Polls help map political terrain but hold pitfalls. Public opinion surveys are becoming an ever more prominent feature of the political landscape. To help readers understand polls during the 1988 presidential campaign, the Monitor will run monthly columns by a recognized expert in the field.
Rank-and-file voters, participating through primaries and caucuses, play a larger role in selecting presidential nominees now than in any earlier period of American history. Given this fact, news media and candidates incessantly survey the public's preferences. A small mountain of poll data has already risen on the 1988 Democratic and Republican contests. Together these polls can help us better understand current politics. But when used carelessly, polls can be an abundant source of confusion. We need to be aware of how it is that competently done surveys can readily spread misinformation.
The Democratic nomination contest is more open and fluid than the one on the Republican side. In late October, Yankelovich/Clancy/Shulman asked Democrats and Democratically-inclined independents who their first choice was among the declared field for the party's presidential nomination. The results: Jesse Jackson, 23 percent; Michael Dukakis, 12; Albert Gore, 8; Paul Simon, 7; Richard Gephardt, 5; Bruce Babbitt, 2; None/Not sure, 43.
The main lesson most observers have drawn is that Mr. Jackson's apparent lead is extremely soft, and that the contest remains remarkably unsettled. Jack-son's apparent margin, which accrues largely from greater name recognition, completely vanishes when Democrats are invited to express their choice among a wider field, including some relatively well-known leaders such as New York Gov. Mario Cuomo. Also, Jackson's consistently high ``negatives'' must be considered: Many Democrats say they just won't vote for him.
The polls show that none of the announced Democratic candidates is even close to front-runner status. A huge proportion of the party's rank and file is still looking for someone else. In this setting, it would be silly to make very much of how one candidate stands compared with another - Dukakis compared with Gore, for example. No candidate has yet locked up substantial public support. That is the only conclusion to emerge firmly from polls on the Democratic contest.
By comparison to the Democrats' race, the Republican contest is fairly settled. For the past year it has had in George Bush a front-runner and in Bob Dole a clear No. 2 contender. In the late October survey by Yankelovich/Clancy/Shulman, Pat Robertson stood at just 6 percent, Jack Kemp at 5 percent, Alexander Haig at 4 percent, and Pete du Pont at 1 percent. And, since Mr. Bush and Mr. Dole are both well known, differences in their standing aren't mere artifacts of differing levels of name recognition.
The accompanying graph shows Bush and Dole support in certain national surveys done between Jan. 1 and Oct. 25 by six leading survey organizations. All of these data are from ``full field'' questions, where rank-and-file Republicans are presented with a list of the declared and semi-declared contenders and asked their first choice. Trial heats of the ``if it came down to Bush vs. Dole'' variety have been excluded. While polls vary as to how the potential Republican electorate is defined - all Republican identifiers, registered Republicans, Republicans who are likely voters, etc., - experimentation shows that these differences are not the source of much variation in results.
The graph tells us several important things about the Republican race and about using polls to describe it. First, the contest does indeed have a clear structure. George Bush is ahead, and while future developments may change this, he has held his lead over Dole throughout 1986, in good times and bad. Bush's edge in October is about what it had been in January.
Second, the graph reminds us how easily errors can creep in when polls are used carelessly. The Republican race is more coherent than that on the Democratic side, but it is still very fluid. Add to this inherent volatility poll variations resulting from differences in sampling, question wording, placement in the larger questionnaire, and other such factors.
Journalists sometimes compare the results of a trial heat posed by one survey organization at Time 1 to those of another at Time 2. Consider the confusion thus perpetrated when Bush and Dole are compared in January, using the CBS News/New York Times data, and in September using the NBC News/Wall Street Journal findings. Or, when they are ranked in January with the NBC/WSJ data and in October with the ABC News/Washington Post survey. The first comparison says that Dole has been gaining; the second that Bush is leaping ahead. In fact, neither is true.
The final lesson contained in the graph is that many Republicans simply have not made up their minds at this point, or have made only the lightest of selections.
When public thinking thus is not fixed - when it is, in effect, an unanchored boat - the slightest puff can move it. These puffs may be supplied by poll methodology or by short-term developments in the political environment. There is a lot of bounce from one poll to another that isn't real, or that is so superficial it should be ignored.
The challenge to poll-based campaign coverage is learning how to distinguish public judgments that are substantial from those that are like cotton candy.
Everett C. Ladd is the executive director of the Roper Center for Public Opinion Research at the University of Connecticut.