More Projections on the Popular Vote for Clinton and Obama

My recent post noting that Clinton is currently ahead in popular vote (with Michigan and Florida included) has inspired many comments in which many reasonable arguments have been advanced on all sides. One criticism that I should note is that the realclearpolitics.com compilation of the popular vote doesn't credit Obama with the votes cast for "Uncommitted" in Michigan, and that if he is credited with those votes Clinton's current popular vote margin disappears. Probably most, though certainly not all, of the votes cast for "Uncommitted" were intended for Obama (John Edwards was still a live candidate when Michigan voted); reasonable people can argue about how many should be attributed to him.

To gauge the possibility of Clinton's current popular vote lead (if it is that) being maintained, I used once again, as I did in my March 28 post Jay Cost's spread sheet, this time plugging in the results from Pennsylvania, which had a higher turnout but a lower Clinton percentage than I had projected—and remember that these projections are optimistic from Clinton's point of view. This time I found Clinton getting a net gain of 545, 298 popular votes, including her actual 207,529 popular vote gain in Pennsylvania. The net post-Pennsylvania gain is 337,769. I have assumed that turnout will be 77.2 percent of the 2004 Kerry vote, as it was in Pennsylvania; obviously in Puerto Rico there was no Kerry vote, so I have arbitrarily assumed a turnout of 1 million voters in a commonwealth with 2.5 million voters; it could go much higher, or lower. In North Carolina, I have predicted that she would do slightly better than current polls, leaving aside the results from the latest poll from PPP, which seems to have been all over the lot in recent races (they had Obama ahead in Pennsylvania). Despite the current polls showing Obama ahead in Indiana, I have projected a narrow Clinton win there, with half the percentage margin she won in Ohio and Pennsylvania. I have assumed, optimistically from Clinton's point of view, that West Virginia and Kentucky will deliver large percentages for her (as all Appalachian counties around those states have done) and that the Oregon results will look more like the desultory victory Obama won in the nonbinding February 19 caucus in Washington rather than his much more robust percentages in the February 5 binding caucuses in that neighboring state. I have kept the same numbers from Puerto Rico, and acknowledge that no one has any real idea of the turnout or the Clinton percentage in this unprecedented contest (the one poll I am aware of shows her, plausibly, with a large percentage lead). I have heeded the advice of many commenters, and the results from one South Dakota poll, and projected Obama rather than Clinton the winner of the June 3 primaries in South Dakota and Montana. The numbers look like this:

Currently, according to realclearpolitics.com, Clinton has a popular vote advantage of 122,728 when you include the Florida and Michigan results. When you offset this by including the imputed turnout (imputed, because state Democratic officials did not tabulate the actual turnout) in the Iowa, Nevada, Maine and Washington caucuses, Clinton has a popular vote advantage of 12,506. What happens to these when you plug in the projected votes cited above?

These projections are not quite enough to eradicate Obama's current popular vote leads without Florida and Michigan, which are without the imputed caucus results in those four states +500,543 and with those results +610,575. But by the former measure they leave Hillary Clinton only short by 40,046 votes, about one-tenth of 1 percent all votes that will be cast in Democratic primaries and caucuses. Admittedly, these projections are optimistic for Clinton, but I would submit not wildly so. And they could significantly underestimate or overestimate her popular vote margins, depending on what happens in eminently unpredictable Puerto Rico.