When polls are weapons

When polls are weapons
By Emma Dumain

(Note: What this says folks is that they can get opinion polls to say anything they want, if they ask the questions in a way that slants the opinion in their direction. So opinion polls are about as useful in telling how people really feel as gasoline to put out a fire. Dave)

http://www.congress.org/news/2010/04/22/when_polls_are_weapons
For a price, political candidates, interest groups, and activists alike can get their hands on a powerful and far-reaching tool to spread their message: a public opinion poll.

Polls were once so costly and labor-intensive that only well-financed companies could afford to conduct them. But today, anyone with a couple hundred dollars can hire a polling company to field questions to a sampling of strangers.

Organizations can instruct pollsters to ask specific questions, phrased in a way that will yield the responses they want. In cases where the results are not as revealing, these organizations can still control their message by publicizing their interpretations in releases or blog posts.

New technology has also made polls faster to produce, cheaper to fund, and easier to circulate.

At the same time, the 24-hour news cycle and online media presence have made it harder for people to distinguish between "good" and "bad" polls, and to discern when polling data is being manipulated for political gain.

"People have been releasing surveys in their interest for decades," said Professor Jon Krosnick of Cornell University. "Unfortunately, there are no policing mechanisms in place to help readers and viewers and listeners distinguish between the different types of polls that are out there today."

Polling used to be an expensive and time-consuming operation. In the 1930s and '40s, pollsters had to be flown around the country to conduct extensive surveys in person.

In the 1960s, as more households became outfitted with telephones, polling companies switched gears: they realized they could save money incurred in travel expenses by hiring more employees to conduct more polls, by phone, in home-based headquarters.

Developments in computer technology in the 1980s set the tone for decades to come. Interviewers could record and calculate their data on personal computers, rather than work off of paper questionnaires. Findings could be collated more quickly.

Many polling firms found new ways to cut costs years later by employing interactive voice response technology: rather than paying people to conduct polls over the phone, a pre-recorded voice instructs participants to dial the corresponding digits when responding to a series of "yes" and "no" questions. Surveys conducted entirely over the internet became popular as well.

The faster, cheaper, public opinion poll was born.

Though polling for partisan purposes—for "message"—has always been around, the volume of these types of polls has increased in recent years, thanks to new technology driving down production costs.

This is positive for polling companies, and for the groups that use polls and the publicity they generate to their advantage.

The liberal news blog Daily Kos relies on polling data to bolster its agenda. It uses a nonpartisan polling firm, Research 2000, to conduct polls to complement its analyses.

Though most polls on the Daily Kos focus on how candidates are stacking up in the hot races of the day, Research 2000 pollsters often ask of Republican respondents if they believe Barack Obama was born in the United States.

When the numbers are on its side, the Daily Kos uses them to criticize Republicans.

Activists who lack a Daily Kos-like platform rely partly on the 24-hour online news cycle to see their agendas publicized. With news websites and blogs working constantly to captivate readers with new content, new poll numbers gain visibility almost instantly.

Public Policy Polling, a Democratic public opinion polling firm, frequently makes headlines with its polls asking questions yielding answers that might make Republicans look foolish.

To advance the Democratic message, all Public Policy Polling has to do is release the numbers. The press, learning that the Republican respondents in a particular poll believe that ACORN "stole" the election for Barack Obama, for example, will spread the word.

Though activists and their pollsters are riding the wave of these new developments, experts are skeptical that they are positive, in the long run, for consumers.

For one thing, the great number of polls released each day makes it difficult to tell which rely on sound or shoddy methodology. Pollsters are not always open about how they conduct their polls, and likewise, many news outlets do not bother to ask.

"With bloggers, there's no reason to believe they have more sophisticated understanding of methodology or ability to discriminate between 'good' and 'bad' data than anybody else," said Michael Traugott, a politics professor at the University of Michigan. "And a write-up on the results is usually very thin on analysis, without explanations for why people might give that response."

Another problem is that the groups that commission these sorts of polls are often themselves unconcerned about the methodology insofar as the results help advance their agendas.

It's no secret that IVR polling presents the problem that, with the elimination of a human voice on one end of the line, there is little way of telling if the respondent on the other end is taking the poll seriously, or is who he or she claims to be. And many computer surveys violate a core principle of solid public opinion polling: instead of polling companies choosing participants at random, it is the participants who choose whether to take a survey, often in exchange for prizes.

"I'm just shocked that people are happy to buy [bad] data," said Krosnick. "They say, 'I can't afford to do real science, I have to do snake oil science' … and for polling firms, even if you use crummy methodology, as long as people are willing to publicize your poll, you will succeed."

Polling experts are also concerned about how questions are phrased, and what it means that the results are being used, more and more frequently, for political gain, rather than as an honest assessment of public sentiment. The pollsters, however, are quick to defend themselves. They say that they should not be held accountable for how their clients use the findings they collected.

"I don't do any analysis," said Del Ali, president of Research 2000, the firm that conducts polls for the Daily Kos. "The client pays for the data, and we're not standing over their shoulders saying, 'Hey, this is what you're gonna write.'"

Tom Jensen of Public Policy Polling also disputes the claim that being a partisan pollster will temper the results of a poll.

"People certainly accuse us of being biased whenever we put out results that they don't like, but that's why our track record is important…we have a good one," Jensen said. "Most of what we've been putting out right now is bad news for Democrats, and more Democrats are getting angry with us than Republicans because of that. But we have to put it out the way we see it, whether we like what we see or not."

Emma Dumain covers health care for Congress.org.