The Pulse

By ELLEN M. DRAN

In this election year, hardly a day goes by without the reporting of results from one or more surveys of public opinion: Do you approve of the job President Bush is doing? Will Bill Clinton's past behavior affect your vote? Are you better or worse off financially than four years ago? If the election were held today ....

Public opinion on candidates and on a lot of issues is fairly well known, but little is known on what the public thinks about the polling process by which its views become known to policymakers and to the public in general. Last fall, the Illinois Policy Survey began to look at the rather neglected area of public opinion about public opinion polls. The intention is to expand this study in future editions of the poll, conducted annually by the Center for Governmental Studies at Northern Illinois University. The focus of this preliminary study was perceptions of the role of opinion surveys as a means of communicating with elected representatives and affecting public policy.

Figure 1.

Trust in accuracy of public opinion polls

How often do you think you can trust the results of public opinion polls to represent what people think about important issues? (n = 800)

When asked, "How often do you think you can trust the results of public opinion polls to represent what people think about important issues?" 6 percent thought poll results are almost always right, and 38 percent said they are right most of the time. A majority took the more negative side, saying polls are right only some time (46 percent) or hardly ever (8 percent). (See figure 1).

Among the demographic groups pollsters usually like to look at, such as gender, income classes, education levels, political party, the only group which had a majority give the more positive responses was young adults. Fifty-five percent of the 18- to 34-year-olds think polls are right almost always or most of the time, probably a manifestation of youthful enthusiasms, or perhaps older folks are more jaded about public events in general.

Also, blacks were more skeptical than whites: 46 percent of white respondents gave one of the positive responses versus just 29 percent of blacks. The data do not provide an explanation for this racial difference, but it may be that African Americans perceive polls as misrepresenting black interests or perhaps as diluting the black perspective in the aggregate results. This is something to ponder in future investigations of public opinion on public opinion.

Otherwise, regardless of their personal characteristics, the same percentages of Illinoisans gave positive and negative assessments of poll accuracy.

The wariness about poll accuracy may be healthy skepticism and an unwillingness to give blind trust to a methodology the public does not fully understand. This caution, however, does not appear to mean the public does not value polling. The survey found only 19 percent indicating that there are too many polls; 48 percent said that about the right amount of polling is done; 26 percent think there are too few polls.

30/July 1992/Illinois Issues

The 54 percent who gave one of the negative responses about poll accuracy were asked a follow-up question on what they think causes polls to be wrong sometimes. The most common complaint was not directed at polls or pollsters but at other survey respondents. More than a fourth (28 percent) indicated that others are not as thoughtful about answering the questions as they are. (See figure 2.) Examples of actual answers: "People aren't thinking, giving answers from the top of their heads"; "It's hard to get an honest answer"; "People . . . lie."

It is unclear how respondents arrived at this assessment. Having listened to innumerable interviews as a polltaker, I am generally impressed at how seriously so many respondents take the questions and how carefully and thoughtfully they answer them, although there are at times those who "aren't thinking."

Some fairly standard misconceptions about survey methodology were evident from the survey: that polltakers talk to the wrong people (20 percent) or that sample sizes are too small (13 percent). Actual answers included: "Only middle-class people respond — no very poor or very rich people"; "People are picked or selected."

Although not perfect (polls tend, for instance, to underrepresent young people and minorities), scientifically drawn samples are more representative of the total population than is the electorate when only about a third of the public typically votes in state elections and a bare majority participate even in presidential elections.

Figure 2. Why polls are sometimes wrong (n=431)

Table 1. Choice of polls or citizens for representation in general and for two issues
	Polls	Citizens
In general	48%	60%
Abortion	46%	62%
Taxes	48%	70%
(number)	(393)	(408)

As for sample size, admittedly it is counter intuitive that 800 survey respondents can faithfully represent the views of 11 million Illinoisans or 3 million Chicagoans or 150,000 in Rockford or even 250 million Americans, all with the same degree of accuracy, but in fact it is so (ask a statistician).

What may perhaps be surprising is that only 20 percent think survey errors are due to some kind of manipulation by pollsters or misuse by the media:

"When it hits the press, it gets changed"; "The reliability of polls depends on the polltaker." These respondents, along with the 8 percent who said that polls ask wrong or bad questions, identify problems that are familiar to all survey researchers. Most pollsters are not only attuned to problems of biased questions and misuse of poll data, but they constantly monitor and evaluate research to minimize them.

The public, however, is right to be ever wary and vigilant about manipulation and misuse. Public opinion polls are now a pervasive factor in American political life. As with voting, polls benefit from an element of equal representation and a capacity to level differences in political influence on public decisions. It is important, therefore, for the polling community to know what the public thinks of polling and to develop ways to educate the public about polling methodology and use.

Now, if the elections were held today ....

Ellen M. Dran is a research associate in the Center for Governmental Studies at Northern Illinois University, which has conducted the Illinois Policy Survey since 1984.