A robust television ad campaign is a critical element in winning an election, which explains why the Obama and Romney campaigns and their parties and allies are in the midst of unleashing the most sustained and concentrated blitz of 30-second commercials in American political history.

But for all of the cash thrown at presidential TV ads — perhaps more than $1 billion between now and November — their impact has historically been relatively small in swaying large swaths of voters in the general election.

"The most fundamental point about political advertising is that it matters at the margins," said Erika Franklin Fowler, director of the Wesleyan Media Project, which tracks and analyzes campaign ads. "It might help in a close election," but factors such as the state of the economy and partisan identification are much more influential, she said.

But if that is so, it raises a fundamental question: Why do presidential campaigns devote such resources to an effort that allegedly yields so little?

The answer may be that they can't afford not to.

The risk of losing even a fraction of the vote by being outspent on advertising largely drives the relentless fundraising and spending of both parties, says Travis Ridout, a political scientist at Washington State University. "You don't want to be left behind," Ridout said. "No one is willing to unilaterally disarm."

"Ask Al Gore if a small advantage matters," said Ken Goldstein, president of the ad-tracking firm Campaign Media Analysis Group of Washington, referring to Gore's 540-vote loss in Florida in 2000.

Advertising will consume most of the millions of dollars raised by the candidates, their parties and the independent super PACs that are the wild card in this year's campaign. Goldstein estimates that at least $1 billion will be spent on TV spots in just a dozen or so swing states that could determine the election. Others guess more — lots more: Bill Burton, a former aide to President Barack Obama, told New York magazine last month that the total raised this year could top $2.6 billion, with much of that total ticketed for advertising.

Fundraising figures released by Obama and Republican contender Mitt Romney last week haven't challenged the most-expensive-election-ever narrative. Romney and his party raised more than $76 million last month, outdrawing Obama and Democrats, who hauled in $60 million. Romney is also getting help from Republican-leaning super PACs that have raised far more than their Democratic-leaning counterparts. (Advertising strategists for Romney and Obama declined to comment for this article).

The notion that all that TV money will buy relatively little is almost as old as TV itself. Dozens of academic examinations since the 1940s have shown that TV ads have limited persuasive effects on voters in general elections, often accounting for no more than a 1 or 2 percent difference, and often considerably less.

What's more, outspending a rival on advertising in a general-election campaign doesn't guarantee anything; John Kerry and his Democratic allies ran almost 200,000 more ads than George W. Bush did in 2004 and lost in a close election. On the other hand, Obama had a narrower advertising advantage over Sen. John McCain in 2008 and won relatively easily.

Political operatives and campaign media advisers don't dispute the general notion of limited impact. But only a few openly question the proliferating ad wars.

"More money is spent on television advertising than any other element of presidential campaigns, and yet it provides the worst return on investment," said Mark McKinnon, the chief media adviser for Bush's presidential campaigns in 2000 and 2004 and an early adviser to McCain in 2008. "I think you could take 90 percent of the spending on television ads and flush it and still have the same net impact."

One illuminating illustration came in a study of the 2008 race by researchers at Washington State and Bowdoin College.

The study compared the vote totals of the candidates in counties in "battleground" states with their totals in counties just across the border in safely red or blue states. People in the latter counties saw the same ads as voters in the battleground counties but weren't subjected to other kinds of campaign messages and activities, such as phone calls, mail or visits by home campaign workers.

On average, a candidate who dominated advertising in a region raised his vote total by just 0.5 percent in the control group of voters in non-battleground counties, Ridout and fellow researcher Michael Franz found. They concluded that advertising "is responsible for only a small portion of the final result."

A University of Texas study of the 2000 and 2004 races found that Bush improved his standing by just 0.1 percent under virtually similar circumstances.

This is not to say that an advertising blitz can't sometimes play a definitive role in some races — or in certain stages of races. Advertising, for example, appears to play a far bigger role in primary and local elections where the candidates aren't as well known to voters and where a well-funded campaign can overwhelm poorer rivals. Gregg Phillips experienced the phenomenon firsthand.

Phillips managed Winning Our Future, the super PAC that backed former House Speaker Newt Gingrich in this year's Republican primaries. He points out that Romney and the independent group backing him all but buried his GOP opponents, particularly Gingrich, in the Florida primary in late January. Romney and his backers spent more than $11 million on TV ads, compared with less than $2 million by Gingrich. "Newt said, 'I threw the kitchen sink at him, but Romney threw a bigger sink at me,' " Phillips said.

Phillips and other strategists acknowledge that ads don't work the same way in a general election. By the summer and fall of an election year, the major candidates are usually well known, and TV advertising is just one way people form perceptions about them. There are many other sources of information: constant news coverage, campaign rallies, phone calls and home visits by campaign workers, as well as input from friends, family and co-workers.

Campaign media advisers say that waging a modern presidential ad campaign is like swinging a giant hammer at a tiny nail. Almost all of the commercials are aimed at a minority of a minority of voters: the few who haven't yet made up their minds in a handful of battleground states. A Gallup poll of likely voters in 12 such swing states in early May found that only 7 percent described themselves as undecided. And the number shrinks the closer it gets to Election Day.

TV ads affect "a narrow sliver of voters in a very narrow segment of the country," says Steve McMahon, who directed former Vermont Gov. Howard Dean's primary advertising in 2004 and the Democratic National Committee's general-election ads in 2008. "Those people receive information in lots of ways, and ads are just one way."

But McMahon says advertising is a key part of a candidate's communications strategy. "All those [messages] create an orchestra. If your campaign is working, all the pieces come together," he said.

McMahon says the great danger for any campaign is to lose your "share of voice" — the ability to match or exceed an opponent's advertising. Such imbalances could crop up more frequently in this election because of unlimited spending by super PACs.

While the bulk of 30-second ads will be directed to a limited number of cities, broadcast stations and programs, the candidates have better ways than ever to target different sets of voters. Local cable stations make it possible to reach narrow subsets of voters, such as Latinos watching Telemundo in Tampa or African Americans watching BET in northern Ohio. Although the Internet is a growing source of targeted campaign advertising, it remains a distant second to traditional broadcasting, said Wesleyan Media Project's Fowler. In part, this is because TV ads are passive, whereas Internet ads require a viewer to click and play — a potential barrier, she notes.

Yet there's little evidence that a memorable TV ad or series of ads can on their own change the dynamic of a presidential race. One of the most famous — Lyndon Johnson's "Daisy" commercial in 1964, in which Johnson implied that his Republican opponent, Sen. Barry Goldwater, would put the world at risk of nuclear war — didn't change Johnson's standing in the polls from the time the ad aired in early September until Election Day. Similarly, the four "Swift Boat" ads that attacked Kerry's war record and integrity in August 2004 had little prolonged impact. Kerry's poll numbers rose and fell before the ads began — and rose and fell right up until the election.