Visiting Scholar  Lawrence B. Lindsey

I'm a numbers guy, especially if they have to do with politics or economics. So even though I am a Republican, analyzing the Democratic primary results has become a great pastime. This is especially true when it comes to the Democrats' dilemma about how to handle Florida and Michigan--two states that broke party rules by holding their primaries before the allowed date, but which probably hold the key to the Democratic presidential nomination. To this numbers guy, the solution is pretty obvious from the data.

But the Democrats appear to be a party of lawyers. Only lawyers could have invented delegate selection rules as complicated and opaque as the ones the Democrats are struggling under. It also looks like only lawyers have a chance at the Democratic nomination. Harvard Law (Obama) and Yale Law (Clinton) candidates have survived, while University of North Carolina Law (Edwards), Syracuse Law (Biden), and the University of Louisville Law (Dodd) have been eliminated. And lawyers at the DNC Rules Committee will decide what happens next.

Still, sometimes lawyers call in numbers guys as expert witnesses.

The first question is whether Florida and Michigan voters acted like these primaries mattered, even though they knew the delegates they chose were not recognized by the national party. This can be discerned from turnout, and in the case of Florida the answer is yes.

Florida had a closed primary in which only registered Democrats could vote; turnout amounted to 46.7% of John Kerry's 2004 popular vote. The primary turnout relative to Kerry's 2004 vote in other closed primaries ranged from 39.8% in New York and 40.8% in Connecticut to 48% in Delaware, 49% in Arizona to 58.5% in Maryland. In other words, Florida Democrats acted as if their primary mattered just as much as other Democrats. By contrast, turnout in Michigan was only 23.7% of Kerry's 2004 vote, and it is an open primary. Michigan Democrats did not act like their primary mattered.

The second question is whether the two states' primary votes were skewed because of their timing, or whether they looked like what would have occurred had they happened on some "legal" day like Super Tuesday. A survey of exit polls from the primaries held so far shows patterns of voting by factors like age, gender, racial and ethnic identification, income, education and religion. This allows us to test whether the Florida and Michigan results looked the way they "should," based on how the voting occurred in other states.

Hillary Clinton beat Barack Obama by 17 points in Florida. If one takes the voting by age in large Super Tuesday states like California and New Jersey and applies it to the demographics of Florida, a predicted margin of 16 points emerges.

The similarities don't end there. For example, Jewish voters made up 9% of the Democratic electorate in Florida and New Jersey. Mrs. Clinton won this group by 32 points in Florida and 26 points in New Jersey. This is not surprising, since many Jewish residents of Florida emigrated from up north, and thus voted the same way their cousins, nieces, nephews and children did.

The statistical evidence strongly suggests that the outcome in Florida reflected what would have occurred had the state voted on Super Tuesday rather than one week earlier.

The voting in Michigan reflects many similarities to other states, but is far less conclusive. Sen. Obama's name was of course not on the Michigan ballot. Yet voters had the option of voting "uncommitted"--and the demographic evidence suggests they understood that voting "uncommitted" was a vote for Mr. Obama, or at least against Mrs. Clinton. California and New Jersey votes by age, where Mr. Obama was on the ballot, were almost exactly the same as in Michigan, where "uncommitted" was the alternative to Mrs. Clinton. A difference does emerge in the over-60 group, which gave Mrs. Clinton a 37-point margin in Michigan compared with 21 in California, 25 in Missouri and 28 in New Jersey. The average of those would have reduced her 15-point overall margin in Michigan to 12 points.

That difference in margin is virtually identical to the key difference between Michigan and other states: less of a racial gap. Among the 23% of Michigan Democrats who identified themselves as black, "uncommitted" beat Clinton by 38 points. Remember that Michigan voted four days before South Carolina, when the racial issue moved to the forefront. In South Carolina and the Super Tuesday states, Mr. Obama beat Mrs. Clinton by margins between 50 and 60 points. Had this happened in Michigan, Clinton's victory margin would have been 11 points instead of 15. Although unprovable without access to the actual polling questionnaires, the likelihood is that older black voters trended decisively to Mr. Obama after Michigan and South Carolina. That same conclusion also appears to be consistent with national polling.

In sum, the Michigan vote was flawed in ways the Florida vote was not. The most statistically valid conclusion would be that changes in voter attitudes in the second half of January would have produced a much narrower win for Mrs. Clinton of 10-12 points (not 15) had the state voted on Super Tuesday instead of Jan. 15. Still, Mrs. Clinton would almost certainly have won.

The behavior of Mrs. Clinton, who went to Michigan to lobby for a revote, and that of the Obama campaign, which worked to thwart a Michigan revote, indicate that both camps know this would be the outcome. Demographically Michigan looks almost identical to Ohio, which gave Clinton a 10-point victory.

Discussion among Democrats on how to deal with Florida and Michigan centers on three options. The first is not to seat them at all. Legally appropriate, but it would doubtless hurt the Democrats in both states in November--which may be why Republicans in the state legislatures found themselves as allies of Mr. Obama in working against a revote.

The second option would be to seat delegations that were evenly split between Mrs. Clinton and Mr. Obama. This would make the votes of 2.3 million Democrats irrelevant, while creating artificial representation for the states. It is very much like the 72 bonus delegates selected by party leaders to "represent" women, ethnic minorities, the gay and lesbian communities and the handicapped.

The third option would be to let the early primary votes stand, and select delegates according to the outcome. On a statistical basis, this is clearly the right result for Florida. The easiest solution for Michigan is to simply award the 45% of the vote uncommitted or for another candidate to Mr. Obama. This appears to be the intent of those voters, as well as the likely result of a rematch. It would reduce Mr. Obama's current edge in pledged delegates to 115 from 167. It would also reduce the adjusted popular-vote margin, that converts caucus votes to primary votes, to an edge for Mr. Obama of 466,000. If Mrs. Clinton wins Pennsylvania by the margin polls now suggest, the two candidates would be essentially tied in popular votes, with an Obama edge in delegates of about 80. That would leave the remaining primaries and the superdelegates to decide the outcome of an essentially tied race.

Democrats are clearly going to have to rewrite their delegate selection rules after this contest, like they did after similar fiascos in 1968 and 1988. Until then, it's up to the lawyers, and may the cleverest lawyer win. My money is on Mr. Obama blocking the statistically based solution described above. After all, as a product of Harvard myself, I know perfectly well that Harvard produces cleverer lawyers than Yale, regardless of what the numbers might say.

Lawrence B. Lindsey is a visiting scholar at AEI.