Monday, November 10, 2008

Early Analysis (Too Early for the Facts)

It's not like any of the pundits worried about facts before the election.

The Tribune dropped the Perspective section this week; the columns now appear at the back of the tabloid-format news section on Sunday. I just don't think my stuff looks as good in a tabloid format, I'm more of a broadsheet kind of guy, but it's a "transforming" industry. Hey, at least newspapers are more profitable than the banks.

Results still aren't final in AZ, and won't be for another week. Turnout should creep up, but there's no way it'll get to pre-election estimates and I strongly doubt it'll reach 2004 levels. The third Corporation Commission seat is still undetermined; the Democrat in third place had it Friday afternoon, but additional votes over the weekend switched to the Republican formerly in fourth place. The margins are now about 3,000 votes out of 5.2 million (800,000 per position) cast, above the mandatory recount threshold.

East Valley Tribune, Nov. 9, 2008

It's still too early for election punditry, because Arizona now has a six-week election "day." Early ballots are mailed a month before, and counting the votes takes an additional week. So fine distinctions (turnout was down from 2004 -- but by how much?) wait until next week.

Maricopa County election officials estimated that 113,000 early ballots were dropped off at polling places on election day (which must be checked before being counted) and another 80,000 to 100,000 voters cast provisional ballots (which have to be checked more thoroughly, and up to 40 percent may not be counted). We have to wait to start the real number-crunching, but here are some preliminary thoughts.

First, we know that (except for Arizona and Alaska) polling works. And with statistical sampling validated by the voters, the next time my doctor orders a blood test, I can get by with that little smear, I don't have to make her take it all.

The biggest innovation for political junkies this cycle was the polling aggregation websites that crunched numerous polls into models and graphical trendlines. You even have your choice: left (, right (, and wonky ( As a 5 percent margin of error means that even a perfectly-designed and executed poll will be wrong once every 20 times, I've become a believer in putting multiple polls together to get a better overall result.

We now know that the 2008 presidential election was easy to predict. Only seven (seven!) states were closer than 5 percentage points; only 15 were closer than 10 percent. Those are notable decreases from 2004 (10 and 21) and 2000 (12 and 22). If margins are that large, it's easier to predict the winner. Still, pollsters are having a difficult time figuring out whether and how to call cellphones; turnout models need to be tweaked for every election. So despite all these graphs and aggregations, nothing stopped me from being very, very nervous until the networks called Ohio for Obama.

Second, I guessed completely wrong on Arizona turnout. Our GOP Secretary of State also thought we'd set a record, based on registration and early ballot requests; instead, turnout was around 65 percent, a bit above the 2006 off-year gubernatorial election (61 percent) but well below the 2004 presidential election (77 percent).

The other state with turnout well under 2004 levels? Alaska, where turnout appears to be well beneath 2004's 66 percent and possibly lower than 2000's 60 percent. So if you think too many people vote, or that ballot-counters need less work, there's a simple solution: Put somebody from your state on the national GOP ticket.

Nationally, turnout soared; (preliminary) national turnout of 64 percent is the highest in more than 40 years, and a considerable leap from 2004's 55 percent and 2000's 51 percent. In Arizona, Democrats assumed that Arizona turnout would rise, and concentrated on urging low-efficacy voters to complete the entire ballot. Instead, with Arizona never really "in play" until the last minute, voter enthusiasm here (and Alaska) never matched what we saw in the rest of the country. New Arizona state slogan: "We found the 2008 election less interesting than Bush vs. Kerry."

Low turnout yielded the most anomalous result, at least for somebody who cut their political teeth here during the 1980's. I grew up seeing national and statewide Democrats get creamed, while local candidates could survive and thrive. Instead, last week national and statewide Democratic candidates did fabulously well -- for Congress, Harry Mitchell wins by 10 points, Ann Kirkpatrick and Gabrielle Giffords by even more, and Democrats sweep the three Corporation Commission races (subject to a potential recount on the third seat) and take control in one fell swoop -- while D's also lost a state Senate seat and two to four state House seats.

Maybe it's just the "native son" effect, which here depresses turnout. Or maybe we have more in common with Utah, Wyoming, and Idaho (where McCain got over 60 percent) than we do with Nevada, New Mexico, and Colorado. You bet I'll want to look at the numbers -- when we finally get them.

No comments: