Home            Back to index

K-State expert says polling here to stay but getting more difficult to do

By Beth Bohn

If national polls found that most Americans would prefer not to be polled about which candidate they favor in the upcoming U.S. presidential election, would that put an end to such polling?

Probably not, according to a Kansas State University expert on public opinion.

Polling has long been a campaign staple — and will probably remain so for a while, said Jim Franke, an associate professor of political science at K-State who studies issues and teaches courses related to America political thought.

"Polling is the second highest expenditure in most presidential campaigns behind television advertising," Franke said.

Polls can help a candidate determine voter awareness and attitudes of their campaigns and what key campaign issues are. News media often use polls to predict election outcomes and attract readers or viewers.

But polling today is meeting with more resistance from the public, Franke said.

"Polls are struggling to some extent today. One of the main problems is with response rates," he said. "Fewer people seem to want to participate. I think the principle issue is cynicism about polling, in part having to do with telemarketing."

Response rates can be key to a poll's legitimacy, Franke said. "Pay attention to the proportion of people who were contacted for the poll and the number of people who actually responded. The lower the poll's response rate, the lower its accuracy rate will likely be," he said.

Most major national polls today are conducted by telephone because at least 95 percent of American homes have a phone, Franke said. Telephone polls also are more cost effective and can provide more timely results than polling people face-to-face.

"During the heyday of telephone polling, in the 1980s and the early 1990s, the polling technology was very solid," he said. "But new call-blocking technologies, such as caller ID, are making it harder for pollsters to reach potential participants."

If we're not hanging up on pollsters, some of us are providing them with false answers to their questions, Franke said. "One reason could be some people think it's nobody business about whom they're going to vote for or how they stand on certain issues."

The accuracy of polls also has been drawn into question by the public, probably because of problems with the close presidential election in 2000, Franke said. Based on exit polling information, the nation's television and cable news networks prematurely declared Al Gore the winner in Florida but then had to take back their declarations when the race proved too close to call.

Exit polling is when pollsters interview voters after they have cast their ballots. Results from exit polls are more accurate than polls taken before an election, Franke said.

"We shouldn't be hoodwinked in what happened in 2000," he said. "The only way to predict accurately what happened in Florida would have been to talk to every voter in Florida. The election was that close."

Franke said the blame for the confusion in Florida is better placed with the television networks and cable news networks and their rush to be first on the air to call a race.

As the 2004 presidential campaign heats up, so will the number of polls. Franke said to remember that it's not the intent of poll sponsors, such as USA Today-Gallup, ABC-Washington Post, Zogby and CBS-New York Times, to fabricate results. Most major national polls are conducted using the same scientific methodology, with some variations.

Most national polls use a sampling size of 1,000 to 1,500 people, Franke said. Their margin of error is typically 3 percent to just under 4 percent. The larger the poll's sampling size is, the lower is its margin of error. But Franke cautions that a poll's margin of error should not be considered its accuracy rate. The margin of error actually reflects a range. So if a poll puts a candidate's support at 55 percent with a margin of error of plus 3 or minus 3 percentage points, the predicted range of support is actually 52 percent to 58 percent.

Most national presidential polls also rely on a randomly selected, computer-generated sample. The sample also is typically demographically stratified, Franke said, meaning that the demographics of those polled reflects the demographics of the American adult voting population by gender, age, ethnicity, geographic location and other factors.

Based on current polling, Franke believes there could be several similarities between the last few presidential elections and the 2004 race. "Watch the undecided vote closely," he said. Undecided voters, often called independents or swing voters, usually make up their mind at the last minute. They are typically the youngest voters and aren't as politically active or politically informed, he said.

Independents usually decide late and on whatever basis, Franke said. "For example, in this election it could come down to a pocketbook issue, such as the price of gas, or it could be the bandwagon effect. If undecided voters sense a certain candidate has the momentum, they may hop on board because simply they want to back a winner," he said.

Summer 2004