While reading Hillary Clinton's increasingly outrageous arguments that the Florida and Michigan primary votes should count (e.g., the Zimbabwe analogy), I decided that rather than just getting angry it might be useful to try to take an objective look that isn't simply based on candidate preference or Democratic Party rules. Did the Democratic voters in these two states make their preferences known in the primaries? Of course, for Michigan the question sounds completely ridiculous because Clinton was the only major candidate on the ballot (Kucinich, Dodd, and Gavel were also listed). However, voters were also able to choose "uncommitted", which 40% of them did, and there was an exit poll that showed 46% for Clinton, 35% for Obama, and 12% for Edwards.

So, I thought that the best starting point for assessing the primaries was to look at the turnouts. What I was expecting to find was that the turnouts for both primaries were extremely low, well below those for the other primaries. In that case, Clinton's argument -- that the voters in some sense would be disenfranchised by not counting the results -- would clearly have no basis.

This is exactly what I found for Michigan. Only 594 thousand voters participated, compared to the 2.5 million who voted for Kerry in the 2004 presidential election, i.e. a turnout rate of 24% (I recognize this isn't the usual way to calculate a turnout, but for the question at hand it doesn't matter). This was well below the average of 61% (through Feb. 19th; the turnout rates actually became even higher after that date) and was by far the lowest in any state (the next lowest was 40% in NM). The NH primary a week earlier had a turnout that was 84% of the 2004 Kerry vote, and the SC primary immediately following Michigan had a turnout rate that was 80%. In conclusion, there is no reasonable basis for counting the Michigan primary votes, even if one were to somehow use the exit poll results. The Democratic voters of Michigan were indeed disenfranchised, but this was due to the decisions of their state party and the DNC on the primary date, and cannot be remedied by validating the primary votes.

The story for Florida is quite different. Here all of the candidates were on the ballot, but (for the most part) didn't campaign within the state. The voters actually turned out in fairly high numbers in this case: 1.75 million, compared to 3.6 million who voted for Kerry in the 2004 presidential election (49%). This is lower than the average of 61%, but within a standard deviation (16%), and higher than the turnout rates in NM, CT, NY, LA, and DE (40%, 41%, 43%, 47%, and 48%. respectively). It is also much higher than the turnout in the 2004 Florida primary (752 thousand), although that primary was fairly late (March 9th, just at the time that Kerry accumulated the necessary number of delegates).

One could argue that the Florida turnout would have been much higher if the DNC hadn't announced that they wouldn't seat the FL delegates at the convention. In favor of this argument is the observation that the 49% Democratic turnout rate was matched almost exactly by the Republican turnout (1.95 million, or 49% of the vote for Bush in the 2004 presidential election), whereas the Democratic turnout overall was much higher than that of the Republicans. However, the party differences were fairly small in a couple of cases (NH and AZ), and there is no way to really know how the DNC decision affected turnout, nor is there any reliable way to determine whether a larger turnout would have significantly affected the outcome (or, for that matter, what was the effect of the absence of in-state campaigning).

In light of these considerations, it is fair and reasonable to count the Florida primary votes. It is also politically wise, in terms of promoting a Democratic (presumably Obama) victory in November. After watching "Recount" on HBO last Saturday and seeing in graphic detail how the Democratic voters in FL were disenfranchised in 2000, largely due to missteps of the Dems on the national level, I think it is especially important to make sure that the votes of the Florida Democrats count now. How many times can we expect them to get excited and go out to vote when their votes don't count?

Inclusion of the Florida votes would have no appreciable effect on the outcome, as Clinton would gain only 38 delegates more than Obama, much less than the present difference of ~158 in pledged delegates (further enhanced by ~32 in superdelegates). If the Edwards delegates go to Obama (as might be expected since Edwards now supports Obama), then Clinton gains only 25 delegates. Even if one chooses to emphasize the popular vote, as Clinton now disingenuously does, it doesn't close the gap -- Obama would still lead by ~274 thousand (vs. ~569 thousand without FL).

As stated at the outset, the preceding analysis did not take Democratic Party rules into account. Is the violation of these rules by the Florida Democratic Party sufficient cause to discount the primary vote in that state? Some may feel that the DNC must hold the line on this -- if Florida is allowed to break the rules and move up their primary date, then this might encourage other states to do so. Many would argue that it is already a problem that the campaign starts so early, and that the problem would be intensified if the primary season shifted even earlier. It could also be argued that future candidates would feel that they had to campaign in rule-breaking states if the votes might ultimately be counted, and that this would add further problems to the process.

These arguments may have merit, but should be left for another day. A relatively easy compromise solution for the case at hand, Florida, is for the DNC to agree to count the primary votes if the Florida Democratic Party agrees to push back its future primaries to super-Tuesday (or later). After this election, serious thought should be given to revamping the entire system, which gives Iowa and New Hampshire an unfairly prominent role and, in most primary seasons, leaves Democrats in later-voting states no real voice.

For now, the overriding consideration should be the November election. If the Florida Democratic primary votes are discounted, it seems likely that Democratic voter turnout will not be high enough to avert a McCain victory there. Gaining this state might be sufficient for McCain to win the national election, which would be disastrous. This is a much more important issue than reining in the states that try to move up their primary dates.