Buy High, Sell Low

[koganbot]

Strange that stock investors seem to be reacting to news that confirms that we're in a recession. Did anyone who pays attention to the economy think that we weren't in one? I wish I had money to invest and enough cushion to sit on my investments, on the assumption that what I buy will be higher in five years.

I live in Colorado, a battleground state, so I've been getting at least one anti-Obama robocall a day from the Republicans; a week ago the theme was that he hangs with terrorists, now it's his inexperience, and of course every few days it's that he's going to raise taxes. Interesting, though: a month ago I was getting a lot of anti-Udall robocalls - Mark Udall being the Democrat running for Senate - mainly about his nefarious plans for spending my money, but those calls have disappeared. Perhaps the anti-Udall money has simply shifted to television. But I'm also wondering if the Republicans have conceded the Senate seat while still thinking the presidential run is in play.* There've been stories of the McCain campaign pulling out of the TV market here, but McCain will be in my city today, and my guess is that even if he's drastically cutting back in Colorado, he doesn't want to make it appear as if he's giving up on any more states.

Obama is being kind of a shithead too, by the way. His anti-McCain sound bites have been that McCain wants to cut taxes on companies that are sending our jobs overseas. The first half of that message is OK, but the second half, about jobs being sent overseas, is irrelevant. I mean, the McCain plain doesn't specifically give extra tax breaks to firms that outsource overseas, does it? (I haven't actually read his plan, if such a thing as a "plan" even exists beyond "I will extend the Bush tax cuts and cut capital gains." I don't know. I'm not actually a well-informed voter.)

I have no expertise when it comes to opinion polls, but based on what I've read from people who sound like they do, here's the situation: if Obama wins all Kerry states plus Iowa, New Mexico, and Colorado, he's won, no matter what happens in the tighter states. Or if he loses New Mexico and Colorado but wins Virginia, he's won. All of these are likely to go to Obama, with Virginia and Colorado being the most vulnerable (FiveThirtyEight projects him up in Virginia by 6.8% and in Colorado by 6.3%). I do worry about the fact that Colorado's Republican Secretary of State purged thousands of voters from the voter rolls, some of the purgings being illegal according to the New York Times, but I assume Obama's lead and the number of people he has on the ground to contest the purgings will overcome that, as will the fact of so much early voting here (if you try to vote in advance and find you've been purged, you've got extra time to do something about it).

Anyway, taking both Kogan states (I live in Colorado and my brother lives in Virginia) would be a daunting task for McCain, and it seems likely that this is not going to be his focus, even though, as I said, he'll be in Denver today. But where he does seem to be pushing hard is in Pennsylvania, one of the Kerry states. This would appear to be an even longer shot - FiveThirtyEight puts Obama up by 10.7%. Throwing resources at this state would seem suicidal. But I've read some interesting and nonridiculous rationales for why McCain's doing it. One would be that he needs to energize his campaign, so saying, "Look, taking every single state that's been touted as swing isn't our only path to the White House, since we think we can take this one." Knock down the idea of Obama inevitability. Second is that maybe some of his own polls show the state tighter than the publicized ones do. Third is that the primary showed Obama vulnerable among white working-class Democrats, especially in the western section of the state. And fourth, Pennsylvania, unlike Colorado and Virginia, has almost no early voting; to get an absentee ballot you've got to show that you genuinely need one. So if you're McCain you've got more potential voters in play up until November 4, hence more time to swing them. He can figure that the three-and-a-half point further behind he is in Pennsylvania than in Colorado is more than compensated for by the fact that he's got more of a chance to gain ground in Pennsylvania - a population more vulnerable to his appeal, with few of the voters preempted by already having voted. (I read an estimate that said something like half of Colorado's votes will already be cast before election day.)

So, let's expect the following. McCain runs hard in Pennsylvania, with racially and culturally coded references to the fact that "Obama just isn't like us" and "He doesn't see the country in the way that we do." Now, indications are that such appeals have backfired so far overall, but if you target them to a narrow enough group of voters, maybe you've got a chance to sway those particular voters, not to mention their near neighbors in Ohio. OK, I really doubt that this strategy is going to work, but let's say it does. McCain lets Virginia and Colorado go but wins Pennsylvania and all the Eastern tossup states (Ohio, Indiana, Missouri, and Florida).

OK, remember what I said a couple of weeks ago, about how in this Internet and 24-Hour News Age, what happens in western Pennsylvania doesn't stay in western Pennsylvania. An appeal with a racial tinge will get out beyond its target audience, but this won't necessarily hurt McCain in the Eastern tossup states. (Well, I think it will, actually, but for this scenario to work we've got to assume that it will help him more than hurt him.) OK, but what about the Western tossup state? By which I mean Nevada. (If North Dakota or Montana ends up being a tossup state, then lets face it, McCain is being swamped and it's all over.) Now, I don't know Nevada, even though I sometimes write for a Nevada publication. But having lived a decade in Colorado, I do have this sense of how things might be. Among undecided voters, you're going to have a lot of registered independents with libertarian leanings who have trended Republican in national elections but are not happy with the party's totalitarian and big-spending impulses under Bush. And you'll have conservative-leaning Hispanics who nonetheless don't trust the Republican Party in its attitudes towards Hispanics (though I must say that this is one issue where you can't put too much blame on McCain and Bush, both of whom have tried to be somewhat reasonable on immigration). And you'll have Mormon/LDS and evangelical Christians who are social conservatives on, say, abortion but not on issues such as poverty and race. I don't know if there are a lot of these, but those who are undecided are likely to be the ones who like the fact that Obama is a Christian who talks about reaching across social barriers. And they won't hold his being black against him in the least, any more than Jesus would. Again being someone who's never been to Nevada, my impression nonetheless is that black-white isn't a big issue, that there aren't a lot of blacks and that fear of black culture and black crime isn't what makes Nevada politics run. And Nevada can elect Democrats, e.g. Harry Reid, the Majority Leader of the U.S. Senate.

So, what happens in western Pennsylvania doesn't stay in western Pennsylvania, in fact makes it to Vegas etc. And I'm surmising - hoping - that it hurts McCain there.

OK, here's the electoral vote if McCain takes Pennsylvania plus the eastern tossup states but loses Virginia and Colorado (and Iowa and New Mexico, which he's almost sure to): McCain 268 to Obama 265 with 5 electoral votes left up for grabs in the western tossup state of Nevada. If Obama takes Nevada, he's won, 270 to 268. Right now FiveThirtyEight is projecting Obama plus 2% in Nevada. (Also shows him about that much ahead or more in all the tossup states not counting North Dakota and Montana, but that's a different story. RealClearPolitics has Obama plus 3.3% in Nevada. Not totally sure what those numbers mean. FiveThirtyEight has a "snapshot" number of Obama up by 3.6%, and I'm guessing that that's the number they're getting if the election were held today, but that they work in some other assumptions to make their projection of 2%, possibly going on how much past elections have tended to tighten near the end.)

Not that I expect this scenario to happen. I don't see how McCain makes up 11 points in Pennsylvania in eleven days (or even, say, 5 points in eleven days, if the race is closer than the polls show) when he's not been gaining in the last month. But if an inordinate amount of campaign dollars starts pouring into Nevada, which has a relatively small population and few electoral votes, this scenario might be why.

NEWSFLASH: I just got my first Democratic robocall, from Senator Ken Salazar, asking me to vote no on some ballot initiatives I've as of yet paid no attention to.

*FiveThirtyEight has Udall ahead by 8.8 and Obama ahead by 6.3, which isn't that big a difference.

Comments

[alexmacpherson.livejournal.com]

I've actually been secretly wishing it was a little bit closer - just a little bit, not too much! - so that our all-night election party would be more suspenseful.

Then again we got a suspenseful party in 2004 and look how that turned out, so I'll take Obama's relatively huge lead. (I can't remember what the polls looked like before the '04 election.)

[alexmacpherson.livejournal.com]

(I vaguely worry that the Bradley effect will come into play, though my flatmate pointed out that it was conceptualised over 20 years ago and hopefully people have moved on a bit since then.

[koganbot]

Well, remember that opinion polls have margins of error of about 4 or 5 percent, and that's not accounting for whatever wrong assumptions are built into this or that poll. State opinion polls can be especially unreliable, and FiveThirtyEight and RealClearPolitics, which try to collate a multitude of polls, point out that some of what they're working with on a state level can be days or weeks out of date. West Virginia seemed to be in play, but that was owing to a single poll that was a couple of weeks old, and when new polls came out, they showed McCain ahead by about 7% (though one still has the state a tie). On the national level, polls are all over the place, depending on how they go about deciding whether someone is a likely voter. One poll has Obama by 2%, another by 14%.

But if you need something to be suspenseful about, here it is: To get a filibuster-proof majority in the Senate, the Democrats need 60 seats. Right now they have 49 plus two independent Senators who caucus with them (though one of those, Joe Lieberman, has been shifting right for years) for a total of 51. So they need to take nine Republican seats; right now they've got two for sure and another two very likely and two more where they're definitely ahead (Oregon and North Carolina), so that's six. There are two tossups (Minnesota and Alaska) which are tilting slightly towards them, which puts them up to eight, one short. There are three states leaning Republican where the Democrats have a shot: Mississippi, Georgia, and Kentucky. (There are two Senatorial contests in Mississippi owing to a vacancy; one is sewn up for the Republicans but the other is closer.) So anyway, that's five contests to get all upset about: Minnesota, Alaska, Mississippi, Georgia, and Kentucky. And two states where the Democrats are leading, North Carolina and Oregon, are hardly locked up. So you're up to seven, a few of which we can assume won't be decided until the wee hours of the American West's morning, if even by then. So you have plenty of opportunity for drawn-out suspense. Also, historically, races have tightened in the last two weeks before voting.

[skyecaptain.livejournal.com]

The NRSC just pulled operations from the Colorado Senate race, which might explain the lack of anti-Dem calls there.

[skyecaptain.livejournal.com]

A lot of people have mentioned this already, but (1) the Bradley effect may not have been a factor even with Bradley (poor polling and a much tighter race than this) and (2) there are a ton of reasons for people who wouldn't vote for Obama to give aside from his race if they were so inclined. I'm optimistic enough to think that there really are far fewer people who actually, deeply care about his race anyway, but even given there ARE enough people to make a difference, they still probably wouldn't be reporting Obama, they'd say McCain and give a reason other than race.

[dubdobdee.livejournal.com]

this is totally from memory, from many days ago, but i believe i read the "tax breaks on shipping jobs overseas" refers to a broader currently existing tax break, which it was suggested be removed from those who shipped jobs overseas (as a way to discourage outsourcing): by voting this rmeoval down, mccain was indeed for "tax breaks on shipping jobs overseas", though it's obviously a slippery and misleading (and thus demagoguey) way of saying it

[dubdobdee.livejournal.com]

MoEs depend on the relationship of the size of the sample to the size of the population in question -- they're by no means all necessarily 4-5%

US polling generally uses smaller (relative) samples than UK polling, interestingly enough: it's been argued this is actually because larger samples, producing more accurate (and therefore -- barring unexpected gamecanging events -- less volatile) polls, as well as being practically more expensive, aren't really what news media need (it's in the interests of competing news services to serves up a constantly shifting narrow battle, with a variance of results between stations, because then people tune in night after night; and stations can compete by producing "news the customer wants to hear") -- though it's not clear why this market logic operates in the US but not the UK (possibly the proliferation of state-level media would be a causative factor)

there are three other factors causing error -- the results a sample gives are weighted in reference to previous elections (which is obviously unhelpful in realignment or sea-change moments); there is no way really to take accurate account of big upswings in voter registration (this doesn't apply in the UK, where right to vote is assumed, providing you have proof of address; the US has much more stringent requirements and potential for challenge and disbarment); and the considerable regional variations in how the voting mechanism operates means there all all kinds of low-level ratfuck voter suppression and/or bamboozlement tactics

i rather like the stratego effect of the electoral college -- the "different ways to get past 270" element is very like a boardgame... but it's wilfully capricious as a system "all can agree on as fair"

[koganbot]

Yes, you're right that they're not all 4% to 5% off, and if a whole number of different polls show, say, Obama ahead by 5%, that probably is pretty accurate, though of course in reality the different polls are all over the place. FiveThirtyEight is interesting reading because the guy isn't doing the polling himself, he's collating other people's polls and therefore is very concerned with weighing them relative to each other, assessing their probable accuracy. And he writes about this, isn't trying to hide his process.

the results a sample gives are weighted in reference to previous elections (which is obviously unhelpful in realignment or sea-change moments); there is no way really to take accurate account of big upswings in voter registration

Well, pollsters can try to take such things into account. If a state has registered another 200,000 Democrats and has lost 25,000 Republicans, you obviously aren't going to assume that the parties balance the way they did in the last election. What's really tricky is figuring out who's a likely voter, since shifts in enthusiasm and decline in cynicism among, say, black voters could mean that rare voters and new voters may be more likely to vote this year than in previous years, a trend that is likely to help the Democrats. But the voting in primaries can give pollsters at least some idea of what's likely to happen on November 4, as can asking the voters. FiveThirtyEight talks about such issues here:

Some Likely Voter Models are Suspect

[mostlyconnect.livejournal.com]

simple poll average in 2004: http://election.princeton.edu/2008/10/24/on-the-track-record-of-simple-poll-aggregation/#more-1976