POLL OF POLLS **: The Chasm Narrows

POLL OF POLLS

The course of George McGovern’s presidential campaign apparently has turned a corner of sorts—at least in the public-opinion polls. He is beginning to narrow the huge lead held by Richard Nixon. Both the Gallup poll and Harris survey now place McGovern 28 percentage points behind Nixon, a gain for the Democrat of six points since the previous surveys by both polling firms.

Those identical figures would seem to contrast with other major polls, such as the most recent Daniel Yankelovich Inc. Poll for TIME, which had McGovern trailing by 39 points, an Albert Sindlinger survey putting him 41 points behind, and a Democratic-sponsored poll by Pat Caddell, showing only a 22-point deficit. Except in the case of Sindlinger, whose polling techniques did not screen out probable nonvoters, the apparent discrepancies seem to be largely a matter of timing. Taken over a period of 21 weeks (Aug. 25-Sept. 12), the Yankelovich study, for example, showed a slight upswing for McGovern in its last week. The most recent Gallup and Harris polls, conducted in late September, confirmed that some pro-McGovern shift in sentiment was taking place.

With Nixon’s 28-point lead, despite the fact that Democrats hold a big voter-registration advantage—the polls are once again under fire as unreliable. This sentiment surfaces repeatedly as the U.S. public quite naturally resists the notion that it can be so neatly and numerically tabbed on the basis of what a relatively few Americans (about 1,500 in the normal sample) tell the pollsters.

The polls have indeed had problems in detecting the drift of such multi-candidate contests as presidential primaries, where voter opinion is much less certain and is affected by more intangibles. But to those who object to having their opinions computerized, the discomforting truth is that the major polls have been astonishingly accurate in predicting presidential elections ever since the miscalculations of the Harry Truman upset of Thomas Dewey in 1948. Since that time, however, the largest discrepancy between the final Gallup reading, for example, and a presidential-election result was the 4.4 points by which Gallup underestimated Dwight Eisenhower’s voter popularity in 1952. After that, Gallup’s biggest miss was the mere 2.7 points by which he overestimated Lyndon Johnson’s 1964 edge over Barry Goldwater. The average Gallup error in a presidential election since 1948 has been only 2.04 points on the winner. Because the sampling techniques of the major polls allow for the statistical possibility of a plus-or minus-three-point error (meaning they could miss the spread between the candidates by 6 points), the record is impressive.

Cheering. Despite such reliability, the polls do raise troubling problems about their impact. There is no doubt that McGovern’s ability to raise funds has been hurt by his low standing in the polls, and this could conceivably propel him further downward. Testifying last week before a House subcommittee on a bill that would require all public-opinion firms to disclose their methods of operation, both George Gallup and Louis Harris disputed the idea that a commanding lead in a poll would automatically produce a bandwagon effect in which the release of the news would further increase the margin. Just as likely to happen, Harris argued, would be a reverse effect caused by sympathy for the underdog and complacency on the part of the leader’s supporters.

Gallup, in fact, recently released statistics showing that the presidential candidate who trailed in early September gained between one and ten points by election time in every race after 1944 with the single exception of Adlai Stevenson’s two-point drop in 1956. Declared Gallup: “Polling experience gained in presidential elections since the 1930s indicates that the present wide lead of Nixon over McGovern could vanish before Election Day.” That may well be one of the most cheering things any pollster has said about the McGovern campaign to date. Some McGovern workers appreciate the black humor of their position as they survey the generally bleak public-opinion reports. Looking at Caddell’s findings, one McGovern aide laughed and conceded: “Here I am chortling over a poll that shows us 22 points behind.”