|
|
|
Polling Place
Quinnipiac's superstar surveyor plays 20 Questions
|
Business New Haven
4/30/2001
By: Susan Cornell
|
Is it highly likely, somewhat likely, or highly unlikely that Hamden is home to one of the top polling institutes cited frequently by journalists, researchers, and public officials? The answer: highly likely, indeed.
The independent Quinnipiac University Poll regularly surveys residents in the tri-state area about national and state political campaigns, and issues of public concern such as taxes, hand-held cell phones for drivers, schools, municipal services, transportation and the environment. And, there are the not-so-regularly conducted polls on topics ranging from who New Yorkers felt would win the Subway Series to Connecticut residents' opinions regarding the news media in the murder trial of Michael Skakel.
It is this polling institute that is referenced in national network news broadcasts and whose findings are featured regularly in the Wall Street Journal, New York Times, Washington Post and USA Today.
Mission & Methodology
The mission of the Quinnipiac University Polling Institute is to “conduct timely and accurate public-opinion polls on politics and public policy in New York, New Jersey and Connecticut as a public service and for academic research.” Simply stated, the Polling Institute informs the public and serves society - and, in the process, helps to market the school to potential students near and far.
Interviewing is conducted from the facilities of the Quinnipiac University Polling Institute at the western edge of the Hamden campus. Professionally trained students and non-students conduct interviews using a Computer-Assisted Telephone Interviewing (CATI) system to collect data from residents of the three states almost every night. This is truly a complete team effort with a corps of about 100 players who may be enrolled in any number of the university's programs of study - business, health sciences, law, liberal arts, education or communications.
A typical public-opinion survey involves interviewing a randomly selected sample of roughly 1,000 adults over a five- or six-day period. According to the Institute, the sampling error is 3.1 percentage points at the 95 percent level of confidence. This means that 95 percent of the time, the results obtained should be no more than 3.1 percentage points above or below the figure that would be obtained by interviewing the entire population.
Oh Yeah, I've Heard of That
One of the Polling Institute's claims to fame is for the most accurate prediction on the Charles Schumer-Alphonse D'Amato New York U.S. Senate race in 1998, won by challenger Schumer. While most of the polls conducted in the Millennium year focused on the elections, polls are conducted on any relevant topic.
Senatorial and Presidential politics dominated the Quinnipiac University polls in 2000. A First Lady running for political office became the featured performer, but the institute also turned its attention to non-campaign issues as well. These included the abortion, the use of undercover police, giving additional state money to New York City schools and capital punishment.
Superstar Surveyor
Et tu, Bruté - time to poll the pollster. BNH interviewed the director of the Quinnipiac poll, Doug Schwartz, on a variety of hot topics in the news regarding polling as well as on the institute's role. Schwartz is an expert in public-opinion research who designs the survey samples, develops the questionnaires, performs the statistical analysis, interprets the data, and summarizes the poll's findings for the media. He previously served as a survey associate with the CBS News Election and Survey Unit and as an election night analyst for CBS senior correspondent Ed Bradley. Schwartz also worked with ConnPoll for the Institute of Social Inquiry at the University of Connecticut.
He is a 1988 cum laude graduate of Connecticut College and holds a master's degree in political science with a concentration in survey research from the University of Connecticut. As answers to many of these “survey questions” could potentially require pages or even textbooks for responses, we limited Schwartz to 100-word answers.
What is the history of polling?
George Gallup began conducting scientific surveys in the 1930s and correctly predicted Roosevelt's easy victory in the 1936 Presidential election. Gallup's correct prediction helped to establish the credibility of scientific surveys. While Gallup got it right, the then-famous non-scientific Literary Digest Poll incorrectly predicted that the Republican Alf Landon would win. Although the Literary Digest Poll received responses from over two million people to its mail survey, its sampling was flawed. The Literary Digest used mail lists that were based on car registration and telephone directory lists. During the Depression, many people with lower incomes could not afford a car or phone, and they tended to be Democrats.
What is the history of the Quinnipiac Poll?
Quinnipiac University began polling in Connecticut on politics and public policy in 1988. Quinnipiac expanded its polling to New York State in 1994 and New Jersey in 1996. The Quinnipiac University Poll is the only poll in the country that conducts regular surveys in three different states. We are also the only poll that conducts regular surveys in New York City.
How can the news media improve its coverage of polls?
In general, I think the media does a very good job in their coverage of polls. When reporting poll results reporters should print the exact question wording of the survey question. How the question is phrased has an effect on the responses. For example, when we asked people whether they favored or opposed the death penalty, about 60 percent said they favored the death penalty. When we asked people to choose between the death penalty and life in prison with no chance of parole for people convicted of murder, people were evenly divided.
According to a poll on polling, approximately 40 percent of the general public refuses to participate in surveys. Why is that so?
People refuse to participate in surveys mainly because they are too busy to participate. Another way to look at it is that given that people have busy lives, why do they take the time to do a survey? One reason is that they have heard about poll results in the news media and want to see how it is actually done. Some people are never asked their opinions about politics and public policy, so they welcome the opportunity to share their opinions with us. Others recognize that by participating in the survey they are making their views known to policymakers.
Why is the public's response to poll interviewers declining?
It may be that more people are reluctant to participate in legitimate opinion surveys because they fear that someone will try to sell them something. In our surveys, we make it clear that we are not selling anything. We are simply interested in people's opinions. I think a bigger problem than higher refusal rates is the greater difficulty pollsters have in simply contacting a respondent. With the advent of answering machines and caller identification, more and more people are screening their calls, making it harder for pollsters to reach them.
What are the effects of the unwillingness of some people to participate in polls?
The refusal rate is a concern of pollsters. However, it is not a problem for pollsters because those who do the survey do not differ in a systematic way from those who do not [respond to] the survey. If those who refused to do the survey differed in a systematic way from those who did the survey, polls would not accurately reflect the views of the population. A good measure of the accuracy of polls is that pre-election polls do a good job not only over predicting who will win, but the margin of victory as well.
Why is the number of polls conducted increasing so?
More polls are being conducted because there is a demand for them. With 24-hours news channels like CNN, MSNBC and Fox, the news media needs more news to fill up the time. Polls are great news stories.
How can the public better understand polls?
Although scientific sampling is accepted by the news media, politicians and academics, the public may still be somewhat skeptical of how a sample of 1,000 can represent the views of millions. Many people say that they have never been called by a pollster and therefore they do not trust the results. Quinnipiac conducts about six polls a year, with about 1,000 respondents in each poll. Given that there are about 2.5 million adults living in Connecticut, the chances are not very high for one individual to be in a survey.
While some pollsters say the Internet allows them to collect public attitudes more quickly and inexpensively than the telephone, other public-opinion researchers say that surveys done strictly online don't measure up. What are the drawbacks to Internet polling?
The major drawbacks of online polling are: A) Only about 40 percent of people have access to the Internet at home. Those who have the Internet tend to be more educated and higher-income. Thus interviewing only people who have access to the Internet would skew the survey's results toward those with higher socio-economic status. B) Even if everyone had access to the Internet, there is no list that includes everyone's e-mail address that a sample could be drawn from. C) The worst kind of online polls are those that are based simply on respondents who happen to visit a particular Web site and want to do the survey. You cannot generalize the results of these polls to the general population, or even to Internet users. You can only talk about those people who did the poll. Such polls should clearly state they are unscientific and unrepresentative.
What are the benefits of online polling?
One use of on-line polling may be to measure public reaction to things like campaign advertising. However, I would still use traditional telephone sampling methods to select the sample. Then I would provide a computer and Internet service to those who did not have it. This is obviously costly, but necessary to have a good random sample of people who would participate in this online poll. Those taking the poll could be shown advertisements - something that obviously could not be done over the phone - and then asked questions about those advertisements.
Will anyone figure out how to draw a random sample of computer users the way traditional pollsters draw a probability sample of the population?
Yes, but it could take a long time before everyone has the Internet and there is a list with everyone's e-mail address. Even then there would be a problem: You can't just send someone a survey over the Internet without their permission. This issue would need to be addressed as well.
Murray Edelman of Voter News Service advised journalists recently, 'Don't give [Internet polls] any credence at all.' Do you agree or disagree?
If the online poll is simply the responses of those who happened to visit a particular Web site, I wouldn't pay any attention at all to it. If the sample was drawn through traditional telephone random sampling methods and everyone who did not have access to the Internet was given a computer and provided access to the Internet, I would pay attention to it. I believe CBS News currently conducts this kind of poll on occasion to measure instant public reaction to an event, such as a Presidential debate.
What are your thoughts on proposed legislation requiring, for example, that exit pollsters stay at least 1,000 feet away from polling places?
As long as it does not make it more difficult for the exit pollsters to get people to do the surveys, I would not have a problem with it.
Did the Voter News Service make the wrong calls in Florida on Election Night or was a series of errors, miscalculations and just plain bad luck to blame?
My understanding is that there was a combination of mistakes that led to VNS incorrectly calling Florida for Gore. Part of the problem was that VNS underestimated the number of voters who would be absentee ballot and tended to be Republican. It is important to remember that VNS has had a very good track of calling thousands of races correctly and missing just a few. Unfortunately for VNS, the one that they missed this time, Florida, determined the outcome of the Presidential election.
What can be done to reduce abuses with exit polling?
The networks should wait until all the polls are closed before releasing exit-poll data from any state. Although there is no conclusive evidence that voters on the West Coast are affected when the networks project the winner in other states, the possibility that they could affect even a few voters is reason enough that they should not do it. One way of accomplishing this would be to have uniform poll closings nationwide. Exit polls are very important not so much because they can be used to project who won, but because they tell us why voters voted the way they did.
What comparisons can be made between Lewinsky-era and Watergate-era public opinion polling?
The public reacted differently to the two scandals: Clinton retained public support, while Nixon eventually lost it. This is not to say the public approved of Clinton's personal behavior. They did not. The public did not believe that Clinton's personal behavior justified having him removed from office. The polls on the Lewinsky scandal contradicted the conventional wisdom at the outset of the scandal. Talking heads thought that public opinion would turn against Clinton and he would be removed. The polls may well have saved Clinton.
Politicians, citizens and the press all take part in pummeling the polls. Why do these groups criticize polls so?
Candidates criticize polls when they do not like the results. No candidate wants to be seen as far behind in a race because that would mean that they will get less money from donors who do not think the candidate can win. The candidate who has bad poll numbers will probably also get more negative press - stories will be about why the candidate is trailing in the polls.
Will poll-bashing ever let up?
No.
Where do polls deliver their greatest value?
Polls are of great value to a democracy because they inform public officials about the concerns of the people. If we did not have scientific public-opinion polls, elected officials would rely upon mail and phone calls to tell them what concerns the public. Yet those who mail or call their elected officials are not representative of the public.
What future changes, if any, do you see for the Quinnipiac Poll?
We will continue to conduct timely and accurate polls in Connecticut, New York and New Jersey.
|
Go FirstGo PreviousGo
NextGo LastGo
to Index
|
|
