Saturday, May 25, 2013
By John Richardson email@example.com
It's nearly Labor Day and there's just over two months left in the presidential campaign, a high-stakes U.S. Senate race and an intense battle for control of Maine's Legislature.
SOME QUESTIONS OF YOUR OWN TO ASK
THE AMERICAN Association for Public Opinion Research and other groups have published tips for voters to separate the good polls from the bad ones. Here are a few.
• Consider who commissioned the poll and who conducted it. If they are not identified, it should immediately raise questions.
• Consider the sample size – 400 people is often considered a minimum sample size in Maine.
• Look for details about how the poll was conducted and who was interviewed. Random samples and personal interviews over telephones and cell phones are generally considered more reliable than the use of automated phone calls or Internet surveys.
• If the poll data is weighted, or statistically adjusted to match voter demographics, it should say so and explain how.
• Be aware of the margin of error. A margin of error of 4 percentage points at a 95 percent confidence level means that if the poll were repeated 100 times, in 95 cases the results would be within 4 percentage points, either way, of those reported.
• Look at the wording of questions to make sure respondents were not encouraged to answer a certain way.
• For more information, go to AAPOR's website: http://www.aapor.org/ Poll_andamp_Survey_FAQ.htm
In other words, beware the political opinion polls. It is going to be hard to miss them.
"We're getting more and more and more polls," said Paul Lavrakas, a research psychologist and president of the American Association for Public Opinion Research. "I've been studying this since the mid- 1980s, and every election cycle it seems like there's more."
As a pollster himself, Lavrakas doesn't think more polls is necessarily a bad thing. Polls are a way to hear the opinions of actual voters during election seasons that are increasingly dominated by talking heads and partisan bloggers, polling experts say.
But as the co-author of "The Voter's Guide to Opinion Polls," Lavrakas said it is increasingly important for voters to know the difference between a reliable poll and a poll that should be ignored.
Results of a poll on Maine's U.S. Senate race made the rounds on political blogs last week, and served as a reminder about the possible pitfalls of polls in election season.
The poll suggested that King's lead slipped, from 28 points in a June poll commissioned by The Portland Press Herald to 18 points in early August. A higher percentage of voters said they were undecided, while Summers' numbers increased slightly from one poll to the other.
The Summers campaign shared the poll results with bloggers in Maine and in Washington, D.C., who wrote that the race had tightened, citing a new poll but not saying that the Summers campaign had paid for it.
Then the Summers campaign sent out a news release and fundraising appeal, citing the media reports about new poll results. It also neglected to say the campaign had paid for the poll, although it didn't hide the fact when asked.
The episode was a typical example of how poll results are manipulated by political campaigns, Lavrakas said. While the poll may be perfectly valid, a voter equipped with basic knowledge of polls would have known not to take the numbers at face value.
Here is what the experts say voters should know about three major types of polls: independent polls, internal polls and push polls.
Independent polls are the ones conducted by professional polling companies and paid for by groups that are not working for a candidate or a campaign. These are often commissioned by media organizations, such as the Press Herald or ABC News or The Wall Street Journal.
Pollsters consider these to be generally the most reliable polls. But they aren't perfect, and some are better than others.
"When it's done well, it's very scientific," Lavrakas said. But, he said, "when you are studying human behavior, you are never precise the way physical science can be."
Polling science is always evolving. For example, the fact that more voters are switching from landline telephones to cellphones has posed new challenges for pollsters who want to get a truly random sample. Pollsters now typically use both kinds of phones to reach voters.
And some aspects of polling, such as screening for likely voters or asking questions in a way that gets to a voter's true feelings, can be more art than science, experts said. Even the order of questions can skew the results.
(Continued on page 2)