Tuesday evening, as the polls were closing in most of the country, a young black Milwaukeean named Andre Douglas tweeted that only 4.7% of blacks had voted in his home state. It’s still not quite clear where he got this incorrect information — my best guess is that he or someone he knew misread a report that only 4.7% of Wisconsin voters had been African American — but the stat spread like wildfire on Twitter that night, quickly losing its connection to Wisconsin and becoming a “fact” about the country as a whole.

The claim was greeted with skepticism by many, but the doubters were overwhelmed by the believers. The story was repeated so often that it — combined with the news that the US Senate would be losing its only black member — turned the term “African Americans” into a Twitter trending topic for much of Wednesday.

Why did this false claim, first made by a guy with fewer than three hundred followers, blow up so big? Three reasons, I think…

First, it couldn’t be easily debunked.

You might think that how many African Americans voted in this week’s election would be a fact you could quickly Google, but if you did, you’d be wrong. It’s not a stat that appeared in any media coverage of the election, and — as I discovered when a friend on Twitter asked if anyone knew the true number — calculating involves tracking down data from a bunch of different sources.

If folks had been able to research the claim and quickly post a link to a more accurate number, they would have done so. But they couldn’t, so they didn’t. (Almost immediately after I posted my own estimate of black turnout — something like 34% — yesterday evening, my site started lighting up with the results of Google searches on the stat.)

Second, midterm election turnout is an obscure subject.

What percentage of Americans turn out to vote in a typical midterm election? How much less (or more) likely are blacks to vote than whites? How much did voter turnout rise in 2008, and what has happened to midterm voting numbers after elections similar to 2008’s?

I could have given you pretty good answers to most of these questions yesterday, but I’m a historian of American social movements who specializes in issues of race and electoral politics. And even I would have been guessing about some of them.

But unless you have this data — unless you’ve got at least a solid hunch about what off-year turnout numbers should look like — you’re not going to be able to form your own opinion about whether the 4.7% stat makes sense.

If you don’t know whether overall turnout in 2010 was 20% or 40% or 60% (and most Americans don’t) and you don’t know whether blacks typically turn out 90% as often as whites or 60% or 30% as often (and again, most Americans don’t), then you’re not going to be able to assess the stat’s plausibility on your own.

Third, it reinforced a belief that was already there.

For whatever reason, the false stat circulated almost exclusively among African American Twitter users. For those who shared it, it served to reinforce beliefs that they already held about low levels of black civic engagement — and in many cases a cynicism about the shallowness of black support for Barack Obama’s presidential campaign.

Interestingly, though, those beliefs are themselves false — or at least exaggerated. It’s true that blacks tend to vote at lower levels than whites, but the differences aren’t dramatic, as figures from recent elections show.

In 2004, when John Kerry ran for president against George W Bush, 65.4% of white citizens voted in the November election, as opposed to 60% of blacks. In the midterm elections of 2006, the figures were 49.7% for whites and 41% for blacks. And in 2008, with Obama on the ballot, blacks were more likely to vote than whites — by a margin of 64.7% to 64.4%.

The gap between white and black turnout does exist, but it’s not huge — in 2006, the last midterm election for which we have data, it was about the same as the gap between 35-44 year olds (45.5%) and 45-54 year olds (53.8%), and considerably smaller than the gap between married people (56.2%) and divorced people (42.9%).