Looking Ahead to the General Election - Part 3: Primaries, Caucuses and Electability
I have no idea who will win the Democratic nomination but I want to briefly discuss the states that the candidates are winning. Daily Kos diarist Poblano published a regression analysis on Saturday on the demographics of those who had voted for Sen. Clinton and Sen. Obama prior to February 9th (interestingly, one of his projections was that Obama would sweep the states in February after Super Tuesday - based on that analysis). His findings are somewhat complex and I recommend you read the diary in its entirety to think about the points he is raising. I want to focus on a portion of his analysis:
1. Caucus versus Primary. Obama does do better in caucus states -- a lot better. In fact, a caucus appears to be about 13 points better for him than a primary, all else being equal. By the way, the type of primary or caucus does not seem to matter. That is, Obama doesn't do any better in "open" primaries or caucuses. Perhaps this will change as [t]he Republican race winds down, but it hasn't mattered so far.
Prior to Sen. Obama's victory in the Maine caucus, Marc Ambinder also noted the sharp difference in caucus victories:
Obama has now won 10 caucuses and nine primaries. Clinton's won 9 primaries and 2 caucuses. [Eriposte note: I believe that Clinton has formally just won one caucus to date (NV).]
The caucus-primary difference is very interesting and I haven't seen a compelling explanation for it yet (I'm trying to find some time to analyze the overall voting populations in this primary compared to the total registered voters). MyDD diarist DaveOinSF claims, based on some data on turnouts, that although the voter turnout in the smaller caucus states has often been much higher than the corresponding turnouts in the 2004 caucuses, the number of caucus voters in many of the states as a percentage of the general election (GE) Democratic voting population is much smaller than in the case of the primary voters in the larger states that Sen. Clinton (and sometimes Sen. Obama) has won - meaning that Sen. Clinton's wins in the primary states have generally involved a much larger turnout of the overall voting population in comparison to Sen. Obama's wins in the caucus states. He provides numerous examples: for instance, the Alaska Democratic caucus that Sen. Obama won involved a total turnout of less than 9000 voters compared to the roughly 111000 voters who voted for John Kerry in Alaska in the 2004 general election (no wonder some of these small state caucus wins can be pretty misleading)! In contrast, in Nevada, which was the only caucus that Sen. Clinton won, roughly 120000 voters showed up in comparison to the roughly 397000 who voted for Kerry in 2004, making that a caucus with one of the highest percentages [*edited] (29.7%, just a tad bit
higher lower than Iowa - see Part 4 for more on this) of caucus-to-GE Democratic voters in this election. According to DaveOinSF, in the case of the large primary states, the turnout is running much higher as a percentage of the GE voting population - the numbers he cites for the Clinton states run in the 50-67% range. This is suggestive - and not entirely surprising - that Sen. Clinton tends to do better overall when a much larger percentage of the voting population is voting in the primary or caucuses (and indicates that her campaign has not done a good job of getting their voters to the caucuses and making them stay till the end of the caucuses).
Let's continue on with Poblano's diary:
7. John Kerry vote share, 2004. Even controlling for all these other variables, Hillary does better in blue states, and Obama does better in red states. Again, the best way I can explain this is that Hillary excels in states with strong, liberal establishments. However...
8. Percentage of Democratic voters who self-identify as Liberal. Obama gains ground in states where a high percentage of the Democratic base identifies as "liberal". This data was gathered from the CNN exit polls that I mentioned above, based on a sample of Kerry voters only. Obama's very best states are states that overall are "red", but where the Democrats that are in the state are very progressive. Idaho is one such state, for example; 42% of Kerry voters in 2004 identified as liberal (the nationwide average is 35%), even though Bush beat him by more than 2:1.
Actually the larger blue states not only have stronger liberal establishments - they also tend to have larger immigrant communities and significant concentrations of working class democrats who are more often supporting Sen. Clinton over Sen. Obama, in contrast to the more liberal voters and youth who are more often supporting Sen. Obama over Sen. Clinton. Sen. Obama has more often won in states that are smaller and tend to be Republican strongholds - the kind of states that Democrats have very little chance of carrying in the Fall. Sen. Clinton, as Poblano notes, has won most of the largest states that Democrats traditionally carry in the general election. Taylor Marsh has a more detailed post and table that talks about past winners of the Democratic and Republican nomination since 1968 and notes that the eventual nominees won most of the delegate-rich big states. [Some interesting sidenotes: Sen. Obama has had difficulty winning in Republican states with strong concentrations of Southern Baptists, according to Poblano. Poblano has not analyzed the Catholic vote, but Sen. Obama is currently losing the Catholic vote by a huge margin to Sen. Clinton - and according to the NYT, "Catholics, who make up about a quarter of the registered voters in the country, have backed the winner of the national popular vote for at least the last nine presidential elections, going back to 1972". More on "swing voter" trends in this MyDD diary.]
The data set is obviously complex and it's hard to make conclusions with certainty. However, what we can say based on the data is that the eventual nominee has generally won most of the largest states, and we should be extremely cautious about using Sen. Obama's wins in the smaller states (especially red states with minority progressive populations and often much smaller turnout percentages as a function of the overall voting population) as a sign that he can somehow win more convincingly in the general election than Sen. Clinton. As Jerome Armstrong said (emphasis mine):
The Obama campaign makes the case:
On Super Tuesday, in six red states that had primaries or caucuses for both Republicans and Democrats, Obama won and got more votes than the top two Republicans combined. These states - Colorado, Georgia, Iowa, Missouri, North Dakota and South Carolina - account for a total of 53 Electoral College votes. In Idaho and Kansas, where there was no Republican primary, Obama won at least a three-to-one victory over Clinton.
That's not serious. To quote North Dakota, where a total of 18,000 or so voted in the Democratic caucus [Eriposte note: Compared to roughly 111000 who voted for John Kerry in the 2004 GE], as proof that Obama could win the state is laughable. That sort of logic puts Nebraska, which had a similar total, and voted for Obama, also in the Democratic column for Obama in November.
Finally, although the Obama camp has been spinning a scenario where more of his voters might not vote for Sen. Clinton than vice versa, it is also quite possible that more of the traditional Democratic vote in the cities (including the youth vote) might transfer to Sen. Clinton in the general election if she is the nominee, relative to the transfer of the more moderate/conservative vote from Sen. Clinton to Sen. Obama if he becomes the eventual nominee (example). So, Sen. Obama's assertion - that he will likely get the voters of Sen. Clinton and that it is questionable whether she would get his voters - may not necessarily turn out to be true.