As U.S. voters face 18 months of constant bombardment from pollsters, it’s time to revisit the serious problem with political prognostication.
Polls have become click-bait — and erroneous click-bait at that. Here’s how bad it has become:
• In the British election, the polls said the race between the Conservative Party and the Labor Party was too close to call. In fact, the Conservatives won a majority of seats in Parliament and defeated Labor by more than 6 percentage points.
• In the Israeli election, the average of pre-election polls showed Prime Minister Benjamin Netanyahu winning only 21 seats in the Knesset, while the primary opposition party, Zionist Union, would get 25 seats. Mr. Netanyahu’s party earned 30 seats, while the Zionist Union won 24.
• In the U.S. midterm elections, the pollsters missed a huge change in the U.S. Senate, including errors in Alaska, Virginia and elsewhere, and mistakes in key gubernatorial selections, such as Illinois and Maryland.
Polls have gotten the vote wrong in past elections. The worst example of a botched result occurred in 1936 when Literary Digest asked its readers about the presidential race between former Kansas Gov. Alf Landon and incumbent Franklin Roosevelt. Two million people answered the straw poll, which forecasted that Mr. Landon would win with 57 percent of the vote. A young, rival pollster, George Gallup, questioned 50,000 voters, who said Mr. Roosevelt would sweep to victory by a sizable margin.
In fact, Mr. Landon won only two states, Maine and Vermont, while Mr. Roosevelt received nearly 61 percent of the popular vote.
The magazine poll had a poor demographic mix of people, which skewed toward the upper classes. Similar sampling errors continue.
In recent years, the television networks called the 2000 presidential race for Sen. Al Gore of Tennessee. After a recount and a court battle, President Bush eventually won Florida and the election. The 2012 pre-election polls gave too much strength to GOP hopeful Mitt Romney.
Writing after the 2014 U.S. midterm elections in The Huffington Post, political analyst Jeanne Zaino argued, “The lesson is, we need to take what interested parties on both sides say about the polls with some skepticism.”
Statistical guru Nate Silver mangled his British projections. “The world may have a polling problem,” Mr. Silver told Politico. “In fact, it’s become harder to find an election in which the polls did all that well.” Former Obama adviser David Axelrod, who worked for the Labor Party that got crushed in the United Kingdom, also complained. “In all my years as [a] journalist and strategist, I’ve never seen as stark a failure of polling as in U.K.,” he tweeted.
So what should voters do? I think people should ignore polls because news organizations have created stories of little value based on the results. If people still want to look at the predictions, make sure to read only those of likely voters. Moreover, look at the polls as trends rather than actual results. Make sure to see whether the sample represents the voting public. Review the margin of error, which is usually 2 to 3 percentage points.
The manner in which questions have been phrased also provides a good assessment of whether the poll gives a valid snapshot. The National Council on Public Polling provides some advice. “Does the question seem fair and unbiased? Does it present a balanced set of choices? Would most people be able to answer the question?” the organization states.
Remember that the only poll that really counts happens on Election Day.
• Christopher Harper is a longtime reporter who teaches journalism at Temple University. He can be contacted at firstname.lastname@example.org and followed on Twitter @charper51.