Making sense of all the recent election polls
The battle of the opinion polls is on. As you have most likely read by now, the different polls show widely varying results regarding the gubernatorial race. Some polls make their “snapshot” (at that moment in the campaign) predictions on the answers supplied by those most likely to vote, others interview just plain old registered voters. As you also undoubtedly know, the polls themselves play a role in the outcome. That’s because we Americans love to be on the winning side on anything and that includes elections. We call this the “bandwagon” effect.
If someone is likely to win, two things will happen. People who hate that candidate will stay home, as will some of that candidate’s supporters who think he or she will win without them. Also, the perceived winning candidate will start to get a lot of donations. Remember, it is as American as apple pie to legally bribe the winning candidate so that when he or she wins, they’ll remember you and give you what you need. And the more money in the war chest, the more to spend on advertising and campaigning. In other words, the pollsters themselves are a huge part of who wins and who loses. We do know that in close races, more people usually show up to vote because they think that their vote counts. Since Democrats tend to vote in larger numbers than Republicans, the more people come out to vote, the better it is for the Democrats. If Andrew Cuomo is way out in front, the Republicans may become dispirited and stay home. But, my bet is that Andrew won’t be happy if a pollster shows a close race. That will excite the Republicans and turn them out.
When the unexpected happens and the very respectable Quinnipiac poll shows the supposedly invincible Andrew Cuomo with only a six point lead over Carl Paladino among likely voters, it becomes news because Americans like a horse race. The closer the race is, the greater the chance that people will get interested and become “likely” voters. That’s why organizations like the equally prestigious Siena College poll wait for a while before they begin to poll for “likely” voters. They just ask everyone who is enrolled. Here you have to remember that only about half the people who are eligible to vote actually turn out.
So how do you find out which people really are likely to vote? Well, you ask them, you find out if they have regularly voted in the past and like that. But what if they lie to you? People do lie to their pollsters, we see it all the time. Most newspapers have branded Paladino a strange (as in “whack job”) dude. So the conventional wisdom is that you should vote for Andrew. But if you are really angry, you don’t tell your mom and you don’t tell your pollster. You just go into the voting booth and pull down the Paladino lever. That’s exactly what seems to have happened before the Lazio-Paladino primary. They were thought to be neck and neck but Paladino whipped Lazio, big time.
So, that’s why polling is both a science and an art. The science part is that you make the calls, ask the questions, and add up the figures. The art part is creating a hypothesis of who will vote. There are a lot of possible variables here. We know that people are really ticked off about jobs, the economy, the Congress, yada, yada, yada. So are those people likely to turn out and vote or are they more likely to stay home because they are angry? That’s what each pollster will have to decide, sooner or later. I really respect the pollsters who don’t chicken out and who take chances. That’s what Quinnipiac has done. On the other hand, people are just deciding now what they want to do. Maybe the Siena people are right to wait just a little longer but clearly their using people as subjects who will not vote at all is hardly persuasive.
Then there is the Marist way — do both. Ask everyone who can vote what they are thinking and do a second column in which you count only the people who you think will vote. That way we get both groups. So now you’ve seen the way it works and when you read the next poll, you’ll be able to explain it to the first person you meet at a cocktail party who has no clue.
Originally published in the Legislative Gazette, 9/27/10