Pollsters Take Political Pulse Of American Voters
Pollsters Take the Political Pulse of American Voters
In 1936, when public opinion polls were in their infancy, the prestigious Literary Digest conducted an ambitious survey using phone books and automobile registrations. Their conclusion: Republican Alf Landon handily would defeat incumbent President Franklin Roosevelt.
Roosevelt, of course, won in a landslide.
That public humiliation caused pollsters to adopt rigorous new standards to increase the accuracy and reliability of their results. Today, public opinion polls have not become infallible, but polling has become ubiquitous in American life, whether measuring attitudes toward presidential candidates or public schools, car insurance or computers.
CANDIDATES AND POLLS
Estimates are that, for the 2008 election cycle, more than 500 state and national polls actively are surveying the American public, whether about the presidential race ("If the election were held today, who would you vote for?"), or about views on issues from Iraq to taxes.
These early poll standings may not identify the next president, but they are an important way of generating interest and money. Moreover, voters use them to begin deciding who seems capable of winning office -- and who cannot escape the marginal realm of single-digit ratings.
California's Field Poll conducted a presidential poll in October, and will run surveys again in December and January. "We want to be able to track the latest movement in public opinion before the primary on February 5," Field Poll director Mark DiCamillo told USINFO November 11.
RDD AND CELL PHONES
The underlying validity of public opinion polling is based on the simple proposition that a genuinely random sampling of a given population will provide an accurate picture of the whole. Simple in concept, but often difficult in execution.
The Literary Digest's downfall, for example, was not faulty random selection, but sampling a relatively affluent population that owned cars and telephones at a time when much of the country had neither.
Today, telephones are the means of choice to conduct timely state or national polls. The typical telephone poll uses a technique called Random Digit Dialing (RDD) in which a computer program generates phone numbers from known area codes and prefixes, a procedure that, theoretically, gives every phone number an equal probability of being selected.
A growing concern, however, is the more than 10 percent of Americans who use only cell phones. RDD only generates landline numbers. So far, studies indicate that the exclusion of cell phone users does not affect the validity of the results, according to the industry's professional organization, the American Association for Public Opinion Research (AAPOR).
National pollsters like Gallup and Zogby may have no choice except RDD, since there is no national voter list to select from. California is different, however, and in 2006, Field Poll changed its "sample frame" from RDD to state voter lists with whatever telephone number the voter includes.
"We use a different script with cell phones," DiCamillo says. "Our first question is, 'How would you like to be contacted,' so they can choose a landline number if they want." Field Poll workers call cell phone numbers only on weekends, he adds, when most users have unlimited minutes.
NEWS MEDIA AND RESPONSE RATES
Another challenge for pollsters is the attitude toward the media in general. Americans have been much more skeptical about the media than in the past -- even with the proliferation of cable television and the Internet, according to the Pew Research Center for the People and the Press. (See related article.)
As a result, the number of people who decline to be interviewed (the "refusal rate") has increased substantially. AAPOR has not found a link between response rates and survey quality, but the trend remains a concern.
DiCamillo says that in the late 1970s, telephone response rates typically ran 70 percent. Today, the response is closer to 30 percent.
The Field Poll relies on being seen as nonpartisan and supported by a wide range of news organizations. Its media subscribers range from the conservative San Diego Union to the liberal San Francisco Chronicle, DiCamillo observed.
The Field Poll words its questions carefully, not only to ensure balance, DiCamillo says, but to keep the language simple and clear. "We even try to keep most words no longer than seven or eight letters if we can," he says with a laugh. "We feel we've earned our credibility."
Any opinion poll is only one snapshot in a dynamic environment, and accurate polling requires not only asking the right questions of the right population, but using time-tested social science methodology as well.
National pollsters, for example, might choose to "weight up" their findings to compensate for those groups that are under-represented in their polls. The Field Poll, which typically interviews 1,000 to 1,200 individuals for each survey, applies certain formulas to ensure that its sample matches California's overall demographics in age, gender, political party and region of the state.
And if the person answers the telephone in Spanish? "We call back with a staff member who is fluent [in Spanish]," says DiCamillo, pointing out that the Field Poll cannot afford to miss the estimated 5 percent of registered voters in California who speak primarily Spanish.