Monday, June 2, 2008

Montana Projection: Obama by 18


As compared with the night of May 20th, when the combination Obama's embarrassing margin of defeat in Kentucky and the long break separating its poll closing from the one in Oregon led to a sort of anticlimactic night, the optics work out relatively well for Obama tomorrow. Tactically, his objective is round up enough superdelegates to hit 2,118 before Chris Matthews goes to sleep. But aesthetically, his goal is probably the following: be the first one to have a state called for him. That could be accomplished either by winning South Dakota outright, or by keeping it close enough that the networks are ready to call Montana first, where polls close an hour later. The general rule of thumb is that for the networks to call a state immediately, they need to see about a 15-point margin in the exit polls. Although it is by no means a completely safe bet for Obama, we expect him to achieve that threshold.

What Barack has going for him: At a surface level, Montana and South Dakota are pretty similar. They are both exceptionally white states, with the principal minority group being Native Americans. To the extent there are differences, however, they tend to favor Obama.

Montana's Democratic electorate is somewhat more progressive than the one in South Dakota, but also more independent-minded and libertarian. Its signature issues are probably being pro-environment and pro-gun; Obama tends to do well on the former issue, whereas Montanans see Hillary as too much of a big government power broker to have much credibility on the latter. Montana has fairly high levels of education, and though its incomes are below average, Obama tends to do well in "bohemian" areas where education runs ahead of income.

While these are not exceptionally large differences, they are compounded by the fact that Obama has spent more time on the ground in Montana than Clinton, and that Montana has an open primary, with no apparent sign of interference from Operation Chaos. Obama has also fundraised reasonably well in Montana, whereas Clinton's numbers are marginal. Also, Montana is a pretty state, and Obama has won every pretty state except California (OK, we made that last part up).

What Hillary has going for her: By any standard definition, Montana is white and working class. Neither of the Democrats' two All-Stars in the state, Brian Schewitzer and Jon Tester, have yet endorsed Obama, so Clinton does not have to contend with the institutional support that Obama has in Tom Daschle's South Dakota. Bill Clinton carried Montana in 1992.

Apart from these superficial markers, however, Clinton isn't a great fit for the state's political culture, which in some ways resembles the Pacific Northwest as much as it does the Prairies. It might be thought that Hillary performs well in rural areas and Montana is quite rural. But she really only performs well in some types of rural areas, generally those associated with cultural conservatism, which are more likely to be found in the Southeastern quadrant of the country (Obama, by contrast, performed well in OR-2, one of Montana's closer comparables). Although Bill has worked Montana aggressively, Hillary has largely abandoned the state for South Dakota, and her campaign's body language reads as a probable double-digit loss.

Projection: Our numbers work out to Obama 59.1, Clinton 40.9, or a victory margin of about 18 points. We expect a heavy turnout of about 166,000 in Montana, which does not have quite the voting culture that you find in the North Central portion of the country, but does have an open primary and a somewhat larger population. As measured in votes, our projection is Obama 98,373, Clinton 68,079.

Montana's delegate allocations are a little funky, with the state divided up into two pseudo-CDs based on Montana's old congressional districts from the 1980s (it has just one now). Each district has five delegates; any margin of victory would give Obama (or Clinton) three of those five, whereas they'd have to hit 70 percent to get a fourth. So in practice, the district-level delegate allocations will very probably be 3-2 Obama and 3-2 Obama. On an extremely good night, Clinton might win the Eastern district, and Obama might hit 70 percent in the Western district, but those outcomes are unlikely.

Obama has a better chance to win a fourth at-large delegate, which would require 62.5 percent of the vote, but we have him falling a bit short. The 2 PLEOs will definitely be split 1-1.

Delegate Projection:
Eastern: Obama 3, Clinton 2 (~10% chance of Obama 4, Clinton 1)
Western: Obama 3, Clinton 2 (~20% chance of Clinton 3, Obama 2)
At-Large: Obama 2, Clinton 2 (~25% chance of Obama 3, Clinton 1)
PLEO: Obama 1, Clinton 1
Total: Obama 9, Clinton 7

23 comments

Chuck in Seattle said...

Thank you!

I'm inclined to think it's possible that Hillary could get 3-2 in the east (more so than Barack getting 4-1 in the west). Billings is the "conservative" big city and all the colleges are in the west. I was not shocked the Barack ran off to Great Falls (in the east) last week to try to shore up that half of the state. Great Falls is a Dem stronghold, but he had a miserable turnout at his event. The local newspaper repeated that it was below expectations.

I'm going with 8-8 in Montana. With higher numbers in the west for Barack, and higher numbers in the east for Hillary. So I'll give you your 59-41.

Anonymous said...

I like the pretty state theory! Only one problem: New Mexico's gorgeous.

Lisa said...

Chuck, Billings is a college town. I don't think Obama will do as well there as he will in Butte, Bozeman, and Missoula, but Billings is not the pro-Hillary country to the extent that the rural Eastern Montana counties are.

Montana allows walk-up registration so we could see a lot of new voters tomorrow.

I'm curious to see how many votes Ron Paul gets. Montana is one of his strongest states.

Lisa said...

oops. scratch that. I forgot that Montana's Republican party had its caucus in February.

Anonymous said...

Does your model include any projections about whether Clinton voters are more motivated to turn out for the primaries now that the nomination result is certain? It seems like most Obama folks realize it's over, while most Clinton voters still believe in the idea of a "protest vote," or, more charitably, a "show of support."

I've wondered whether the KY, OR, and PR numbers have been skewed by this effect. And perhaps even more so in SD and MT.

Anonymous said...

Did Obama really get pathetic turnout for his rally in Great Falls? Ouch. Great Falls is one of the biggest cities in Montana and a bigtime Dem stronghold. What was the turnout anyone?

Anonymous said...

Hey the turnout wasn't that bad for Obama in Great Falls, he got 2400 to show up. Granted, they didn't fill the 6000-seat venue like they wanted to, but 2400 in Great Falls is pretty great.

Anonymous said...

Hey, Poblano, do you have the final site layout and the final version of your regression model finalized in case Clinton drops tomorrow? If you don't I have a few suggestions/requests:

1. Kerry's run alone clearly hasn't been that great for predicting voting trends and turn out. You should include two losing years and two winning years in the 538 regression (i.e. Kerry, Gore, Clinton, and Clinton) and also weight them according to their age as you do with polls.

2. The 538 Regression should lose weight as the data it is based on becomes older, otherwise everything will regress toward Kerry's run no matter what ridiculous stuff happens this year, and Obama is no Kerry despite McCain being a pretty good Bush.

3. The nation-wide popular vote isn't as interesting as the state-by-state popular vote. Since your left column appears sized for your ads you might consider putting this in where Clinton's percentage is.

4. The swing state analysis and win percentage are interesting but they poise as many questions as they answer. It would be cool if clicking on one of the states at right brought up a small window (like this post window) which gave a more complete state analysis, possibly including a graph of the distribution of popular votes in the 10,000 runs as well as a graph for how the state has been predicted to go over time, similar to those you have for the entire election. Also a demographic and political affiliation breakdown would be interesting. Alternatively you could bring up a single page which had in-depth analyses for all the states so that they could be compared easily. This would make it easier to see which states appear to be fluctuating more and causing the biggest changes in how the election will go (i.e. maybe not the most important swing state at any point in time but the states which should still be looked at as swingable versus those which never move despite appearing competitive).

5. It would also be very cool if you could write a program which will use your newest regression model and your newest pollster weightings after all of the primary results come in to regress what the polls actually said about the general election win percentage each day (or week) since Wisconsin. If you could have this automated then you could simply re-run it at any time for any set of polling data or any election really and I feel it would be a powerful analytical tool for prescribing campaign tactics in general.

6. Free frogurt for your long-time readers.

Anonymous said...

7. Oh, I almost forgot this but that's because it would be more as a gag as per your earlier post, but you could include the chance of each state necessitating a recount and then calculate the chance that a recount would be needed anywhere in the country. You could even consider this for one of your six pie chart spots at the top of the right column. :)

Anonymous said...

Your projection for SD is off by 31 from ARG (Clinton +26) and your projection for MT is off from ARG by 14 (O +4).

Using my sophisticated analysis (HA!), I find that ARG has polled 23 states and their mean error in favor of Clinton has been 10%. The median and mode are both 7% errors in the margin in Clinton's favor (IN, MD, PA, and WI). The range has been 17 off in Obama's favor (NH) to 26 off in Clinton's favor (SC). In six states, they've overestimated the margin in favor of Obama (NH, RI, VT, CA, MO, TX) and in 17 they've overestimated the results in Clinton's favor.

In four contests they predicted the wrong winner (NH, DE - off by 13, CT - off by 17, and IA - off by 17).

So in order for your analysis to be correct, ARG would have to have their most colossal blunder yet in their SD prediction.

I'm guessing the median/mode kicks in and it's Clinton by 19 in SD and Obama by 11 in Montana.

My analysis took much less time and energy than yours -let's see who's right:)

Pablo

Anonymous said...

One more comment: I am tired of seeing analyses, such as one in Huffington Post saying that Obama won because of caucuses or in 538 saying that Clinton is winning the GE simulations. These all presume that everything else is static. Obama won in caucuses, yes, but he also diverted attention to caucuses during Super Tuesday instead of working to the end in California. He might have had a different strategy if caucuses weren't there. The point is that he gave it thought, showing more nimble strategy that would serve a president well.

And the GE simulations also presume much. First, her supporters are pissed off now - I can't for the lift of me figure it out - it's a whiny bunch. First they say if you can't take the heat, stay out of the kitchen, I love the heat, this is the fun part, etc, etc, but when they lose, they start crying about how unfair it is. For the love of God, they are working hard to permanently alienate their supporters for Obama. But some of that will change and when it does, the GE prospects for Obama will improve. Obama's supporters aren't as unhappy, and thus tell pollsters they'd gladly vote for Clinton if she was the nominee.

Why that isn't part of the analysis, I don't know.

Pablo

hosertohoosier said...

Pablo,

That is a fair point on caucuses (especially on Super Tuesday), except that some states have had caucuses and primaries. In Texas, those were held on the same day - Obama won the caucus by 12 and lost the primary by 4.

Anonymous said...

That's in part because Limbaugh only instructed his supporters to vote, not caucus.

Let's face it, Limbaugh turned both the Texas and Indiana races in her favor. Without that, this race is over much earlier - Limbaugh has done exactly what he wanted to do - prolong this race.

carl29 said...

That ARG poll is joke. I think that indeed Hillary's best shot is South Dakota, the 8th oldest state in the nation. But 20+ spread seems pretty odd. I think that her supporters may be energize today, trying to see her going on a "high note." I think that it is close because Hillary's people are not talking about mar