Can Anyone Explain How Alvin Greene Actually Won?

Does anything feel more proudly patriotic than voting? You stride into your local polling place, past fellow citizens smoothing "I Voted!" stickers to their chests. You snap the booth's curtain shut and turn to the ballot, ready to execute what Samuel Adams called "the most solemn trust in human society," only to realize:

Crap. Who are these guys?

Academics call it "low-information voting": the psychology of what goes through our heads when confronted with a ballot full of names or issues we know very little about. Long established in political-science circles, the topic is newly relevant with the surprise election last week of Democrat Alvin Greene in the South Carolina Senate primary. It was a baffling 18-point victory for an unknown, unemployed, inarticulate Army veteran who gave no speeches, distributed no literature, and won no endorsements in a noncampaign against experienced pol Victor Rawl. How did it happen? How did 59 percent of 169,542 South Carolinians decide to cast a vote for a man they'd never heard of?

Though there have been allegations of fraud, nothing has been proved. But there are plenty of theories and while none is significant enough by itself to explain an 18-point margin of victory, taken together they may begin to explain Greene's surprise win.

The most popular hypothesis so far is that Greene's name simply appeared before his opponents. Known as the "ballot-position effect" or the "name-order effect," the theory holds that being listed first increases a candidate's performance by an average of 2.5 percentage points. That was the finding of a 1998 study by political scientists Joanne Miller and Jon A. Krosnick, who analyzed the returns from 118 races in Ohio from 1992, finding the name-order effect in nearly half of them.

"The psychology here is the 'primacy effect,' " says Miller, now an associate professor of political science at the University of Minnesota. "When people have to make a choice from a visual list, the theory is that they start at the top, consider the first choice, and if they can think of a reason, just stop there, and not necessarily go to the second choice, the third choice, and so forth." In a two-person race like the Greene-Rawl contest in South Carolina, the close-to-three-point bump for one candidate becomes a six-point spread.

That still leaves 12 points of Greene's margin of victory to account for. Jennifer A. Steen, an assistant professor of political science at Arizona State, thinks she can explain some of it. In a 2004 study with Jonathan GS Koppell, she found that the name-order effect increased as voters get farther down the ballot to races they know little about. In other words, the ordering bump is even more pronounced in what are functionally no-information elections. That's a fair description of the South Carolina Senate primary, as important as a Senate seat is. Republican incumbent Jim DeMint is expected to easily win this November, and Democrats didn't recruit anyone remotely impressive to challenge him. Though Rawl is better known than the total stranger Greene, it's not by much: he had only a 4 percent favorable rating among Democrats in one pre-primary poll.

"That's what makes this more like a primary for something like dogcatcher instead of Senate," says Steen. "It's anomalous even in the situation where the incumbent is expected to run away with it."

Given a near-total lack of data about the candidates, voters begin to exhibit some techniques whose rigor would not impress the Founding Fathers. One, says Stanford's Krosnick, is known as the "mere exposure effect"—the notion that the more a person is exposed to a certain stimulus, the more he or she is inclined to like it. "Greene" is a more common word than "Rawl." Another factor is the "name-letter effect," which holds that people prefer names that share their own initials. There are more A.G.s in South Carolina than V.R.s.

Do American voters really treat their votes so cavalierly? Actually, voters are fairly responsible when it comes to leaving blank those down-ballot races about which they have no clue. There's even a political term of art for it, "roll-off." But the roll-off in the South Carolina Democratic primary was just 10 percent between the first contest on the ballot, the gubernatorial race, and the third contest, the Greene-Rawl Senate match. That's a curiously low figure, national political operatives say, even for a primary election, in which voters tend to be better informed than November voters.

Why would that be? The likeliest explanation is that South Carolina uses electronic voting machines statewide. On a traditional paper ballot, voters can simply leave blank those races about which they are uninformed. The iVotronic machines used across the Palmetto State, however, remind voters if they have left any contests blank. On a final review screen, the candidates a person has selected are highlighted in yellow with a big green check; an office for which no choice has been made gets a white background, with "No selection made" in red letters. "It does remind you before you cast your ballot, 'hey, you didn't vote in this office,' " says Chris Whitmire, a spokesman for the state election commission. Could voters interpret that to mean that an incomplete ballot is an invalid one, and feel compelled to revisit races they know utterly nothing about?

Whitmire says he is unaware of any such complaints. But a 1995 study by Stephen M. Nichols and Gregory A. Strizek found that electronic voting machines "sharply attenuate roll-off—particularly in lower visibility contests." Nichols, now an associate professor of political science at California State University San Marcos, says that pretty much describes the Greene-Rawl race. He first encountered the issue in Ohio in 1992—the same elections studied by Miller and Krosnick—when the state was testing machines that blinked a red light next to unvoted contests. "It was sort of like having your mother standing over you, nagging you to finish voting," Nichols says. "I thought, man, people must be voting in county auditor, treasurer, tax collector-type races just to turn the damn light off."

"This could exacerbate something like a primacy effect, where some people who would have ordinarily rolled off because they don't know anything feel like they have to finish, and feel they need to make a choice somehow," says Miller. "It's totally plausible," says Steen.

"Ballot-position effect" times "no-information voting," plus "mere exposure effect" and "name- letter effect," minus expected roll-off—does that equal an 18-point victory for a nobody like Alvin Greene? Maybe. Even the people who study this kind of thing are scratching their heads, and they'll never know for sure. The window for collecting reliable data passed with Election Day. "There's no way to [poll voters] after the fact, now that everyone has read all the publicity. The results are not going to bear any relationship to the actual election tally," says Steen. "I think it's mostly gonna be a mystery."