Imagine standing on the rim of the Grand Canyon while a guide says: "You probably think there is a canyon here, and to the untutored eye there appears to be a canyon, but I have data that demonstrate the canyon is not here." There have been various attempts to demonstrate that nothing much has happened in American politics in the 1980s, but no attempt as determined as the article in the May issue of The Atlantic, "The Myth of America's Turn to the Right."

The authors, Thomas Ferguson and Joel Rogers, deny that "a majority of the public has reached a stable, well-informed consensus" in favor of conservative policies. Note the adjectives "stable" and "well-informed." They are loopholes.

There often is a gap between the rhetorical and the real politics of Americans, between the way they talk and the way they really want to be governed. The authors say Americans are "ideologically conservative and programmatically liberal." Certainly conservatism often is the belief that it is time to cut my neighbor's subsidy.

But Ferguson and Rogers argue not just that Americans remain "programmatically liberal." They say there has been a "generally liberal trend" regarding domestic issues, and the "movement of public opinion has been directly opposite to the movement of public policy."

They cite an increase in the belief that there is too much concentration of economic power, and the belief that business is making "too much profit," and that government should limit profits. They cite increasing tolerance about such matters as abortion and women's rights. They even find evidence of the "liberal trend" in the fact that 89 percent of the people believe "there can be no winner in an all-out nuclear war," and the fact -- whatever it means -- of a growing "aversion" to nuclear weapons.

If the authors are right, the political system is radically defective. It is producing radically unrepresentative government. If the authors are wrong, and Democrats decide the authors are right, Democrats could produce another disaster.

However, a clue to the weakness of the authors' thesis is that they argue it too aggressively, explaining away even Ronald Reagan's popularity. They say that if one "controls for economic conditions," Reagan's popularity is not significantly higher than Jimmy Carter's. Well, yes: Reagan would not be popular if he produced stagflation. The authors even cause the 1984 landslide to disappear: they note that Reagan received the votes of only 32.3 percent "of the potential electorate." With such intellectual overreaching, the authors have provided Democrats with a road map for another drive off a cliff. It is an old conservative map, reversed.

In 1964, conservatives nominated Barry Goldwater, the last candidate unreconciled to the post-New Deal role of the federal government as an engine of distributive justice. Conservatives had a theory -- the "conservatives in the woodwork" theory. Conservatives thought: one reason millions of Americans do not vote is that they are forced to choose between two liberals. Give them a choice, not an echo, and conservatives will pour out of the woodwork, into voting booths.

The Atlantic article suggests there are lots of liberals in the woodwork. The authors seem slightly scandalized that Mondale "became the first Democratic nominee in many years to fail even to put forward a major jobs program." They say "there is no particular mystery" why, when offered a choice between candidates so similar, people rejected the one who promised to raise taxes, "or declined to vote at all."

The data in the article are not startling. Americans always want more of everything from government, except taxes and war. However, the article illustrates the limits of quantification in political analysis. Political scientists want to count things, such as the numbers of people who subscribe to particular propositions. But although they produce numbers, polls about propositions may be blunt instruments for measuring national moods.

The voter's presidential choice is personal and complex, involving a mixture of hopes and vision and a sense of being comfortable with a candidate. Such an alloy of impulses is difficult to quantify. But the alloy is not necessarily nonrational. Any consensus can be dismissed as "unstable" and "uninformed." But it is wise to start political analysis with this axiom: when the data are utterly inconsistent with common sense, doubt the data.

That axiom flows from this one: the voters and the politicians know a thing or two, including each other. The former usually get the latter what they want, and the latter generally give the former what they want.

If your data say the Grand Canyon is not really there in front of you, you can get new data, or you can prove your confidence by striding forward. Republicans urge Democrats to read the article, trust the data and take 10 steps forward.