Sorry for the delay!
I've been away on business for a week, and hadn't posted new numbers for 2 weeks before that, so I've got lots of new information. Sorry it took me so long to get caught up!
There have been some dramatic numbers changes. Back on the 20th, I predicted, at his current rate, Dean would have enough delegates to "clinch" (2162) by New Year's.
I was wrong.
In fact, Clark has made some significant gains in the Delegate counts, based on his consistent high performance in recent polls.
While trying to use polls to predict delegates is silly, using delegate changes over time to show trends is probably even sillier, but here goes:
Dean had 2003, and Clark 623, on the 20th. At that point Dean had been gaining at the rate of about 30 per day. I had been tracking from November 28th to 12/20 (a period of 22 days.)
Another 22 days later, based on current polls, Dean now has 1859, and Clark 920. Dean has dropped about 6.5 delegates a day since 12/20.
However, since 11/28, he's gained an average of 11.2 delegates a day. It's anyone's guess what will happen in the next 8 days: whether his current short term trend of losses will continue, or whether his longer term trend will return. After Iowa, however, there are going to be some dramatic shifts almost immediately, and it all depends on how the caucuses go.
Click on "There's more" to view further analysis and the graphic showing the numbers and breakouts by primary date.
Full numbers
To view the latest numbers in detail, click here (306 KBytes)
States won by date (Based on polls.)
New Additions: SuperDelegates
I've added the information from the latest poll of DNC Members to help improve the accuracy of my SuperDelegates numbers. However, the poll data are not "hard counts", so I've provided both numbers in my totals: Confirmed and Likely are the orange and yellow lines. Confirmed are based on announcements and other hard research. Yellow are based on the latest poll of DNC Members. See http://www.grumpyoldbear.com/Superdelegates.htm for more detailed explanation and breakouts.
One big thing that sucks is that ABC actually contacted all of the Supers and asked them who they were supporting. They got 90 or so for Dean, whereas I have only been able to identify 66. They also had more for Gep and Clark than I had (although I had all of their Kerry people, woo hoo, largely thanks to the Kerry Web site actually posting them instead of keeping them secret.)
If anyone can help me figure out the remaining 30 Dean people and the missing Clark and Gep people, I would greatly appreciate it.
New Additions: Meta Analysis of National Numbers
I got some criticism that using only one national poll to fill in the missing states was less accurate than aggregating several national polls. ohwilleke gave me some advice on how to do some quick and dirty meta analysis. The results have already confused some people, so let me try and explain briefly.
- I'm taking the percentage for each candidate and the total # of respondents.
- I'm multiplying those together to get "Total Respondents for Each Canddiate" in that poll.
- I'm adding all the respondents from each poll together, both for each candidate and total.
- The percentage of total respondents for each candidate, divided by the total respondents for all candidates, is the "meta" number.
The end result is, in this case, 378.3 respondents for Dean, out of 1395 total. This is more than 27%, which is his highest poll number. The reason some people got confused is they said "if you're averaging 4 polls, and he got less than 27% in 3 of them, then he can't get more than 27% total".
If I were averaging the results from each poll, that would be accurate: Dean would have 23.8 in that case. Using the total # of respondents to weight the polls is what changes things, and the cause of the confusion is that the candidate in second place varies among polls. Most have Clark, but some have had Lieberman or Gephardt. When that happens, Clark gets fewer total respondents than he would if he were always in 2nd place, which means that Dean gets proportionately more.
I think this is actually an eye-opening reflection on the realities of the race: Dean isn't fighting a single second place challenger. He's fighting 4 regional candidates and a national candidate who hasn't reached his full strength. He's fighting Kerry in New England, Gephardt in the Midwest, Edwards in the South, Lieberman in the Mid-Atlantic, and Clark in a variety of states throughout the country. Clark is the strongest nationally, and I believe he will be the eventual "Bill Bradley" to Dean's "Al Gore" (or the reverse, if Dean doesn't perform as advertised.)
