Those "Worthless" Polls

Predicting the future has never been a reliable craft, whether practiced by tea-leaf readers or political pollsters. The 1980 presidential primary season has been an especially difficult time for the latter. TIME'S national political correspondent John Stacks explains why:

The headline in the New York Daily News just four days before the state's primary election left little doubt about what the result would be. Proclaimed the News: CARTER'S THE ONE--IN A POLLSLIDE. That conclusion was based on an opinion survey of 800 potential voters in New York State conducted by Louis Harris. The results showed Carter beating Kennedy 61%-34%. By the weekend before the election, Harris had new data, gathered from a survey of 600 potential voters, showing the gap narrowed to 56%-36%. Then came the results from the real polls on Tuesday night, and the size of Harris' error was clear: Kennedy had won by 59%-41%. The Harris polling error: 38 percentage points.

Harris was by no means alone in his discomfiture this election year. Polls in Iowa before the January caucuses showed Ronald Reagan slightly ahead of George Bush; instead Bush won by 31%-29%. In the Maine caucuses, the Bangor Daily News had Carter ahead of Kennedy by 19 points the weekend before the voting; Carter beat Kennedy by only 6%. In New Hampshire, a Boston Globe poll put Reagan and Bush almost neck and neck the Sunday before the election; Reagan walloped Bush 50%-23%.

Public opinion researchers are nearly unanimous in arguing that the problem lies in the special circumstances of primary elections. Unlike general elections, primaries are used by voters to protest, to slow front runners, to send messages of dissatisfaction to incumbents, without anyone bearing the burden of actually deciding who should be President. Ruth Clark, senior vice president of the firm Yankelovich, Skelly and White, says of the much trumpeted primary samplings: "I just don't believe in them."

The difficulty, says Clark, is that those polled represent the entire electorate, not those who will actually vote. "No one has found a reliable way of identifying those people most likely to come out to vote," explains Clark. "In general elections, people remember voting before, and you can rely on their memories. They tend to forget their own behavior patterns in previous primary elections."

This was the problem that Louis Harris faced in New York. He was surprised by the low turnout around the state, which left people who felt tepidly pro-Carter at home while people who were angry turned out to vote against him. Laments Harris: "I just wish to hell we'd used a lower turnout figure. We failed on the magnitude of the swing against Carter, but we caught its direction."

Harris faced another problem, that opinion after he took his last poll was still shifting fast. Says Clark: "In a poll, you are only measuring opinion at a single point in time. Things change and opinions change." It was just such a shift that threw the New Hampshire polls off. Bush and Reagan were doubtless close, but on the Saturday night before the election, Bush got trapped in his refusal to let other G.O.P. candidates join his debate with Reagan, and opinion quickly shifted against him.

The fallibility of preprimary polls leads some to dismiss their utility altogether. Says Lucien Haas, an aide to Senate Majority Whip Alan Cranston of California: "The polls are absolutely worthless." That is an exaggeration. The real problem is that since polling can assess the views of a body of voters but not which of those voters will actually vote, a preprimary sampling is only an approximation of what is likely to happen. Any politician or pundit who attributes to such a poll more accuracy or importance than it can realistically have does so at his own risk.

This file is automatically generated by a robot program, so viewer discretion is required.