Can the Polls Be Right?

The election this November is the crucial one for the major polling organizations. They were wrong in New Hampshire, wrong in Oregon, closer, but still wrong, in California. If they are seriously inaccurate again in November, it will be a long time before any politician again trusts polls as Lyndon Johnson seems to.

But the polling organizations seem to be fighting their crucial battle under adverse conditions. The same problems that apparently baffled them in the primaries face them again in the election, and there is no evidence that any of the problems has been solved.

Louis Harris answered criticisms of his apparent failures in the the primaries by saying that the job of a poll was that of a reporter and not that of a prophet. By polling a given number of people, Harris said, he can determine how the country feels about an issue at the time the poll is taken. In a situation where feelings change rapidly, however, polls cannot be taken close enough to the election to reflect the full impact of the changes. Harris and Dr. Gallup were able to predict in New Hampshire that a trend was running towards Lodge, but they underestimated it. The same was true of the Goldwater trend in California.

In the Presidential election, the polls will once again be covering unstable emotional ground as November 3 approaches. For one thing, polls have never had great success in predicting elections in which race is a major issue. It has been observed in Southern elections that the more segregationist candidate often comes on strongest in the last two weeks of a campaign: this is the kind of change that goes unnoticed by polls.

Last year, for example, Edward Breathitt defeated A.B. Chandler in the Democratic primary for governor of Kentucky. Race was not the most important issue in the campaign and the polls taken for Breathitt predicted his margin precisely.

Before the election in November, however, Democratic Governor Bert Combs issued an executive order outlawing segregation in businesses licensed by the state. Breathitt's Republican opponent, Louis Nunn, attacked him on this issue vigorously. Still Breathitt's polls showed him leading Nunn with 58 per cent up until election day. Breathitt won by a bare majority.

Race is not the only emotional issue in the campaign, nor are the Republicans the only party likely to try to take advantage of an emotional surge in the last few weeks. The issue of Goldwater's involving the country in a nuclear war could change a lot of votes in a hurry. So could a revival of sentiment in the name of John Kennedy. Goldwater might not be averse to trying to work up an emotional landslide on the issue of Presidential morality.

Professor Pettigrew has made some recent remarks which illustrate another circumstance that might distort the predictions. In denying that a "white backlash" had sprung up recently, Pettigrew pointed out that an abnormally large number of voters had turned out in Northern primaries where Governor Wallace was a candidate. Wallace's support, he suggested, had not come from a sudden change of heart on the part of people who voted regularly, but as a result of what he called the "out from under the rocks effect"--people who had always disliked Negroes now had a chance to express their dislike with their votes.

Something similar to Pettigrew's "rocks" effect had been a central tenet of Goldwater's conservative thesis long before he won the nomination. It was the Senator's contention that a majority of American voters are in his sense of the world, conservatives. This did not show up in the election returns, he said, because conservatives stayed away from the polls when both parties nominated liberal candidates.

Whether or not one agrees with the Goldwater theory (and the results of the New Hampshire and Oregon primaries seem to belie it), it is clear that if Goldwater is correct, the polls will fail to reflect this. Goldwater is relying on the support of habitual non-voters; Dr. Gallup discounts the votes of anyone who has not voted in any of the last three elections, and Mr. Harris must have some similar provision. In other words, it is a necessary corollary to the Goldwater thesis that polls will fail to predict accurately the result of a Presidential election in which a conservative opposes a liberal.

The poll-takers have undoubtedly tried to correct the errors that made them miscalculate in the primaries. They may change the size or the nature of their samplings; they may speed up the process of taking their final polls., taking their samples closer to election day itself. But polling organizations are in a risky business, where a good Election Day rainstorm in a Democratic area can reverse carefully predicted results. In an election with as many unpredictables as this one, it will be a major accomplishment if the polls are able to foresee the outcome.