The Gallup Organization recently asked a representative sample of Americans this question about the crisis in the Persian Gulf:

"Do you think the president should stick closely to American public opinion when deciding what steps to take next -- including the results of polls like this one -- or should President Bush do what he thinks is best regardless of what the American public thinks?"

More than half -- 56 percent -- told Bush to read the public's lips to decide Middle East policy. Only 41 percent said Bush should follow his conscience.

A look at survey data collected over the past five months and interviews with experts on public opinion and war suggest, however, that policy-making by the polls would be difficult, if not impossible -- if not downright dangerous.

In fact, some experts on public opinion and public policy say they believe Bush has far more latitude to shape and lead public opinion on the Middle East than he or his advisers might realize.

The biggest difficulty facing those who would use the polls to shape policy is that the survey results appear contradictory, reflecting widely shared ambivalence about what the United States should do next in the gulf.

Pushed by fast-changing events and pulled by differing methodologies, public support for war has been reported to be below 50 percent in some recent polls while topping 70 percent in others. An explanation for these differences lies in how the questions were asked.

The impact of question wording on levels of support for military action against Iraq is clearly seen in a recent survey by The Washington Post. That poll attempted to determine why two questions measuring attitudes toward use of military force produce seemingly different results.

Each question has come to play a role, albeit a minor one, in the ongoing policy debate between those who favor quick military action against Iraq and those who currently oppose the war option.

The president's supporters generally recite results from what has been dubbed the "hawk" question by some pollsters. It is asked by a number of news organizations including The Washington Post and ABC News.

This question, asked of half the sample in the recent Post test, reads: "As you may know, the United Nations Security Council has authorized the use of force against Iraq if it doesn't withdraw from Kuwait by Jan. 15. If Iraq does not withdraw from Kuwait, should the United States go to war against Iraq to force it out of Kuwait at some point after Jan. 15 or not?"

The result: 62 percent said yes, the United States should go to war sometime after Tuesday's deadline.

A different question is most often used by people who prefer to use sanctions against Iraq -- as well as those, for reasons that will be clear later, who back an immediate military strike.

One version of what some pollsters call the "dove" question reads: "The United Nations has passed a resolution authorizing the use of military force against Iraq if they do not withdraw their troops from Kuwait by Jan. 15. If Iraq does not withdraw from Kuwait by then, do you think the United States should start military actions against Iraq, or should the United States wait longer to see if the trade embargo and other economic sanctions work?"

In the test poll, 49 percent favored beginning military action if Iraq was not out of Kuwait by Tuesday, while 47 percent wanted to wait.

So which is it: Do 62 percent or 49 percent of all Americans favor the use of force in the Middle East?

The apparent answer is that both are accurate because they appear to measure subtly different attitudes. If anything, the "dove" question may produce the more belligerent result.

That's because the dove question offers a choice between immediate military action when the deadline expires and continued reliance on sanctions. Read that way -- as many who favor instant war have done -- the result suggests about half the country wants war on Tuesday or soon thereafter if Iraq isn't out of Kuwait.

In contrast, following up the reputed hawk question with one asking how long the United States should wait after Tuesday before going to war produces a comparatively dove-like response: Just 18 percent of those questioned said they wanted the conflict to begin immediately.

Another 28 percent would support going to war within a month. When the numbers are added together, they suggest that about 46 percent of the American people favor war within a month of deadline day, or almost exactly the same percentage of those asked the dove question who favored starting military actions immediately.

The test also showed that 81 percent of everyone asked the dove question would support some form of military action if Iraq hasn't withdrawn within three months. That's significantly more than the 62 percent who favored going to war at any time after Jan. 15, as measured by the hawk question.

So now it's clear enough: The hawk question is the dove question, and the dove question is the hawk.

Probably not. In fact, some experts who briefly reviewed the test data suggested that both questions probably are telling the same story, the differences in the numbers notwithstanding.

"Those numbers seem to me to be about the same," said John Mueller, a professor of political science at the University of Rochester and an expert on public opinion and war. "The difference between using the word 'war' in the first question and 'military actions' in the second probably accounts for the difference in the numbers."

"When you do the breakdowns and when you follow up the sanctions question it indicates that a substantial majority, on the order of two-thirds, entertains the use of military force to get Saddam Hussein out of Kuwait," said Thomas Mann, director of governmental studies at the Brookings Institution.

So one problem for policy-makers is solved. A fairly large majority would accept military conflict reasonably soon after the deadline expires, though probably not on deadline day.

That may be the correct answer now -- but not once the shooting actually starts. Surveys done by The Post and ABC News and others suggest that support for war falls quickly when people are confronted with the reality of combat.

In fact, a recent Post-ABC survey and one by USA Today both found that initial support for war after Jan. 15 dropped from well over half to just over 40 percent when respondents were asked if they would support such a war if it meant that 1,000 American troops were to die.

Mann and Mueller say they believe Bush has a clear opportunity to lead public opinion rather than be driven by it. They say the polls suggest most supporters of the use of military force retain doubts about going to war. At the same time, surveys also show that most opponents of military action strongly disapprove of Iraq's occupation of Kuwait.

The result is that a broad range of policy options, including relatively quick military action or continued use of sanctions, likely would be initially acceptable to most Americans.

"The public is there to be led by its political leaders and by events," Mann said. "The president and Congress are not greatly constrained by public opinion, they are very much in a position to shape it."

That's precisely what happened just over a year ago in Panama. A Gallup poll in October 1989 asked a national sampling of adults whether U.S. military forces should "invade Panama and overthrow {military strongman Manuel Antonio} Noriega." Only 26 percent supported committing troops.

Just two months later, Bush did precisely that -- and 80 percent of those surveyed by Gallup after the invasion supported the decision.

When analyzing the results of public opinion polls, first impressions may be misleading.

The first pair of questions produce a seemingly larger share of support for military action than the second set of questions.

Or do they?

In fact, the follow-up question in the second set suggests that just the opposite may be the case. The second set of questions produces a significantly larger percentage of persons who favor immediate military action and a somewhat larger share who favor military action within three months of the Jan. 15 deadline if Iraq doesn't withdraw from Kuwait.

Q. As you may know, the U.N. Security Council has authorized the use of force against Iraq if it doesn't withdraw from Kuwait by Jan. 15. If Iraq does not withdraw from Kuwait, should the United States go to war against Iraq to force it out of Kuwait at some point after Jan. 15 or not?

Go to war sometime after Jan. 15 62%

No, do not go to war 32

Q. How long after Jan. 15 should the United States wait for Iraq to withdraw from Kuwait before going to war to force it out?

Do not favor war at any point 32%

Immediately 18

Less than one month 28

1-3 months 8

4 months or longer 2

Estimated percentage of Americans who support going to war immediately after the deadline expires if Iraq does not withdraw: 18 percent

Estimated percentage of Americans who support going to war within three months if Iraq does not withdraw: 54 percent

Q. The United Nations has passed a resolution authorizing the use of military force against Iraq if they do not withdraw their troops from Kuwait by Jan. 15.

If Iraq does not withdraw from Kuwait by then, do you think the United States should start military actions against Iraq, or should the United States wait longer to see if the trade embargo and economic sanctions work?

U.S. should start military actions

if no withdrawal by Jan. 15 49%

U.S. should wait long to see if embargo

and sanctions work 47

Q. How long after Jan. 15 should the United States wait for Iraq to withdraw from Kuwait before going to war to force it out?

U.S. should start military actions

if no withdrawal by Jan. 15 49%

Percentage who support sanctions but would wait:

Less than a month 15%

1-3 months 17

4 months or longer 9

Estimated percentage of Americans who support going to war immediately after the deadline expires if Iraq does not withdraw: 49 percent

Estimated percentage of Americans who support going to war within three months if Iraq does not withdraw: 81 percent

NOTE: Percent who expressed no opinion not shown.

Figures are based on interviews with 1,011 randomly selected adults interviewed by telephone Jan. 9-12. The total sample was divided in half, with each half asked a different sent of questions. Margin of sampling error for the overall results is plus or minus 5 percentage points. Interviewing was conducted by International Communications Research of Media, Pa.