If the United States were a Third World country, our Nov. 2 election would not pass certification by international monitors. As former President Carter has explained on National Public Radio, we lack a central, nonpartisan election commission to guarantee fair and equal treatment of all voters nationwide, our candidates do not receive free and equal access to the media to deliver their message, voting procedures are not uniform throughout the county, and there is not a "paper trail" available in all cases to guarantee an honest recount where called for.

To insure against fraud in the counting and reporting of election results, international monitors depend on the same kind of election-day exit polls as were conducted in the United States. Unlike earlier polls that attempt to predict a future election outcome, which are therefore subject to all manner of potential errors, exit polls estimate the characteristics of a population which has already voted. It is as if you had a huge jar of M&Ms and you took out several handfuls at random, counted the proportion of each color in your sample, and knowing the total number of M&Ms in the jar, used these results to estimate how many were red, brown, green or yellow in the jar as a whole. If your sample was reasonably large and randomly drawn from the jar, it would estimate these totals with a high degree of accuracy.

In our own recent presidential election, exit polls were conducted nationwide for the media by two of the world's most respected professional exit-polling firms: Edison Media Research and Mitofsky International. Pollsters were sent to carefully selected, representative polling locations throughout each state. They then interviewed about every fifth voter emerging from the polling place during random periods throughout the day. Total samples from each state were large -- about 2,000 or more voters -- and the error of estimate was small -- plus or minus less than half of 1 percent in 99 cases out of 100.

By agreement among the networks, the results of these exit polls were not reported to the public on election day, so as not to influence the ongoing voting process or lead to embarrassing "premature" calling of outcomes by the networks, as happened for Florida in 2000. But they were shared with -- and believed by -- campaign officials and by the candidates themselves, and they were widely reported over the Internet. As we now know, on the basis of these exit polls, Kerry was expected to win.

Then the "actual" tallies began to pour in.

This was the "November surprise." In state after state, Kerry saw his expected lead shrink or vanish. And when he lost Ohio -- which exit polls estimated he would win by 4.5 percent -- he "lost" the election. According to Steven Freeman, who teaches research methods at the University of Pennsylvania, for 10 exit polls among the 11 battleground states he analyzed to be this far off as a result of random error, particularly when all discrepancies favored Bush, is essentially impossible.

As officials testified this week at a forum called by Rep. John Conyers Jr., D-Mich., and for an investigation by Congress' General Accounting Office, electronic voting theft is incredibly easy. There have also been widespread reports of election irregularities -- more than 38,000 nationwide at last count, according to Verified Voting Foundation's election-incident reporting system. Most of those irregularities appear to favor Republicans.

If this were an election taking place in a Third World dictatorship, or a former part of the Soviet Union (Georgia and Ukraine, for recent examples) people would be in the streets screaming "fraud" and demanding the president's resignation.

Democratic pundits have been wringing their hands, trying to figure out the best tactics for future victory. The answer is simple: Make sure every eligible voter gets a chance to vote, and that every vote gets recorded, counted and accurately reported -- and that a secure paper trail exists to ensure the validity of any required recount.

Suspicions of election fraud undermine the very foundation of our democracy and need to be addressed.

Theodore D. Graves (tgraves@monitor.net) a retired professor of anthropology and social psychology, has taught research methods at the University of Colorado, UCLA and the University of Auckland in New Zealand.

© 2004 The San Francisco Chronicle