LAWRENCE: Time, technology changing methods of voter polling

By Lea Kahn, Staff Writer
   The polling field has come a long way since the polls of the Literary Digest and the Gallup Poll dueled over the winner of the 1936 presidential election — Alf Landon or Franklin D. Roosevelt.
   The Literary Digest magazine’s straw poll generated 2 million responses, which predicted that Mr. Landon would be the victor. But the American Institute of Public Opinion — the forerunner of the Gallup Poll — queried thousands of households and said Mr. Roosevelt would win.
   What made the difference? For one, the Literary Digest based its sample on its subscription list, the telephone directory and motor vehicle registrations — in short, wealthy people. But the American Institute of Public Opinion surveyed about 60,000 people at random, nationwide.
   The result of the election is history — and that was the end of non-scientific polling, according to Edward Freeland, director of the Survey Research Center at Princeton University and a Lawrence Township resident.
   George Gallup’s polling organization translated mathematical concepts devised by Polish statistician and mathematician Jerzy Neyman, who taught at the University of California at Berkeley, into precise estimates based on smaller samples of potential respondents who were chosen at random, Mr. Freeland said in an interview at his Princeton University office.
   ”This was like standing in front of a freight train and not seeing it coming at you,” Mr. Freeland said, comparing the magazine’s polling methodology with that of the American Institute of Public Opinion. For a survey to be accurate, all people must have at least a chance to be chosen to participate in it, he said.
   Following the 1936 election, pollsters became reliant on telephoning likely voters at random, Mr. Freeland said. It was possible to reach nearly all of the American population because of the proliferation of telephones, he said.
   ”For 20 or 30 years, you could reach people with telephones. It was the golden age of the telephone interview. You had access to 97 percent of the population. Random dialing (meant) great coverage. But we will never have that again,” Mr. Freeland said.
   Fast forward to 2012, and technology presents a different set of problems, he said. Now, close to one-third of Americans use only a telephone, while another one-third of the population uses only cell phones. About 16 percent of Americans use a combination of telephones and cell phones, but mostly rely on their cell phones and rarely answer their telephone, he said.
   To compensate for the increased use of cell phones, pollsters such as the Rasmussen Reports reach out to respondents on the Internet on a random basis.
   Early voting in the presidential election presents “another challenge,” Mr. Freeland said. Now, pollsters have to ask not only whether the respondent is a likely voter, but whether that person has in fact already voted by mail-in ballot — and it will become more challenging for pollsters as Election Day gets closer, he said.
   Mr. Freeland said pollsters and candidates also pay more attention to statewide poll results than national poll results in presidential elections, because the popular vote does not determine the winner. It is the Electoral College votes cast by the states that count, and that’s why candidates need to track the states — and electoral college votes — that they want to win.
   Meanwhile, there is some skepticism of polls, possibly as a result of the 2000 presidential election, Mr. Freeland said. The two presidential candidates — George W. Bush and Al Gore — were running virtually neck-and-neck. When the polling results are about 50/50, “it is almost impossible to predict the winner,” he said.
   But Mr. Freeland said polls can be trusted “for the most part,” although people have to be “careful consumers.” They should not be swayed by viewer or readership polls, or by the polls that appear in the margin of a web site. Those polls are almost always going to be biased, he said.
   Nevertheless, polls are important in a democracy because they form a system of communication between the voters and the candidates, Mr. Freeland said. Polls help candidates to learn how their message is resonating with voters.,and it allows them to tweak their campaign.
   ”It means that candidates respond to voter preferences and you see it in their election campaigns. It is the one time when candidates have to respond to the concerns that ordinary people have,” he said.
   ”The next time you get a survey poll, pay attention. It is important. It is not a nuisance,” Mr. Freeland said.