OPINION

BRENDA LOOPER: Poll position

The survey says ...

Brenda Looper
Brenda Looper

Last week was not a particularly good one for the president, at least judging by his Twitter feed. Not only was the impeachment inquiry proceeding apace, public opinion polls were showing increasing support for impeachment. To add insult to injury, one of those polls, from Fox News, showed that a majority (51 percent) now support impeachment and removal.

That prompted the president to tweet: "From the day I announced I was running for President, I have NEVER had a good @FoxNews Poll. Whoever their Pollster is, they suck. But @FoxNews is also much different than it used to be in the good old days. ..."

The Washington Post's Philip Bump wrote in response: "It's not true that Trump has never had a good poll from Fox's pollsters; in January 2016, he celebrated a Fox poll that showed him leading in the Iowa caucuses. Nor is it the case that Fox's poll team generally mirrors the network's coverage. The pollsters represent the sort of objective analysis that the network's pundits only claim to espouse."

Not that it matters to some, as I saw several replies to the president's tweet and to the Fox News story on the poll that essentially encapsulated what many people say when faced with a poll result that doesn't square with their worldview: Well, nobody polled me or my neighbors or friends, so that poll doesn't represent me.

Au contraire, mon frère. And shame on you for making me flash back to my college statistical analysis and research methods courses.

Pioneering pollster George Gallup reportedly once said the odds of a specific person being polled are about the same as the likelihood of being struck by lightning.

On the polling firm's site, Gallup says (emphasis Gallup's) it "accurately represents the views of the U.S. adult population because it has a system in place to randomly select survey participants. The precise methods Gallup uses are designed to give every household in America an equal chance of being selected for every survey. This ensures that Gallup has a random cross-section of the population that represents adults of every gender, race, religion, region, party affiliation, etc., in the country and in the correct proportions.

"Essentially, even though Gallup may not include you in the survey, your views are represented by other Americans sharing your views who are included."

Patrick Murray, director of the Monmouth University Polling Institute, one of the most accurate pollsters, told The New York Times' Maggie Astor: "One of the first things that you learn in probability statistics is that the accuracy of your sample is based on the size of the sample itself and not the proportion of the population that it represents."

As Astor explained the process, "Imagine you wanted to know how many blades of grass were in your lawn. If you divided the lawn into 10,000 equally sized squares and counted the blades in 500 of those squares, you could extrapolate that number to the whole lawn. If you made a million squares, you could still extrapolate by picking 500 of them. The total number of squares on the lawn doesn't matter; what matters is that you randomly choose 500 of them to sample."

Sample size can make a difference at lower levels, but as samples grow larger, size matters little as long as the samples are random. Wrote Astor: "You can't conduct a reliable poll with volunteers, because the people who take the initiative to volunteer won't be representative of the full electorate." That's one of the big problems with a lot of online polls, especially those that have no way to prevent people from participating multiple times.

When a segment of the population is over-represented, like say, college-educated women, pollsters weight that segment to more accurately reflect the percentage of the voting population. In more reliable polls, the adjustments are typically minor because the algorithm (not Al Gore, calm down) used to build the samples is already close to the actual percentage.

Still, I hear you say (those of you who aren't snoring by now; thank you for that): "I don't care if there are people like me in the samples! Why should I trust polls?"

Well, that would be because the high-quality ones are pretty darn accurate. Remember, the national polls in 2016 came extremely close to the actual vote percentage, well within the margin of error. (An aside here: If poll results don't tell you--at minimum--the margin or error, sample size, methodology and time frame, ignore them. The best ones include all that and the survey questions as asked.)

When multiple polls are consistently close in numbers, that means they fairly closely reflect reality. I tend to watch the workups done by outfits like FiveThirtyEight and RealClearPolitics, which account not only for individual polls (including the outliers), but take an average of those polls, which is generally reliable and anything but fake news. And that average (FiveThirtyEight's rolling average as of Monday was 50.3 percent support) backs up the Fox News poll that upset the president.

But yeah, if you don't agree, fake news. Obviously.

------------v------------

Assistant Editor Brenda Looper is editor of the Voices page. Read her blog at blooper0223.wordpress.com. Email her at [email protected].

Editorial on 10/16/2019

Upcoming Events