With the fall election season about to begin, we'll regularly see candidates and journalists using polls to embolden their followers or favorites, and to discourage others. What should we know about the construction of polls? Scott Rasmussen, born in 1956, is president of Rasmussen Reports, which Fordham University found to be the most accurate polling organization in the 2008 presidential contest. Election expert Michael Barone calls Rasmussen, an evangelical, "one of America's most innovative pollsters."
With polls these days conducted by phone, how do you develop your call list? List selection is really critical. One reason is that people answer phones differently. People in urban areas are less likely to answer than people in a rural setting, so you have to select a call list that will get out to all different areas of the country and get you a good mix of people. At the same time, it's supposed to be totally random. So you have to have a process where you select numbers at random, and then make sure that you get enough calls from each segment of the country. It is a real challenge to set up the call lists and at the beginning of the year the protocols for using them.
When you call, what percentage of people answer the phone-and of those who answer, what percentage will answer questions? Half the phones in the country have an answering machine that picks up before you even get through, and then with 10 percent you get a fax tone or something else. The number of people who pick up the phone who will complete a survey will typically be 1 out of every 3-5 people. Also, we screen people out. If you call in the early afternoon, about a third of your respondents will be under 18. They're kids who are home from school. We screen them out.
Which people are more likely to respond? Women answer the phone more than men, older people more than younger people. Men will stay on longer if they are interested in the survey.
Do you use a recorded voice or a live questioner? Companies used to do polls by having a room full of people dialing up people and asking questions. We have a single voice recorded. It's played out to everybody. It's a 30-something female Midwestern voice, which we picked because we tested sports announcers, opera singers, professional announcers, anything you can think of-and a Midwestern female voice got the best response. Non-threatening, easy to understand, gives us a consistency that the operator-assisted polls can't possibly have.
How do you decide when you're preparing the poll the percentage of Republicans and that of Democrats to ask? If we're polling on a non-political matter (for example, Tiger Woods), we poll all adults. We go to the Census Bureau and it's very easy. When you get to a political poll, we balance Republicans and Democrats-but the real question is who will show up and vote. If I ask someone over the phone, "Are you going to vote?" there's a civic pressure to say yes: Some people who aren't actually going to vote will lie to themselves and to us and say yes. We have to ask how often they've voted in the past and how interested they are in the issues. We have a formula to estimate who's actually going to show up and vote. It's an ever-changing process.
Do Republicans and Democrats have different response rates? Not noticeably. The hardest group to get in touch with is moms with young children. Senior citizens are the easiest to find at home and the most willing to take a survey, so if you called on a totally random basis you would have way too many seniors in your sample. Cell phones are the next big problem for us. It's an industry-wide problem that a majority of people under 35 no longer use landlines as their principal phones.
On some questions are people more likely to give an answer they think is politically correct? In election polling, that's not much of an issue. It used to be a bigger issue in candidates. Minority candidates would typically overpoll: People would say, "I'll vote for someone," and might not, but that seems to have faded a bit. This is also less of a problem with automated polls. On some issues you have to be concerned about honesty. If I ask, "Have you used illegal drugs in the past 24 hours?" I'm probably not going to get a really good response. But if you ask, "Have any of your close friends used illegal drugs in the past 24 hours?" or, "Do you know anyone who's used illegal drugs in the past 24 hours?" you can sometimes get a better estimate. We did something in 2008 during the last election cycle: "Would you be willing to vote for a woman for president?" Everybody said yes. But then we asked, "Would your close friends be willing to vote for a woman for president?" Many senior citizens said no.
Let's talk a little bit about the philosophy of polling. Is it a problem when people take polls too seriously? Politics in particular has become the new sport. Polls become the statistics of the day, so it's easy to see them abused. Having said that, it's really more important now than ever to have polling data out there that lets public opinion be heard. In our country we're struggling. We have a political class in Washington that is semi-permanent and doesn't have a clue about what life outside of D.C. is like. Anything that can awaken them to those perspectives is a plus. I can think of examples big and small. It never occurred to anyone in Washington officially that the bailouts late in 2008 were a bad idea, but only 28 percent of Americans thought they were a good idea. Polling is a vitally important way to give a voice to the people.Are you concerned about what the Founding Fathers called "mobocracy?" In my new book, In Search of Self-Government, my concern is not that we have too much control by the mob but that the self-governing instincts are being limited. Self-governance means you make the most important decisions of your life by yourself. In some broad manner, whatever you believe the policy should be, you should be able to sell it to the populace. The government needs to have that connection with the people.