I can’t believe what just happened. I’m trying to make sense of it all, to make sense of the magnitude of the electoral shift that happened last night. I’m going to give it a fair, honest go.

First, a disclaimer, one that was addressed when I started writing these columns for the Press and Frontiersman three months ago. Ethan Berkowitz is a close friend of mine. I came to know him when my wife and I moved into the condo next door to him back in 1994. He is my son’s godfather, and he will always be my good, good friend. As I sit here now, the day after the night before, I’m in pain thinking of his disappointment, after a year of such valiant effort.

Advertisement

Having said that, I never at any point allowed any bias or favoritism to change the work that I did. When we did our polls, I crunched the numbers according to a strict protocol, and never once did anything to make his numbers look any better than the way they actually looked. When I wrote my columns, I was as forthcoming with criticism of him as I was with praise. Anyone who claims that I showed any bias needs to put down the Kool-Aid and look at the bigger picture—the picture of all the polls that were done by a variety of pollsters in the run-up to this election.

The last survey I did, fielded two weekends ago, showed McCain with an 11-point lead over Obama. It showed Begich with a one-point lead over Stevens. And it showed Berkowitz with an eight-point lead over Young. My arrangement with the Press, Frontiersman, KTUU and KENI did not include any further polling after that point.

A week after we finished fielding that survey, Ted Stevens was found guilty of all charges in federal court in Washington DC. That of course, changed everything in the U.S. Senate race. Four polls that I know of were done after the guilty verdict came down. The first was done by national pollster Scott Rasmussen on October 28, the day after the verdict, the second by Research 2000 between October 28 and 30, the third done by Hays Research on November 2, and the fourth by Dittman Research in the last few days of the election, between October 29 and November 2, with results released as polls closed on Election Day.

In the Presidential race, these four polls showed McCain leading by 16, 19, three and five respectively. The 16- and 19-point leads by the national guys were in line with the high end of margins in prior surveys, most of which showed McCain leading by a margin in the teens. The three and five were big surprises, which certainly seemed to indicate that the presidential race was closing.

In the Senate race, Begich led by eight, 22, seven and eight. The 22-point lead by Research 2000 seemed to be an outlier, but certainly the evidence from the rest seemed to be consistent—that Begich had opened up a post-conviction lead in the high single digits.

Rasmussen didn’t poll the House race, but the other three showed Berkowitz with leads of nine, six and seven. Consistent numbers again, which seemingly showed Berkowitz to be the favorite in that race.

Now, going back to the presidential race: If the leading pollster in Alaska, Dave Dittman, who’s been doing this stuff for much longer than I have, and who, I might add, I have a lot of professional respect for, has McCain up by five points right before the election, and it ends up being 25 points on the day, then obviously something very, very strange has happened.

Two things are clear to me. First, it’s not a “margin of error” issue. This kind of shift is so far outside what could conceivably happen by chance as to be impossible. And anyway, everyone was showing much the same “error”.

Second, people just don’t suddenly change their minds en masse like this and do something different from what they were saying they were going to do just a few days earlier. Some of the shift may have been real, but not all, not to that extent, and certainly not in such a small timeframe.

So did we just get our samples all wrong? Well, yes and no.

Consider this. In all the political polls we do, we ask people “What is your registered party affiliation?” I statistically weight all my statewide survey results to a rolling average of party ID, to smooth out the bumps of getting odd samples by chance that might favor one party or another. The weighted party ID result in the last survey we did two weeks ago, was 19 percent Democrat, 33 percent Republican. Compare this with actual percentages of party affiliation of Alaska voters, which is currently about 15 percent D, 25 percent R. Self-identified party ID is generally higher because some non-affiliated voters will identify with one party or the other despite the fact that they are registered non-partisan or undeclared. The important thing is that the R:D ratio in actual registration of 1.67 is actually reflected pretty closely by the survey’s party ID ratio of 1.74. So as far as party affiliation is concerned, this was a pretty representative sample of Alaskans.

But here’s the rub. It clearly wasn’t a representative sample of the folks who actually showed up to vote on Election Day. What I’m convinced we saw yesterday was a huge surge of Republican voters across the state, voters who were compelled to come out and cast a vote for Sarah Palin, and perhaps secondarily to cast a vote in support of Ted Stevens. This was the Palin Effect us pundits had been asked about repeatedly since August 29. In the end, it was profound, and was not confined to just Wasilla, but occurred statewide. The Palinbots, as they’re affectionately known, won the day.

In nearly 20 years of polling in Alaska, I’ve only seen a phenomenon even approaching this once before, back in 1994, when Tony Knowles was running against Jim Campbell for Governor. In the last two weeks of the election, Tony Knowles opened up a lead that grew every day, to the point where he led by 16 points going into the election. He ended up winning the race, but by only 500 votes, less than half a percent.

That was the year of Newt Gingrich and the Contract with America. It was two years into Bill Clinton’s first term. The U.S. Senate and Congress went red that year, and in Alaska, the state Legislature turned sharply towards the Republicans, giving them majorities in both houses that have lasted until today.

Long and the short, sometimes stuff happens that just turns things upside down, and this was one of those times. Thankfully, it doesn’t happen very often.

—Ivan Moore