As a kid, I dreamed about driving a Formula One racing car. Having found politics a less perilous route to those adrenaline rushes, I never thought the phrase “Danger, Do Not Attempt This at Home” would apply to my occupation.

Yet it appears that crosstabs in untrained hands can produce some dangerously inaccurate interpretations.

Immediately after the Potomac Primaries, some pundits argued that Barack Obama had fundamentally altered the demographic dynamics of the primary race, breaching the Clinton constituency by winning women and white men.

While Obama clearly walloped Clinton in D.C., Maryland and Virginia, he did so mainly by keeping his own large and robust coalition in tact.

The pundits’ problems begin with misstating the nature of the Clinton coalition in previous primaries. Neither “women” nor “white men” have been core elements of the Clinton constituency. She lost women in a welter of states, from Alabama and Georgia to Missouri and Utah. Clinton lost white men across the Super Tuesday states, including in places like California, Massachusetts and New Mexico.

Properly specifying the constituents of the coalition makes a real difference. White women, not women generally, have been the central Clinton demographic, and Clinton won white women by 18 points in Maryland and by a much narrower six points in Virginia.

Looking simply at “women” is a mistake because African-American women are a key part of Obama’s constituency, and when African-American women constitute 40 percent of the female electorate, as they did in Maryland, or a third, as they did in Virginia, Clinton is likely to lose the broader category even if she holds her own with white women, the segment that has been her base.

Similarly, the correct specification of Clinton’s male demographic is not “white men,” but non-college-educated white men. In Maryland, she won white men without a college diploma by 26 points and eked out a 3-point victory with this group in Virginia. Obama enjoys a substantial edge with college-educated white men. So the mix of college- and non-college-educated white men matters. In Virginia, 70 percent of white men had a college degree, compared to just 35 percent in Alabama on Super Tuesday.

While Clinton held her true base in both states, the difference between Maryland and Virginia remains — in Maryland she won her voters handily, but in the Old Dominion, just barely. Why the dramatic difference?

In many ways the demography of these states is similar, with nearly identical numbers of liberals, marrieds and voters under 30 and over 65.

Perhaps the most important difference was the partisan composition of the electorate. In Maryland, 84 percent of Democratic primary voters identified themselves as Democrats, compared to 70 percent in Virginia. The rules matter, producing different results. Virginia is an open primary, with independents, and even Republicans, free to participate, whereas Maryland is closed, restricting participation to registered Democrats.

That difference leads to one more important conclusion about the candidates’ bases. Sen. Clinton wins white Democrats. She won that segment by 12 points in Virginia and by a remarkably similar 13 points in Maryland, while Obama triumphed among independents and Republicans in Virginia by a stunning 42-point margin.

Higher proportions of blacks among women and more college grads among white men in both states, along with more non-Democrats in Virginia, created the illusion of a disintegrating Clinton coalition.

Looking ahead, though, Wisconsin, Ohio and Texas are not closed primaries. Experience suggests those systems should advantage Obama, even while the basic demography may favor Clinton.

So while the excitement continues, handle those crosstabs with care.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the majority leaders of both the House and Senate.