conversations about gun control in Butte, Montana; immigration in New Hampshire; and religion in Nigeria. Representatives are next headed to North Carolina, where they’ll moderate meetings between the Black Lives Matter movement and the police.

“Is it frustrating? Yes. Is it hard? Of course,” says Weaver. “But I think those conversations across the table are ultimately what gives us a sense of what’s possible at a larger scale.”

The hope is that such methods will hit home, from large communities down to individuals. “If one side thinks more and the other side listens more, then you’d be in a much better place than you are,” says Nick. “Politics is not the hard part of the relationship.”

Smart Investing,

David Dittman
Editorial Director, Wall Street Daily

You May Also Be Interested In:

Stock Defies Gravity in Hated Market

Hard as it may be to believe, public opinion polling is actually grounded in science. Sure, vagaries of statistical analysis create meaningful — and exploitable — gaps. But let’s not let demagogues feed our paranoia. Today is the first day of the next four years of American political life. It’s been a messy election. And...

David Dittman

, Contributing Writer

View More By David Dittman