DENNIS MYERS Against the Grain Advertisement

In 1968 the presidential campaign of Hubert Humphrey, lagging badly in all reliable opinion surveys, came up with a careful scheme to manipulate the press.

Humphrey's executive assistant William Connell planned to scatter positive poll results around to selected journalists. Since reporters are, almost by definition, unsophisticated in reading survey results, soon the nation was seeing reports of a Humphrey "upsurge."

It was all nonsense. The polls the Humphrey campaign distributed were internal polls done by the campaign's own consultant, Joseph Napolitan. In addition, the campaign's negative surveys were withheld.

The ploy had its impact. There's an opinion pattern in politics (and in other fields) called the bandwagon effect. When polls show a swing to a candidate, people tend to flock to that candidate. One 1998 study showed that when one candidate "is an initial favorite by a slim margin, reports of polls showing that candidate as the leader in the race will increase his or her favorable margin."

Humphrey was soon in an essentially tied race. As three British journalists wrote, the operation "was highly successful. Most newspapers swallowed it hook, line, and sinker."

This kind of manipulation is one of the reasons journalists have an affirmative obligation to handle opinion surveys with care and caution. Which is not what happened in Nevada in the past couple of weeks. An opinion poll was shopped around to various media entities, many of which featured it prominently. It supposedly showed that Gov. Jim Gibbons jumped from an approval rating of 29 to 49 percent, a dramatic gain in a period when he was bungling several issues and the Nevada Legislature was slapping him around.

The poll was used in numerous publications, sometimes on the front page. A few newspapers did not carry it, and as best I can tell the local bureaus of the Associated Press did not pick it up.

The survey was not an independent poll. It was an internal poll, commissioned by the Nevada Republican Party from the pollster who worked for Jim Gibbons in the 2006 campaign. But it's likely to create an upsurge in more legitimate later surveys.

The Columbia Journalism Review (CJR), a magazine that monitors journalism behavior, asked a few months ago, "In the history of American politics has a poll ever been conducted that political reporters were unwilling to cite?" That question appeared in an article entitled, "Why Do Journalists Bother to Report on 'Internal Polls'?"

One of the reasons news coverage of surveys is often irresponsible is that few reporters who write about polls actually know how to read them. Columbia University communications professor Todd Gitlin says many journalists suffer from innumeracy, "mathematical illiteracy, inability to reason with numbers." CJR reports that leading national pollster Richard Bennett of the American Research Group says that journalists "routinely misinterpret the confidence intervals that accompany the margin of error."

Few journalists know anything about the central limit theorem governing surveys, yet they play with poll numbers like kids with matches. The stories about the internal Gibbons poll by a Nevada firm compared it to a previous independent survey by a national firm, when there was no information suggesting the two are comparable in questions or methodology. I asked Gitlin why he thought we journalists are so indiscriminately bewitched by polls.

"The bewitchery is the interesting question," he replied, "and I think it's partly a function of the fact that people don't understand polling. Numbers seem firm, precise, hard. In this country (and others), in the great soup of guesswork about 'what people think,' people are often impressed by a result that looks definite. Because we believe (usually) that the people's opinion should prevail, polls combine the cachet of mathematical precision with the aura of holiness that attaches to vox populi."

Of course, public opinion is not precise or firm. That's all the more reason to treat polls gingerly. It would be nice if reporters put the same care into their research that campaign consultants put into manipulating reporters.