In former times, people would have to wait up all night to find out who won a close election. Even the candidates themselves would sometimes go to sleep in the early hours of the morning not knowing whether they would be awakened to the words, “Good morning, Mr. President,” or not.
In the never-ending search for instant gratification, exit polls were created. These samplings, based on interviews with voters right after they exited the polling places, were used to furnish instant projections. The hours of nail-biting over the final results were to be reduced to mere seconds after the official closing of the polls. It was all over but the counting.
But the counting, it turned out all too often, did not confirm the exit polls. This happened in 2000, when exit polls showed Democratic presidential nominee Al Gore winning states like Alabama and Georgia (he lost them). In 2004, Democratic nominee John Kerry did better in exit polls than in the real thing (he lost). In 2008, exit polls “overstated” Barack Obama’s performance by an average of about 7 points (he still won).
In Israel last April, one of the country’s most respected pollsters, Mina Tzemach, suffered the embarrassment of publishing the results of her exit poll giving the Blue and White party a victory of four Knesset seats over Israeli Prime Minister Binyamin Netanyahu’s Likud. On the strength of Tzemach’s prediction, Blue and White chairman Benny Gantz made a victory speech — which he was forced to retract a few hours later after the actual votes were counted.
Initially, there was hope that exit polls would be more reliable predictors than pre-election polls. Among other things, they have the advantage of eliminating such variables as voters changing their minds before casting their votes, if they even make it to the polling place when the big day comes.
A leading American pollster, Nate Silver, has listed several reasons why exit polls are actually less reliable than conventional pre-election polls. Among them: They use a smaller sample of respondents, Democrats tend to be more willing to participate in polls than Republicans, and then there is the “social desirability bias,” also known as “the Bradley effect.”
It was named after Los Angeles Mayor Tom Bradley, an African-American who lost the 1982 California gubernatorial election despite leading in pre-election polls. It seems that many voters were reluctant to say they were voting for the white candidate, because being pro-Bradley was more politically correct. Hence, the misleading findings.
A similar thing may have happened in 2016, when a large number of people who voted for Donald Trump apparently told pollsters that they would not, for the same reason. They didn’t want to reveal to a stranger that they were for Trump.
In other words, people will lie about their choices, and polls assuming they told the truth will give out skewed results.
In the April debacle, Tzemach blamed it on Likud voters. “We had 60 exit polls in our survey, and in all of them the predictions were accurate for Labor, Shas, UTJ, the New Right. There was only one problem: Too many votes for Blue and White and a lack of votes for the Likud. When we saw that, we became angry. Likud voters simply lied to us, putting a Blue and White voting slip into the sample box,” she said.
So this week she announced that she will not be conducting any election polls, thereby avoiding the risk of another such embarrassment.
Yet, other Israeli pollsters were not as far off the mark as she was, and she was the only one to scapegoat the voters.
In fact, polling firms are well aware that people do sometimes lie, and pollsters do their best to take that into account. Figured into the margin of error, routinely published along with the confidently predicted outcomes, should be a margin of lying, as well.
Tzemach admitted her ineptitude. “It seems everyone knew about the trickery except for us. We acted so stupidly. It was our fault that we did not prepare for tricks. The whole time I said, ‘They’re not going to trick us, no one tricks.’ I ran away from it psychologically. That was my issue… a real failure.”
Nate Silver, in his blog, noted that an independent panel set up by CNN after the 2000 election recommended that the network “completely ignore exit polls when calling particular states. I suggest that you do the same,” he wrote to the citizenry at large.
But given the insatiable hunger for political news — real, fake or anywhere in between — it’s not realistic to think that people would voluntarily abstain from exit polls or any other polls.
Given the big business that it is, one cannot expect the traffickers in polls to voluntarily end the practice.
So far, no one has followed Mina Tzemach’s example, and they probably won’t. The average citizen must decide for himself that the polls may be just a waste of time.