According to a recent Zogby International poll, 81 percent of voters think "crime and violence" should be either a "very high" or the "highest" priority for the president. Yet according to the U.S. Constitution, the president has no authority over the kinds of crimes--car break-ins, burglaries, muggings, rape--that worry the average American.

Do voters think the president ought to patrol their neighborhoods? Do they expect him to stand guard over their homes or walk them home late at night?

More likely, declaring crime a "presidential priority" is simply a way of expressing concern about it. This sort of locution has troubling implications for anyone old-fashioned enough to believe in a federal government of strictly limited powers. But it also raises the question of how much weight to give "public opinion" when it is ill-informed or, as in this case, completely divorced from reality.

Elsewhere in the same poll, the respondents were asked to choose between Candidate A, who "says there should be more gun control laws," and Candidate B, who "says that instead of new laws there should be mandatory jail time for those who commit a crime with a handgun." By a margin of three to one (72 percent to 24 percent), voters favored Candidate A.

But suppose you asked these same people to describe existing gun control laws at the federal level and in their own states and cities. Very few of them would have more than a vague idea of what these laws say.

So what does their view that we need more gun control really mean? Like the assertion that crime should be a high presidential priority, it may simply be an expression of fear (in itself hard to reconcile with the facts, since violent crime has been declining steadily in recent years).

Suppose the gun control question were preceded by a statement like this one: "Despite thousands of federal, state, and local gun control laws, there is no convincing evidence that such legislation reduces crime." Suddenly, Candidate A might look less attractive.

Another item in the Zogby poll suggests the importance of slanting the question just so. Regarding the gun control legislation that Congress has been considering, the respondents were asked to choose between two statements:

A. "These laws will prohibit acts of gun violence like what happened recently in Atlanta, Columbine High School, and Los Angeles."

B. "These laws are cosmetic and Congress is just reacting to the press, which is blaming gun owners for the actions of criminals and the mentally ill."

Not surprisingly, a majority (55 percent) agreed with Statement B. The astonishing thing is that Statement A, which seems to suggest that prohibiting bad acts is enough to prevent them, attracted support from nearly a third of the respondents.

I happen to agree with the majority in this case, but it’s clear that the pollsters made no attempt to state the anti-gun position in a plausible way. A fairer gloss would have begun with a qualifying phrase like, "Although no legislation can prevent all acts of violence..."

Medical marijuana is another issue where the phrasing of the question can have a decisive impact on poll results. Surveys generally find that most Americans think patients who can benefit from marijuana should be able to obtain it. In a March 1999 Gallup poll, for example, 73 percent of respondents said they supported "making marijuana legally available for doctors to prescribe in order to reduce pain and suffering."

By contrast, in a survey conducted the previous month by the Family Research Council, a plurality of 48 percent opposed making marijuana available as a medicine. Unlike the other pollsters, the FRC introduced its question by telling respondents that marijuana smoke contains more carcinogens than tobacco smoke and that a synthetic form of THC, marijuana’s main active ingredient, is available by prescription.

The relevance of these points is questionable: The cancer risk applies only to people who smoke heavily for many years, and patients often prefer smoked marijuana to THC capsules because it’s easier to ingest, it takes effect more quickly, the dosage is easier to control, and the psychoactive effects are less disturbing.

But the FRC presumably would argue that the support for medical marijuana found in the other polls is based on ignorance. Its attempt to remedy that problem is yet another illustration of the fine line between measuring public opinion and shaping it.