Oregon is a uniquely convenient state for Barack Obama to have as a firewall. It's not really the state's demographics -- which are, statistically speaking, very white and relatively working class (Oregon is below the national average in per capita income). Rather, it's the fact that all of Oregon's voting is conducted by mail.
EDIT: I've been giving out some bad information in this thread. It occurs to me that Oregon has a Republican primary too, so that's why some of the numbers seemed so high. Certain passages have been corrected.
As of yesterday, the state had already received 360,219 ballots. We don't know precisely how many of those are Republican ballots and how many are Democratic ballots, but I'd guess that over half of them are Democratic ballots; perhaps somewhere in the 200,000 range. That's out of about 800,000 registered Democrats in the state, and total turnout that should fall somewhere in the range of 600,000. And these are only the ballots that have been received; ballots are coming in at a rate of about 80,000 per day. Oregonians can't afford to procrastinate, by the way, because a ballot must be received by election day -- the postmark doesn't matter (although there are special drop-boxes available on election day itself).
So let's say there's some big event that takes place over the weekend and swings momentum toward Clinton. It won't matter, but for the small percentage of Oregonians who use the drop-box option (in 2004, about 13 percent of the primary ballots were received on election day itself, or about 29 percent of the total returned). For all intents and purposes, the election in Oregon is about two-thirds over. That's why Barack Obama is campaigning in Michigan today; the election is literally in the (mail)bag.
EDIT #1: Here's something else kind of fun: the final appeal made by each candidate in the Oregon Voter's Pamphlet.
Clinton hits you over the head with resume and facts, whereas Obama's approach is less prosaic. If the two candidates were in college together, you could imagine Clinton getting really mad at Obama because she spent days working on her paper, and he whipped his together that morning and wound up with the better grade.
Wednesday, May 14, 2008
Hillary's Oregon Problem
Subscribe to:
Post Comments (Atom)
27 comments
At least in Oregon they have to be received by election day. Here in Washington they need to be POSTMARKED by election day. WHAT A MESS!!!
Your analogy is great! Clinton as the butt kissing overachiever vs. Obama as the natural who just gets it. Very astute.
I love how First Lady is governmental experience.
Then again Obama could have a problem with the balloting issue that has arisen there. Newly registered voters have gotten both a partisan and non-partisan ballot and only one counts. Who though Florida was the only state with an over-voting problem?
It looks like there's a serious problem with your regression model's estimate of turnout. While your numbers have been incredibly close percentage-wise with respect to the vote (congrats, by the way), that could substantially change if your model doesn't start to take *number* of votes more seriously. After all, a 41 (or even 39) percent win with 100,000 more votes is a significantly bigger win numerically speaking. Thus, being correct at 39% with 105,000 vote margin is very different from being ahead at 41% with a 140,000 vote margin. Your model assumed a turnout of around 275,000, yet the actual turnout was closer to 360,000. That means you the actual turnout was a third higher than you expected. I really enjoy your site and I hope you keep up the good work, but I'm wondering if there's any type of variable you could include to remedy this problem?
I could be wrong but I don't think the winning percent prediction is integrally tied to the turnout prediction. This isn't a system of simultaneous equations. Nor is the model predicting the number of votes received by each candidate as the dependent variable; rather it's predicting the percentage of votes won by each candidate.
I think Poblano mentioned in a comment last night that he had goofed on the turnout estimate in WV because he used as a baseline the 2004 Kerry vote instead of the much larger number that includes registered Democrats (many of whom voted for Bush). I don't think getting that "right," however, would have affected the predicted percentages for the candidates. Implicitly, I believe, the model assumes that turnout in the current election will be similar to that of the previous elections that are in his baseline.
(I could be wrong about this but this is what I think is going on.)
The biggest problem with Hillary's plan at this point is that it hinges on the people of Oregon hanging on the result of the primary in West Virginia--and throwing out the results in nearly all of the other contests.
Good luck with that.
The vote share model is really totally separate from the turnout model. The vote share model says: "Obama will win 46% of the vote in this Congressional District". And then we multiply 46% by whatever number of votes we expect there to be to come up with his vote total.
If there are only 800,000 registered Democrats in Oregon, how do you figure turnout will be up to 1.1 million? Is this a closed primary or not? Assuming 66% turnout (compared to 59% in PA), Turnout would be less than 600,000.
preader,
One of the MSNBC pundits -- Chuck Todd, I think -- observed last night that the Clinton people have sort of stopped talking up Oregon, which they'd been doing before IN/NC. And they have Bill campaigning today in Montana and South Dakota, and Hillary meeting her donors, whereas this is *really* critical time in Oregon if you think you have a shot at the state.
I would tend to agree that Oregon would the last chance for them to really swing momentum, but if their internal polls show them way behind among people that have already voted, there's physically not much they can do to change that result. And I think this is one case where the actions of both campaigns speak louder than their words.
SUSA poll says they are tied among those who have already voted as of May 12, 2008, and that Obama's margin is made up entirely of likely voters in the week to come.
http://www.surveyusa.com/client/PollReport.aspx?g=94e9005d-d8d6-46e2-b09a-697a4e23a900
Poblano, you can't just multiply a straight percentage across congressional districts regardless of turnout for a couple of reasons. They simply just aren't independent variables.
Suppose a state has only 2 districts: (1) District 1 which had 95% turnout in 2004 and (2) District 2 which had 65% turnout in 2004. A greater turnout across the state would heavily favor District 2's winning candidate because there's no room for District 1 to improve meaning it must have been District 2 that added 10-20% more voters. For example, if the race was a dead heat with District 1's candidate winning 52/48 with the old turnout numbers, greater turnout could easily result in as much as a 10 point swing to District 2's candidate, thus completely skewing the election results from the prediction model.
Additionally, let's suppose District 1 and 2 had the same number and percentage of voters in 2004. We can't just apply an equal percentage across independent of turnout because that assumes the new turnout would have the same ratio of voters per candidate as the old turnout. For example, if the turnout increases 10% in both districts, we can't assume the turnout demographics and preferences match the previous 60% that we expected (indeed you yourself admit that additional turnout is likely to favor youth and black americans, which, in turn favors Obama over either Hillary or McCain).
I think for these reasons, the turnout model should be incorporated in some way to the percentage based model.
Interesting bios. Never knew "wife" got counted as executive experience until Hillary told me. Guess that means I can glom some experience from my lawyer husband. Not impressed by the woman that told her high school valedictorian she was smarter. Unbridled lust for power is highly unattractive and ineffective. Gah
Oregon's a funny state. We have a history of electing opposites, e.g. Hatfield and Packwood who basically canceled each other out in the Senate. We are geographically large, with different economies in each region, and could easily be several different states. We have (in the larger urban/suburban areas) a ton of young, college-educated, entrepreneurial-type people who moved here from other states because they see Oregon as a place of opportunity, natural beauty, and openness. Obama appeals to our sense of hope. Then we have the cattle ranchers, out-of-work loggers and fishermen, more hippies than most folks have seen since the 60s.... Anything could happen.
On the turnout/share relation question.
You calculate the two separately. The question is, how can we have a good prediction of share when we cannot predict turnout?
There are several possible answers.
1. In the world (and not only in Poblano’s modeling) the two are independent. When turnout goes up, so do the shares of both candidates, by the same ratio. You can be bad about predicting turnout and still good about predicting share.
2. In the world (though this was not captured correctly so far in Poblano’s modeling) turnout did not change. The disposition to vote among different groups was exactly the same in pre-PA as in PA-and-later votes. Thus it is immaterial to try to calculate turnout (as it is a constant, not a variable).
3. In the world, turnout did go up, but share remained the same. This can be accounted for in two ways (I recap from a comment made to the WV results thread):
3a. Same as 1 above.
3b. The “constant share” observation is in fact an illusion: the errors in predicting share and in predicting turnout cancel each other out. Since in this case it seems extremely likely that higher turnout favors Clinton, this works out as an underprediction of Obama share canceling out an underprediction of turnout. I.e.: in the world, disposition to vote overall went up, and so did disposition to vote Obama.
I have the sense you, Poblano, believe in 1. But it just doesn’t seem right conceptually and seems indeed to be at least somewhat refuted empirically by the evidence of Obama’s overperformance in small turnouts (i.e. caucuses). Thus it is quite important to know whether 2 is right or not. Is there a way of going back to data and tease out in some way the “Reality” of disposition to vote? Perhaps finding very similar counties that voted before and after March?
Houston,
Obviously to combine the results at the statewide level you need to know the turnout in each district. But as long as the model gets the *relative* turnout right between the various districts in the state, the statewide vote share prediction will not be affected.
Thanks for this, Poblano. I hate to be catty but looking at my voters pamphlet I noticed that HRC's photo appears to be retouched. It's a lovely photo, but she looks to be about 30. The photos are supposed to be taken within the last year and un-retouched. Oh well...
Anyhow, I hear that both candidates will be back in the state this weekend. Based on the wholly unscientific measure of how many Obama to Hillary signs are up here in Portland (about 20 to 1), I do expect him to win by at least 10 pts. He will dominate in the population centers but I hear is also doing well in rural areas. I hope he will take another swing thru central/eastern Oregon. He gave great, specific, local-issue-focused interviews to our local media when he was here last.
I also noticed that HRC looked really young in the photo. But another interesting thing to me was the fact that in her picture she is looking somewhere out there, while Obama is looking straight at you in the picture: very interesting contrast and to me very telling of who they are.
For John Putnam, who posted "Newly registered voters have gotten both a partisan and non-partisan ballot and only one counts. Who though Florida was the only state with an over-voting problem?": Only one ballot can be voted. The voter signs the outside of the outer envelope. When the ballot comes in the signature is checked against the one from the voter's registration, and the voter is marked off as having turned in a ballot. Then the inner envelope, which contains the ballot and has no voter ID information, goes on to be counted.
I don't think there's anything wrong with Hillary listing First Lady as political experience. She certainly had serious a policy role, unlike other First Ladies. The problem in the campaign was that their vastly exaggerating how much of policy role she had, not in claiming that she had one at all.
As an Oregonian living in the major swing county (Washington) in the state,let me address some "concerns" mentioned in earlier posts.
First, Oregonians are quite comfortable and educated on our voting system. Many folks wait until election day to drop off their ballots in designated secure sites and/or sit in long auto lines at the election dept(s) to drive through and drop off their ballots ala April 14th...tax day. It generates a weird sort of community spirit not unlike what others say they feel by voting at a polling site. More than likely, turnout will exceed 70% of democrats according to various sources if not higher. Obviously, interest is high here for our primary which hasn't accounted for much in years.
Second, yes, the Secretary of state screwed up by sending out two ballots for (only) persons switching their party affiliation. This has been highly publicized in the media so any worries of that hurting Obama is misdirected. The county elections offices have an almost fool-proof plan to solve this problem. I won't go into that in this post.
We Oregonians have embraced our version of vote-by-mail. I no longer miss signing in at the polls and going to a booth where there are distractions and or polling problems (think Ohio and Florida on numerous occasions). Voters here have up to two weeks to study our Voter's Guide (a wonderful resource), make your decisions about candidates and ballot initiatives and mail in or drop off your ballot. Gosh, you can vote over a period of days if you wish before turning in your ballot.
Almost all ballots are counted on election day prior to the "Polls closing" and a winner will be known soon after the "polls close" at 8:00 pm on May 20th. At least 50% of state-wide vote totals should be available within the first hour after the cut-off. It's a great system with little room for fraud despite what right-wing bloviators profess on teevee.
It has proven to boost turnout....but we know that Goopers don't like that.
Looking at those Voter Pamphlet appeal texts, the main thing that jumps out at me is:
Paragraph Breaks!
And the second thing is: Don't break words with hyphens!
My god, don't the Clinton campaign staff know a high-school-level writing instructor? She may lose a few fractions of a percent just for submitting that indigestable mess.
As of yesterday, the state had already received 360,219 ballots. We don't know precisely how many of those are Republican ballots and how many are Democratic ballots, but I'd guess that over half of them are Democratic ballots; perhaps somewhere in the 200,000 range.
If voters return ballots in proportion to their registration, your math is probably good. As of March, about 53% of the state's voters were registered Dem. The number jumped some more after that, but probably not a lot more than the 56% you mention.
On the other hand, there's very little reason to think that they are coming in in proportion. Dems have the presidential election, a US Senate primary, statewide primaries in AG, Secretary of State, Treasurer, and OR-5. Republicans aren't running candidates in AG or Secretary of State, have a Senate incumbent, and no presidential rooting interest.
So it seems unlikely that things are playing out proportionally, nor will the final tally look proportional.
FWIW, the total population of Dems in Oregon is something on the order of 850,000. So for the popular vote tally, your over/under's probably 75,000. Obama could gain more, and it would be considered a big loss if he didn't gain that many (that's 10 points on a population of 750,000).
As another long-time Oregonian, I can say that most voters do not make their decisions regarding the Presidential race based on the Voter's Pamphlet--- the race is so late in the season that most voters have already decided. The pamphlet is used primarily for learning more about candidates for state elections, as well as ballot measures.
Obama will most likely win Oregon by at least 10 points because he plays relatively well in "downstate" counties such as Marion and Linn that are key to negating large margins from metro Portland. Expect him to dominate in the college influenced counties of Benton and Lane, both of which went for Jesse Jackson in '88 when Dukakis had already effectively clinched the nomination.
Those areas in Southern Oregon, where she might be expected to do best (Douglas, Coos, Curry, Jackson, and Josephine counties), are also areas where the timber industry remains a major employer and Bill Clinton is still seen as an even-handed player during the Timber Wars of the early '90s.
However the statewide Democratic division from that era over logging policies is precisely why many of these former Democrats switched to voting Republican, if not changing their party registration altogether. This decreases the chance of any huge surge of Hillary votes from these areas, despite respect for Bill Clinton pushing through "Option Nine" as a compromise on the Spotted Owl and keeping Federal money coming in to rural areas whose tax base was decimated by decreased logging levels.
The Portland suburban counties of Clackamas and Washington will be a key battleground, but again even if Hillary ekes out a small win there, it will not be enough to come close to carrying the state.
Dear Friends
After Hillary's 41-point landslide win in West Virginia, she's proven that she is the candidate who will win the critical swing states, get to 270 electoral votes, and beat John McCain in November.
Now we turn our sights to Oregon and Kentucky. Voters in Oregon are mailing their ballots now and voters in Kentucky are preparing to head to the polls this Tuesday, May 20. Hillary needs your help in both states, and we need you to connect personally with Oregon and Kentucky voters and let them know they this race is up to them -- not the pundits.
When our best supporters like you use Activate to make calls to voters, it makes a huge difference in helping us get every last vote out for Hillary. And our Activate dialing system is so easy to use -- it makes the calls for you, meaning you can talk to more voters in one two-hour shift. Hillary needs you more than ever. Can you sign up to make calls for at least one two-hour shift?
Click here to sign up for a two-hour shift.
Over the next five days, Hillary and her senior staff will hold a series of nightly conference calls with our Activate volunteers to discuss how the calls you make are a critical part of our plan for victory. We'll also provide you with training for the Activate system if you haven't used it before. We'll tell you how to join our conference calls when you sign up for a shift.
Click here to join us -- sign up for a calling shift today!
As president, Hillary is going to fight for every one of us every single day. It's up to us to fight for her now.
Thanks for all you do for the campaign.
Sincerely,
Aaron Ver
Deputy National Field Director
Hillary Clinton for President
M Waheed Jadoon
WORLD DEMOCRACY MEDIA GROUP
New York
It's a democracy (or supposed to be). Vote for who you wish... but be informed. There are things you probably don't know about Obama.
http://www.talkleft.com/story/2008/3/23/224059/069
http://www.houstonpress.com/2008-02-28/news/barack-obama-screamed-at-me
http://www.youtube.com/watch?v=bqH-BB6brKg&NR=1
http://www.youtube.com/watch?v=Juy9NwI8_i0
http://www.youtube.com/watch?v=q-q9w_H2ACA
http://www.youtube.com/watch?v=2kkGGqufu2c&feature=related
for a few...
Also be aware that the media is extremely biased and very much against Clinton, as are online polls and posts. Moderators frequently do not allow Clinton supporters to vote in the polls or post on blogs... the machine is at work... looks like we might be in for another "appointed" presidency, but at least this time they will make it look more like the guy was elected.
Post a Comment