Election Day 08

Hi everyone, this is Rob! I've been enabled as a guest blogger on Becky's blog ever since I took over parental leave from her two weeks ago, with the intention of giving updates on Clare's further adventures and my parenting activities. If you're a regular reader, you'll note the absence of any such entries from me before today (and as you might guess from the title of this entry, I'm not going to be writing about any of those things here). Before I move on to the main topic of this post, I'll give you a quick Clare fix. Clare is doing really well, becoming more mobile and active by the day. I've noticed during our playtime that she really becomes focused on whatever I'm doing; if I'm playing with the blocks, she inevitably abandons whatever she's doing to play blocks with me (and by "play", I mean "tear apart whatever I've built and start chewing on a block"). If I move on to open up a book, she abandons the blocks and comes to flip pages with me. As Becky mentioned before, she really is quite a social little girl, and I'm really beginning to think we need to find more opportunities for her to play with other kids. Besides that, she's also becoming more and more adept at feeding herself the little chunks of bread and vegetables that we give her at dinnertime. No more taking whole chunks of food into her little mouth; for the most part, she's learned to bite off manageable chunks of food to chew before taking in some more. You may remember that Becky and I have been trying to keep her diet gluten-free until she's a little older. To that end, we've been feeding her mostly rice and rice pasta/noodles for her carbohydrates, as well as spelt bread and couscous. If you'll follow those last two links and do a text search for gluten (as we should have done), you'll see that we haven't been that successful.

As for the actual subject of this post, today is the first Tuesday following the first Monday in the month of November in an election year (for the U.S.A), meaning it's Election Day south of the border. While we had our own election just a few short weeks ago, for me, I find U.S. politics to be much more engaging and interesting to follow. Every four years, when this first Tuesday in November comes around, I spend most of my free time for the day watching the American networks and cable channels and their election coverage. I still remember Bush/Gore 2000 and spending the whole night watching the returns come in, seeing most of the networks call Florida for Gore even though Bush was ahead in the actual count (on the strength of uncounted returns from heavily Democrat counties), then having them retract those calls and going to bed not knowing who won. On the other hand, I usually flip back and forth to CBC and CTV on Canadian election nights to keep up on the current tally, but otherwise spend most of the night in my normal routine.

If you've been following the election coverage for Obama/McCain 08, you'll know that all of the national tracking polls show Obama with a comfortable lead (anywhere from 5 to 11 points). Of course, if you know a little about how American presidential elections work, you'll know that the popular vote (which the national polls are attempting to track) aren't the whole story, but such a sizeable lead almost always will translate into a comfortable lead in the electoral college. FiveThirtyEight.com, a political blog run by a baseball statistician (and admitted Democrat partisan) that takes the available state and national polling data and runs it through a proprietary election model (based on historical polling and election results), currently predicts that Obama has over a 98% chance of winning, with an expected electoral college tally of 346.5 (of the 538 available). Not quite landslide territory (a landslide would be around 400), but a comfortable win compared to the last two Bush wins.

If you're a non-American (which I assume is generally true for people reading this), I'd guess you'd consider this pretty good news, since most of the rest of the world favours him by a wide margin. However, if you've been following the election coverage at all, you may have noticed some Republican partisans claiming that the polls are inaccurate, as well as Democrat hand-wringing amongst those who won't feel safe until their man is sitting in the White House. A lot of people remember the exit polling from 2000 and 2004 that seemed to point to a Democratic win, and so there's an undercurrent of residual mistrust of the polling data that's out there. There are also complicating factors like the so-called Bradley effect and Shy Tory effect that Republican partisans believe may overstate the Obama support in these polls. However, while the presence of these effects seems to be somewhat dubious (and possibly counterbalanced by the cellphone effect that should favour Obama), it seems that if the polls do indeed get it wrong and McCain pulls out a win, it will be because of how the pollsters model the electorate in coming up with their "top sheet" numbers.

If you've ever hung up on someone calling you to get you to participate in a poll, you can probably guess that the it may be difficult for a pollster to get a representative sample for a given poll. Ideally, if you polled 1000 people, you'd like for those 1000 people to proportionally represent the expected or likely voting population on election day. Two obvious ways you could break down this sample are by age, and by party affiliation. So, for example, out of 1000 people, you might get 50 people from age 18-24 answering your poll. If, in your likely voter model, you expect 100 out of every 1000 voters to be in that age range, then your poll may be undercounting Obama support (since Obama tends to dominate in this age range). Similarly, you may get equal numbers of registered Democrats and registered Republicans to respond to your polls. If your model suggests that more Democrats will vote on election day than Republicans, then your poll again underrepresents Obama's support.

Of course, almost every pollster doesn't just take the raw results and use that for their top sheet. Almost everyone will take what they have and then weight the responses by their likely voter model. And this is where a lot of the angst about the polls (on both sides) tends to come from. I won't go into details about why this is, but Obama has had a huge advantage in terms of money to spend on his campaign compared to McCain, and has spent a lot of money on registering new voters, and organizing GOTV (get out the vote) efforts to ensure that these voters actually vote on election day. Because of this, many pollsters have been using a likely voter model that leans Democrat (anywhere from D+3 to D+9, i.e., Democrats having between a 3 and 9 percent larger slice of the likely voter pie). This flies in the face of what has been happening for the last 20 years, where turnout has been generally even, or leaned Republican. Most pollsters have also been expecting a higher youth turnout because of Obama's GOTV organization, a group that has been historically difficult to motivate (I still remember people trumpeting the youth vote as the reason that Kerry would win in 2004). The upshot is that if the actual turnout doesn't match what the pollsters have been predicting, this is going to be a lot closer election that the polls seem to indicate.

Kudos to you if you're still reading after this long dry post on election modeling. I think this stuff is pretty interesting, and I find American politics interesting in general. I'd actually intended to write something a little different rather than the somewhat dry explanation about polling above. It's really kind of interesting to see Obama take a page out of Karl Rove and George Bush's winning campaigns from the last two elections (Rove really focused on GOTV and rather than winning independants -- Obama hasn't ignored independants, but the so-called "ground game" and getting his supporters to the polls has been a major focus for his campaign). It's also interesting to see how McCain completely abandoned his centrist appeal and went "Rovian", trying to get the Republican base on his side with the Palin pick and the negative rhetoric. Like I said above, Rove really won Bush's last two elections by motivating the base and ensuring that they all showed up to the polls. With Obama likely to win 90+% of the Democrat vote, and McCain likely to do the same with the Republican vote, it pretty much will come down to which party turns out in larger numbers.