deadlines amuse me

exploring how the world works and why it works that way …

Presidential preference polls: how media create a fake horse race

leave a comment »

You can smell that foul odor wafting through the air — presidential politics. Wannabees who won’t say they wannabee are peddling books. Sharply dressed and coiffed “I haven’t decided yet” politicians descend on Iowa and New Hampshire. Explorations of exploratory committees are explored. Websites and Facebook fan pages and Twitter accounts multiply like lobbyists at a fundraiser.

And, if it’s the beginning of the presidential campaign season, then it’s the beginning of the presidential polling season as well. Newspapers and broadcast entities partner with polling organizations to tap likely voters’ preferences for candidates. Even though this is early in 2011 and the election is in late 2012, poll respondents are expected to know now whom they’ll pencil onto their ballot.

So the horse race begins. But it’s fixed. All because of one question:

If the election were held today, who would you vote for?

That question about the 2012 presidential race was asked by:

• Quinnipiac University Poll on March 22-28, 2011, of 2,069 registered voters nationwide; 12 percent were unsure of a candidate choice.

• Allstate/National Journal Heartland Monitor Poll on March 4-8, 2011, of 872 registered voters nationwide; 10 percent of respondents were unsure or refused to answer.

• NBC News/Wall Street Journal Poll on Feb. 24-28, 2011, of 1,000 adults nationwide; 5 percent were unsure of a candidate choice.

• Newsweek/Daily Beast Poll on Feb. 12-15, 2011, of 918 likely voters nationwide; 9 percent were unsure of a candidate choice.

• Democracy Corps poll on Feb. 7-9, 2011, of 500 likely voters nationwide; 2 percent were unsure of a candidate choice and 2 percent refused to answer.

• FOX News Poll on Feb. 7-9, 2011, of 911 registered voters nationwide; 5 percent were unsure of a candidate choice.

• FOX News/Opinion Dynamics Poll on Jan. 18-19, 2011, of 900 registered voters nationwide; 7 percent felt either it was too soon to decide or were unsure of a candidate choice.

• Resurgent Republic poll on Jan. 12-16, 2011, of 1,000 registered voters nationwide; 15 percent were unsure of a candidate choice.

How likely is it that between 85 percent and 96 percent of Americans polled say they have chosen whom they’ll vote for up to 20 months before the election?

They haven’t, says David W. Moore, author of “The Opinion Makers: An Insider Exposes the Truth behind the Polls.” Moore, a former managing editor of the Gallup Poll, offers several reasons why presidential election polls undertaken by press organizations are largely useless.

Point one: The polls above share a flaw: the forced-choice question. Simply put, a respondent’s range of answers to a question is falsely limited. Example: “Do you or do you not support America’s participation in a NATO-led alliance that has established a no-fly zone over Libya?” Either you do or you don’t. The pollster wants a yes or a no. But perhaps you’d prefer more than a no-fly zone. Perhaps you support American participation but object to NATO leadership. Perhaps you feel you have insufficient information to offer an opinion.

In presidential preference polling, you are asked for a voting preference. Never mind that you’ve never heard of some of the names the pollster reads to you. Never mind that you have no idea what policy positions the presumed candidates hold. Never mind that the election is not being held today. Who do you prefer? the pollster insists.

Moore argues that forced-choice questions in presidential preference polling mislead the public. From page 62:

[W]hat the press tends to overlook is that polls do an atrocious job during the campaign, when polling results frequently conflict with one another, and pollsters deliberately suppress important information about the voters. The net result is that media polls often present a highly misleading if not outright false picture of how the candidates are faring and what the voters are thinking.

That’s because pollsters rarely ask one significant question before asking the preference question: Do you have a preference for a candidate for president in the 2012 election? If this question is asked, then it’s highly likely (and Moore cites as an example the dead-wrong front-runner status of Rudy Giuliani in January 2008) that the percentage of “no preference” or “no opinion” or “not sure” will be significantly higher than any preferred candidate. As you can see from the 2012 polls above, the press says that more than three-fourths of respondents have made up their minds already.

Why don’t media polls provide the percentage of respondents who have no preference? That’s because a lack of preference isn’t news. Polling companies that partner with media organizations know this. Polls cost money, so the results need to be newsworthy. Thus press reports that do not reveal the true extent of the disengaged electorate are manufacturing news. Identifying front runners, not an apathetic or ignorant electorate, is news. But this false consciousness about preference placed into the population has consequences. Campaign donors make decisions based on an imperfect political picture. Volunteers choose whom to work for based on misleading information.

Forced-choice polling vastly underestimates the size of the disengaged electorate. Thus the press, trapped in its culture of horse-race reporting, misleads us with many sham polls.

Point two: Presidential preference polls use samples drawn from the national population. The assumption is that a “national electorate” exists and can be used to determine the preference of likely voters. But the presidential election is not a national one; it is predicated on what states candidates win, and thus how many electoral college votes the candidates earn. A poll of a national electorate is meaningless, a “gimmick” as Moore calls it.

So why do media organizations and their polling partners do these “national” polls? They’re easy to do and save money. Writes Moore (p. 148):

The fact that the country doesn’t vote together and all at once is just an inconvenient detail. By taking an inaccurate and misleading macro view, pollsters can avoid the “more assiduous efforts” required for polling in primary states. [emphasis in original]

Polls are not cheap. A few poli-sci profs tell me that the two principal costs in polling are creating (or, more likely, buying) sample lists and actually making the calls. Other cost factors include sampling method and the number of questions asked. Asking 50 questions (which risks respondent fatigue) costs much more than asking five. National poll samples run between 1,000 to 2,000 respondents to get a seemingly legitimizing 3 percent margin of error. Figure $25-$50 per respondent, and that poll may cost between $25,000 and $100,000. Statewide polls might use a smaller sample, say 500 to 1,000, so they would be somewhat cheaper.

But the Republicans plan 31 state primaries for 2012. Obviously, presidential preference polls state-by-state that ask “do you have a preference” and are conducted much closer to the election will likely be far more reflective of the electorates. But the high cost drives media organizations away from the more legitimate state-by-state polling. So the fictitious national electorate created by the press and its polling partners lives on:

Pretending that a national electorate exists is no longer just a harmless gimmick, but a real detriment to the democratic process. [Moore, p. 150]

Point three: Pollsters assume we are smarter or more informed than we actually are. Well, they don’t, actually. They just don’t tell us the percentage of responses for “don’t know” or “don’t care” or “say what?”

That one fact about the collective us — that we are disengaged — is too often absent from poll results. In the case of presidential preference, many if not most likely voters are rationally ignorant. Various media suggest that at least a dozen Republicans are considering running for president. Probably a dozen more are in the wings, watching. At 20 months before the election, the cost in time and money for a likely voter to track each candidate (and his or her various vacillations in policy pronouncements) far outweighs the benefit of that tracking.

Yet poll after poll about presidential preference says at this early date more than three-quarters of Americans have chosen a candidate to support. Few respondents, these polls claim, are unsure; few are apathetic. These claims are patently false, leaving much public opinion about presidential candidates as sheer fiction.

George Bishop in “The Illusion of Public Opinion” adds other confounding factors to preference polling: “(1) widespread public ignorance about public affairs, (2) the inherent vagueness and ambiguity used in most survey questions and (3) the unpredictable influence of variations in question form, wording, and context.” (p. xvii)

Take media polls about anything political with a large grain of salt. Examine how the news stories interpret the results. Find the questions asked of respondents. Find out how much information was provided to respondents — especially if the poll seeks answers to rather arcane policy issues. Identify the pollster: Many work only for Democratic candidates and positions or only for Republicans. Check for bias.

Don’t be duped by spurious, unethical polling by media organizations. They should be reporting news, not creating it through misleading polling.

xpost: Scholars and Rogues

Written by Dr. Denny Wilkins

April 4, 2011 at 9:08 am

Posted in Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: