Over the airwaves, the suspect charges fly: John Kerry has voted to raise taxes 98 times in his 20-year Senate career. George W. Bush imposed the biggest Medicare premium increase in history. Kerry is proposing a government-run health care plan. Bush endorses the outsourcing of jobs. Kerry didn't deserve his Vietnam war medals. The war in Iraq has cost U.S. taxpayers $200 billion.

Totally true, or completely false? Well, not entirely one or the other, as it happens. In the modern political ad wars, "truth" is a relative commodity, and admakers tweak our perception of it to gain maximum advantage for their clients. Perhaps, if you live in the current election battleground, political attack ads like the ones I've just described have already subtly influenced your views of Bush and Kerry, cementing impressions, raising questions or sowing doubts.

This year, complaints about advertising distortions seem louder than ever. The media generally sling the word "negative" at attack ads and complain about them, even when they're few in number and the specifics are largely accurate. But this time around, our research shows, the grousing reporters and pundits are on to something. Questionable claims in Republican attack ads abound this year, in part because of the sheer volume of Bush attacks on Kerry's record. And the barrage of attacks on Bush from independent "527" groups supporting Kerry, as well as the Swift boat veterans' claims against Kerry, have multiplied the suspect assertions floating around this campaign season.

Voters caught in the crossfire might well wonder, how do ads deceive or mislead us? When they do, does it matter? Do they work, and why? And who or what protects us from their wiles?

The candidates and their supporters will spend about a billion dollars to produce and distribute ads this year. To the campaigns, the expense is worth it. Though ads alone can't determine the outcome of an election, they can be a decisive factor. And attack ads are especially effective. Surveys repeatedly show that negative ads register with voters more quickly than positive information and are more readily recalled as well.

But are such ads of any value to voters? It depends. Yes, they sometimes resort to ambiguities and even deception to create an impression, but many also hit on the central themes of a campaign. In particular, ads that contrast the views of both candidates in a single spot often provide voters with policy specifics and accurate information that will forecast how a given candidate will govern.

Political ads don't usually spread outright lies. Many of their troubling statements contain an element of truth. It's true that Kerry has voted to raise taxes -- but not as often or as much as the Republican ads making that claim would have us believe. Yes, jobs have been lost during the Bush administration -- but not as many as Democrats claim. Contrary to a Democratic ad, the war in Iraq has not yet cost $200 billion, although it may in the future.

Such deception is lamentable, even unnecessary -- after all, simply stating the truth would make the same points -- but campaigns exaggerate or oversimplify the truth because it's effective. It gets the voters' attention, builds on widespread assumptions -- Democrats are tax-and-spenders, Republicans are pro-business, pro-defense and against the little guy -- and it sticks with people unless it's solidly and conspicuously debunked.

Ads often sin by omission as well. It's true, as Republican ads note, that Kerry has opposed funding some weapons systems. What they don't mention is that both President George H.W. Bush and his secretary of defense, Dick Cheney, opposed some of the same weapons. Democrats may try to compare current job losses with those of the Great Depression, but the unemployment rate under Bush today is actually about what it was when Clinton was campaigning for a second term. Some deceptions don't exist in the ads per se but in the meaning the audience takes from them. Admakers know that evocative images short-circuit our ability to analyze them critically. That's what's going on in a MoveOn PAC ad focusing on Kerry's support for extending the assault weapons ban. As an AK-47 rifle materializes on-screen, the announcer intones, "This is an assault weapon." True. "It can fire up to 300 rounds a minute." Also true, for the fully automatic version. "In the hands of terrorists it could kill hundreds." True again. But contrary to the inference the ad invites, the fully automatic weapon it displays and describes was not affected by the assault weapons ban, which applied to semiautomatic weapons only. Access to fully automatic weapons has been restricted since 1934 by the National Firearms Act.

As polls repeatedly report, the public hates negative ads. But refine the question, and the answer changes. An Annenberg survey suggests that voters accept civil, accurate, issue-based attack ads. Such ads not only do exist, they also work. One of Ronald Reagan's effective ads in 1980 asked, "Can we afford four more years of broken promises? In 1976, Jimmy Carter promised to hold inflation to 4 percent. Today it is 14 percent. He promised to fight unemployment. But today there are 8.5 million Americans out of work." All true, and all presented in a factual way, without any twist to make the point sharper. Voters also report that they prefer comparative or contrast ads that provide a reason to vote against one candidate while also justifying a vote for the candidate running the ad.

Attack ads have a bad reputation because reporters and the public believe that a candidate generally uses them to spread deception about his rival. Yet in past presidential elections, more distortions appeared in ads bragging about the candidate running the ad than in those blasting the opponent. Although the jury is still out on 2004, cases of self-inflation include Kerry's advertised claim that he "cast a decisive vote that created 20 million new jobs" and Bush's suggestion that he (not Ronald Reagan) provided "the largest tax relief in history."

When it comes to advertising in general and attack ads in particular, of course, money matters. Our 2000 National Annenberg Election Survey found that by outspending Al Gore and buying attack ads questioning the vice president's credibility in the final weeks of the last campaign, Bush blunted Gore's accurate charge that the GOP's Social Security plan would fall short by $1 trillion of doing what it promised. Had the Republicans not gained that last-minute advantage in the battleground states, Gore might have won the electoral college as well as the popular vote.

Similarly, our April survey of voters in the ad-saturated battleground states found that 61 percent had been gulled by the false claim that Bush "favors sending American jobs overseas," 56 percent by the notion that Kerry "voted for higher taxes 350 times" and 72 percent by the assertion that 3 million jobs have been lost in the Bush presidency (at the depth of the slump, the number was 2.7 million).

In political advertising, truth only has an advantage when a credible source makes it known. That's chiefly the media's job. When reporters hold a candidate accountable for the campaigning done in his name, it works. For instance, during the 2000 primary campaign, when Bush was asked whether he actually believed his ad's claim that his rival, Sen. John McCain, opposed funding breast cancer research, he admitted that he didn't.

When a candidate ducks questions, the journalistic fact check is another line of voter defense. In the 1992 Georgia presidential primary, viewers penalized GOP hopeful Patrick Buchanan for a deceptive attack against incumbent President George H.W. Bush after the ad was widely criticized on the national and local news. A CNN-Gallup poll showed that 23 percent of voters reported that the ad had increased support not for Buchanan, but for Bush.

When presidential ads are bombarded with critiques from news outlets, their sponsors and "unaffiliated" supporters usually change the challenged claim. The Bush campaign did this when it reduced its claim that Kerry had voted for tax increases from 350 times to 98. A MoveOn PAC ad did the same in revising the cost of the Iraq war downward from $200 billion to $150 billion in a critiqued Kerry ad.

But the road to accuracy is one of a thousand steps, and hundreds remain to be taken. As Annenberg FactCheck.org shows, the Republicans' figure of 98 tax-increase votes by Kerry is still an exaggeration, and $150 million inflates the current cost of the Iraq war, which is still under $120 billion, according to the Office of Management and Budget. Confident that voters will continue to find the claims plausible and convinced that frequent airing will drown out news corrections, those on both sides continue to repeat some misleading claims.

Yet a focus on the distortions in ads and the failure of news organizations to keep the worst tendencies in check can obscure the fact that campaign ads do include useful information. In fact, those who want substance in ads should applaud attack and contrast ads. Our research shows that in every year from 1952 to 2000, there have been more policy specifics in contrast ads than in so-called positive ads.

In 2000, for example, the Bush contrast ads forecast a prescription drug benefit, tax cuts, educational reforms and private investment accounts for Social Security. As president, Bush kept the first three promises and has renewed the fourth. This year, among other things, the candidate ads feature alternative plans to increase access to health insurance. This probably ensures that the next president, whoever he is, will work to increase the number of insured.

In a model election, candidates would stick to the high road, and their declarations and ads on Social Security, health care and Iraq would be scrutinized by news organizations that weren't sidetracked by the political horse race, hurricanes and the fate of Dan Rather. Our country has conducted campaigns worthy of emulation, notably 1960 (Kennedy vs. Nixon) and 1980 (Reagan vs. Carter). It has also suffered through others that serve as negative examples, especially 1964 (Johnson vs. Goldwater) and 1988 (Bush vs. Dukakis). It's not too late for the candidates, campaign consultants and reporters to move this election in the direction of the former and away from the latter.

Kathleen Hall Jamieson, professor of communication and director of the Annenberg Public Policy Center of the University of Pennsylvania, is the co-author of "The 2000 Presidential Election and the Foundations of Party Politics" (Cambridge University Press).

Fuzzy facts: This automatic weapon looks frightening, but it's already prohibited under a 1934 law, the author points out.