There are various popular vote totals for the Democratic nomination race floating around. MSNBC's has gotten a lot of attention in particular. While they aren't clear about what their methodologies are exactly, they all miss at least some of the "popular vote" in the caucuses. So I decided to do my own calculations from scratch, for both primaries and caucuses. Wherever possible, I used official figures, the state election agency for primaries and the Minnesota caucus, state Democratic parties for the other caucuses. Altogether, I was able to find official results for 30 of the 45 contests held so far (through the Wyoming caucus, March 8). For the other 15 I used media sources (mostly the New York Times), and other web sources (mostly Dave Leip's excellent presidential election page: http://www.uselectionatlas.org/). I have not included either the results from the Washington state "beauty contest" primary, or the Texas caucuses. In the former case, it's hard to know what to make of the results, and in the latter the available results are not even close to complete. In any case, I have included the Washington caucuses and the Texas primary, and avoided "double-counting" either state.

I know you are all anxious to see the results (below), but this is a somewhat tricky and controversial endeavor, and the following is the minimal explanation necessary to read the results intelligently. Gary Rees has posted his own estimates here (http://tpmcafe.talkingpointsmemo.com/talk/2008/03/the-popular-vote-obama-leads-b.php) using a very different approach than mine.

While caucus results in terms of votes are often described as impossible to obtain, this mostly isn't actually true. The situation is certainly confusing, and the media hasn't helped. For example, for Kansas, the New York Times presents the results as state delegates, when they actually are the popular vote. For Washington state, they make the opposite mistake. However, it actually isn't that hard to get most of the popular votes, with a little digging. For 12 of the 16 caucuses so far, the state parties have released the numbers. For the other four (Iowa, Nevada, Washington state, and Maine), they have only released counts of delegates elected to state or local conventions; the popular votes are either being withheld or were never even collected. For these states, I have estimated the popular vote by multiplying the percentage of state or local delegates for each candidate by the total turnout. I am aware this isn't a perfect method, but I'm confident it's a reasonable estimate. Luckily for three of these states, the state parties did release precise turnouts. The Washington state party only gave the turnout as "more than 250,000," so I have used 250,000 in my estimate, to be conservative.

I have presented three sets of results, one without the Michigan and Florida primaries, one with Florida and not Michigan, and one with both. I don't want to get into a discussion as to the legitimacy of these primaries here, with one exception. For my estimate that includes Michigan, I have made an estimate of Obama's vote (method described below) rather than giving him zero votes. A result giving Obama zero votes in Michigan seems to me to be clearly worthless and misleading. Obviously, many of the people who voted for "Uncommitted" would have voted for Obama had he been on the ballot. I can't think of an ideal way to estimate his "fair share" of Uncommitted. Obviously at that point in the race, some of the Uncommitted vote would have gone to Edwards, and a tiny share to other candidates. What I did was assume that the Uncommitted vote in Michigan would have split among the four candidates not on the ballot (Obama, Edwards, Biden, and Richardson) in the same proportions as their actual votes in Florida. This gives Obama just over 67 percent of the Uncommitted vote, or 159,859 votes in Michigan. I won't even pretend that this is "right," but I can't think of an obviously better method, and it does seem plausible to me. I welcome suggestions of better approaches.

Here are the results:

Total without Michigan and Florida
Clinton 12,647,767 (46.7%)
Obama 13,353,753 (49.3%)
Obama lead 705,986

Total with Florida but not Michigan
Clinton 13,518,753 (46.9%)
Obama 13,929,967 (48.3%)
Obama lead 411,214

Total including Florida and Michigan
Clinton 13,847,062 (47.0%)
Obama 14,089,827 (47.9%)
Obama lead 242,765

Here is the breakdown of primaries versus caucuses. I am only showing the results without Florida and Michigan, but you can easily calculate the primary results for the other versions by simply subtracting the caucus results from the totals above.

Without Michigan and Florida
Primaries
Clinton 12,194,118 (47.3%)
Obama 12,598,794 (48.9%)
Obama lead 404,676

Caucuses
Clinton 453,649 (34.7%)
Obama 754,959 (57.8%)
Obama lead 301,310

As you can see, despite the hugely greater number of votes cast in primaries, Obama's large percentage lead in caucuses allows him to nearly double his overall popular vote lead.

I think that this is the best possible way to calculate the popular vote. I am well aware that if the caucus states had held primaries instead, the results might have been very different (certainly turnout would have been much higher), and I am not brave (or foolish) enough to attempt an estimate of what they would have been. This estimate is my best attempt to count the actual votes, for better or worse. I welcome comments and constructive criticism.

The spreadsheet that contains the raw numbers and calculations is available for download at: http://arc1.uwindsor.ca/~dedelste/popular_vote_dem_primary_08.xls
I have not attempted to clean it up. I think it is reasonably self-explanatory, except for the "scratch" tab that contains unidentified calculations for various random states. You now have everything I do.

I leave you to contemplate the sad fact that around 250,000 people voted in the Washington state Democratic caucuses, that actually chose that state's delegates, while almost 700,000 voted in the meaningless Washington state Democratic primary. I hope that whatever our allegiances, we can all agree that the current system of choosing nominees desperately needs reform.

Disclaimer: I have tried to be as objective as possible in this exercise, and especially to be clear as to methods and assumptions, and to present alternate sets of results to allow people to decide for themselves which they consider most valid. I am a former strong Edwards supporter who now supports Obama cautiously. I would vote for either Obama or Hillary over McCain without doubts, but not necessarily with enthusiasm.