[Update 2/27: After getting sucked into this discussion (where I can't comment because they haven't approved my registration) by way of this post, I went ahead and tried to come up with a rational estimate of popular support among registered Democrats for each candidate in the caucus states. But in trying to nail down the caucus ambiguity, I realized there is another, more basic ambiguity that blows it out of the water and makes this whole exercise pointless: Eight primary states and one caucus state do not have registered party affiliations, at all! Over 38 million registered voters!
Riddle me this: How many registered Democrats voted in the primaries in Alabama, Arkansas, Georgia, Illinois, Missouri, South Carolina, Tennessee, Virginia and Wisconsin? The answer is the same for all of them: Zero, because there are no registered Democrats in those states. So what the hell have I been adding up? Exit poll percentages - which are subject to sampling error themselves, by the way - of voters’ self-reported party affiliation. If there is no way to objectively define what we mean by "registered Democrat", then trying to add up "registered Democrat" votes is impossible.
Just by the way, until yesterday I thought Obama fanatics were the worst in terms of being rude and divisive, but then I discovered that thread at corrente. Holy shit, they are crazy.]
Booman is very upset at this post by Jeralyn Merritt:
Who's really picking our Democratic nominee? If it's the Democratic youth or African American voters, I'm okay with that. That's fair. If it's Republicans, I'm not. We'll get trounced in November. Without reliable stats to show Obama's support is from those who will for [sic] the Democrat in November, I'd say the best way to ensure Republicans stay out of our race and don't steal another election from us is for Dems to vote for Hillary to be the nominee.
Her argument is really a non sequitur, because the article she points to says that Obama has won 64% of the independents, not Republicans, in the Democratic primaries so far. Rising to the bait anyway, Booman retorts:
In South Carolina Barack Obama won among registered Democrats 57%-28%-14%. In Virginia, he won among registered Democrats 62%-38%. In Maryland he won among registered Democrats 59%-40%. In Wisconsin, he won the vote among registered Democrats 53%-46%. In spite of this, I spent a good part of my day explaining to Clinton supporters (in email) that Barack Obama is not winning the nomination on the backs of Republican voters who won't show up for him in the fall. It's a ludicrous assertion.
So just for completeness' sake, I went ahead and tallied all of the Democratic votes in the primaries so far. (I left out caucus states.) If you leave Michigan and Florida aside, Hillary has received about 7.4 million Democratic votes and Obama has about 7.1 million. If you include Michigan and Florida (and assume all "uncommitted" votes in Michigan would have gone to Obama), the totals increase to 8.4 million and 7.7 million, respectively.
On the merits, Jerlayn is clearly off base. Republicans have played a trivial role in the Democratic primaries. She's essentially arguing that winning a huge majority of independents is a bad thing for the nominee, which is absurd. Also, open primaries are open because the state Democratic parties want them to be. But the implication of Jerlayn's argument is that open primaries are illegitimate, and that we should only have closed primaries. Why? Finally, just like the issue of whether the Florida and Michigan delegates should be seated, note that Clinton supporters are arguing once again that the rules should be changed mid-game so that their candidate can win.
I relied on CNN for my vote numbers. Here's the spreadsheet I created:
Update: upyernoz says in comments: "why exclude the caucuses? i realize there isn't 'vote' numbers in the usual sense, but there are exit polls that can be extrapolated into something like a vote breakdown."
I excluded caucuses because I can't come up with a reliable way to correlate caucus votes to popular vote. For example, Obama won 75% of the caucus votes in Alaska, a margin he's not gotten near in any primary other than D.C. In Hawaii, same story: 76% Obama, 24% Clinton. Idaho: 79% Obama. Kansas: 74% Obama to 26% Clinton. Nebraska: 68% Obama - 32% Clinton. In Colorado, he got 67% of the caucus votes to 32% for Hillary, which is more believable but still really high. Maine had the most believable margin: 59% Obama - 40% Clinton. In contrast, the highest primary vote percentage Obama got - outside of D.C. - among registered Democrats is 57%. In the D.C. primary, Obama got 75% of the vote. To believe that caucus percentages are representative of the voters, you'd have to believe that coincidentally all the caucus states had Democratic voters who favored Obama as much as voters in D.C. (with 57% African-American population). Much more likely is that a combination of factors favored Obama in caucuses: a better ground game, a different demographic going to caucuses rather than primaries (say, fewer older voters), and possibly a "me too" effect of younger, more passionate Obama caucus-goers pulling people to their side of the room.
The point here isn't that caucus wins "don't count" or are illegitimate, the point is that while we can say confidently that Obama would have won the caucus states if they had held primaries, we can't confidently say by how much because caucuses are unrepresentative of the voters at large. You just can't multiply out caucus vote percentages by numbers of registered Democrats and say that's the number of voters each candidate would have won, just as you can't include Michigan and Florida.
why exclude the caucuses? i realize there isn't "vote" numbers in the usual sense, but there are exit polls that can be extrapolated into something like a vote breakdown.
it just seemed like jeralyn was fixing the whole thing when she excluded the caucuses (coincidentally the states where obama did best). and it echoes the clinton campaign's lines that certain states (e.g. states with a lot of black people, states where obama has lived, states with caucuses, states with open primaries, "red states", midwestern states, etc) don't count whereas other states (e.g. states where clinton has lived, states with a lot of hispanic people, states with closed primaries, non-southern states on the coasts, states that clinton hasn't lost yet) do.
when you start gerrymandering the results to suit your candidate that much, it just looks like a pretty weak argument.
Posted by: upyernoz | February 25, 2008 at 11:55 AM
Well, I initially excluded the caucus states because it muddles the picture, not because it skews the results. The poll numbers before primaries have been somewhat inaccurate, but the poll numbers before caucuses are totally inaccurate, which leads me to think the sentiment among caucus-goers doesn't reflect that of voters.
Posted by: Mithras | February 25, 2008 at 12:21 PM
but the poll numbers before caucuses are totally inaccurate
then why not use the poll numbers after the caucus, i.e. the exit polls.
exit polls aren't always perfect, but they do tend to be more accurate than pre-election polling simply because they ask what someone has already done and not to predict what they are likely to do in the future.
Posted by: upyernoz | February 25, 2008 at 02:00 PM
Um, because the exit polls are polling the same people who participated in the caucus, and caucus-goers seem to be more pro-Obama than primary voters would be. Wait, I'll do an update.
Posted by: Mithras | February 25, 2008 at 03:06 PM
Update posted.
Posted by: Mithras | February 25, 2008 at 03:48 PM
re: the update
those are all fair points. you're right that there probably isn't a good way to capture how a primary vote would have gone if it were a primary and not caucus state.
but why try to convert everything to a primary state? while primaries are secret ballot and thus more like the general election, they don't actually reflect who will turn out in the general election. primaries always have lower turnout than the general. also, if you're trying to build a model that is most like the primaries, then wouldn't you want to count crossover voters (i.e. republicans and independents voting for the democratic candidate) as they are often the swing voters in the general election?
but that cuts to the heart of jeralyn's premise--which i realize you don't completely buy--that the ability to draw non-democrats shouldn't be a factor in the primary process. as you said above, that doesn't really make sense.
Posted by: upyernoz | February 25, 2008 at 08:05 PM
Yeah, the whole point here was to count registered Democratic voters, which I found somewhat interesting. I am just saying there is no accurate way to do that in caucus states.
Posted by: Mithras | February 25, 2008 at 10:26 PM
http://ivrpolls.com/index.php?option=com_content&task=view&id=36&Itemid=1
Why are so many of the older white male Obama primary voters saying they'll vote against him in the general election?
Posted by: Susie from Philly | February 26, 2008 at 02:36 PM
For the same reason 6% of Clinton primary voters say they'll vote for McCain. And the same reason a large number of Republicans say they will stay home on election day rather than vote for "that RINO McCain":
Because the election is still over 8 months away, and predicting what people will actually do that far in advance is useless.
Posted by: Mithras | February 26, 2008 at 03:01 PM