Impact of Opinion Polls: In History and Today
In an election year, it’s impossible to escape public opinion polls. These polls have been tracing the rise and fall of President Obama‘s approval rating and predicting primary results. In the wake of Rick Santorum’s withdrawal from the race, pollsters will soon be matching up Mitt Romney’s chances against Obama in the general election.
Why do we watch polls? Because we want to know what other people think. Polls also allow us to attempt to predict the future. For those holding or seeking elected office, polls have even more significance.
Polls and Money
Political candidates often use poll numbers to help them allocate time and resources. Newt Gingrich’s low poll numbers in Michigan and Arizona caused him to virtually ignore February’s primaries there.
Low poll numbers may lead a campaign to target certain areas for heavy advertising or more appearances by the candidate. However, polls can also cause donations to dry up, making it difficult for a candidate to continue the race.
Signs Along the Campaign Trail
Polls can also help incumbents and candidates identify issues that are important to voters. Such information helps them shape their messages as they travel along the campaign trail.
In Arizona, where polls showed strong support for the state’s tough new immigration law, Romney took a more aggressive stand against illegal immigration than his Republican challengers. That message likely contributed to his solid victory there in February.
The Famous 1936 Poll
One of the most famous polls in political history occurred in the 1936 election year. Hired by a New York advertising firm, journalist George Gallup conducted a poll that correctly predicted the landslide reelection of Franklin D. Roosevelt. The Gallup Poll contradicted the popular Literary Digest straw poll that predicted an overwhelming victory for Roosevelt’s challenger (see side column)
Gallup went on to become a pioneer in the field of scientific public opinion polling.
Today, a number of independent scientific polling organizations exist. Both political parties have their own pollsters, as do many campaign organizations. Survey techniques have become so refined that a well-constructed poll of a few hundred people can accurately express the views of millions of voters, with a very small margin of error.
The All-Important Margin
Margin of error is the maximum expected difference between the result for the people polled—called the “sample”—and what the result would be if the entire population had been polled. The size of the margin of error is related to the size of the sample.
Generally, the larger the sample size, the smaller the margin of error. For example, if 500 people are polled, the margin of error is 4 percent. If 50 percent of the people polled say they support Obama, pollsters could accurately conclude that between 46 and 54 percent of the population supports the president. If 2,000 people are polled, the margin of error is 2 percent.
It’s important to know a poll’s margin of error in order to assess the accuracy of its results. Think carefully if you hear a reporter say that a candidate is ahead in a race, but that the opponent may be within the leader’s margin or error. For example, if 52 percent of those surveyed favor Candidate A and 48 percent favor Candidate B, and the margin of error is 6 percent, Candidate B might actually be ahead!
Keeping It Real
The accuracy of poll results depends on other factors as well. Reputable pollsters must ensure that their sample is randomly chosen and representative of the group being surveyed.
The way questions are worded and the order in which they are asked can also affect the accuracy of poll results. Consider the difference in the wording of these two poll questions:
“Do you think the president is doing a good job?”
“What kind of a job is the president doing—good, fair, or poor?”
For a fun activity on polling, check out this Classroom Simulation “How to Conduct a Poll”
Errors introduced by polling methods are called bias. Bias can sometimes be difficult to avoid. For example, election polls commonly only include “likely” voters in their samples, that is, voters who say they intend to vote. However, polls show that when asked if they are likely to vote, 80 percent of registered voters—good citizens, after all—respond “yes” whether they really intend to or not. Polls that measure only good intentions generally don’t produce the most reliable results!
Fundamentals of Polling (Roper Center)
A Guide to Sample Size and Margin of Error (Public Agenda)