Sunday, October 31, 2004

Final Weekend – U.S. Election

The Presidential race is too close to call. The reason is simple: the popular vote is almost tied nationally (you can check out the updated national polls here and here), and the polls in states that could tip the Electoral College one way or the other are also very close.

Here are my observations about all this:

1. A blogger called the Mystery Pollster – a Democratic polling professional – has averaged the national polls. His latest calculation shows a one point Bush lead of 48% to 47% for Kerry. One polling rule of thumb is that the incumbent will actually receive his final poll share on election day, while most of the ‘undecideds/won’t ‘say go to the challenger, but it doesn’t always apply.

2. On many minds this weekend is the question: what impact will the Osama tape have? My own view is that it is likely to have little or no impact because most voters have had plenty of time to decide how they feel about all the terrorism/war in Iraq related issues. However, there is one poll out from a Democratic pollster, which suggests, if anything, that the tape will help Kerry.

3. It should be noted that most polls just prior to the 2000 election showed Bush ahead, while Gore actually ended up winning the popular vote – a fact which adds to my impression that U.S. polls are often wrong. That’s partly because there are simply more of them, but also because many use small sample sizes and can’t figure out which respondents will actually vote.

4. This contributor to the Daily Kos site has averaged the polls in the battleground states. He concludes that Kerry will win 311 electoral votes to Bush’s 227. I think the author, a pro-Kerry Democrat, lets his partisan enthusiasm get the better of his judgement. While he may be too optimistic about Kerry in drawing his conclusions, his calculations do say unequivocally that the election is very close and the numbers are worth looking at.

5. All this year I have thought it would be tight. There is a great deal of evidence that tells us that the United States is split down the middle politically. My intuition has been that Bush would win, even though I think the longer term trend is against the Republicans. Now that the last weekend is here, my intuition is wavering: the reality is that this election is extremely close, too close to call. Factors we cannot now see (but will be clear after the balloting) could tip it one way or another.

State by State
6. My reading of the polls is that the following five states are toss-ups and will likely determine the outcome – Ohio, Florida, Iowa, Minnesota and Wisconsin.

There are other close races: in New Hampshire, Michigan and Pennsylvania where Kerry appears to be narrowly ahead. New Hampshire voted for Bush in 2000, the other two for Gore. In New Mexico, a state that voted by a razor-thin margin for Gore, my reading is that the polls suggest Bush will win. If all this comes to pass, it would result in a net gain of one electoral vote for Bush.

There are some other close races in Nevada, Arkansas and Colorado, which are probably going to be won by Bush but not by much.

7. One of the most important aspects of electoral behaviour is turnout. It is also one of the things that gives American pollsters fits because turnout is generally low. Hence pollsters build in “likely voter” screens into their questionnaires. The Mystery Pollster has a good discussion of this.

One of the problems with “likely voter” models is that this year there is every prospect of a dramatically increased turnout so that models based on past behaviour would not apply. There is already plenty of anecdotal evidence of increased turnout including strong early voting. Some of this is motivated by strong antipathy to Bush. I know of several Americans living here who are voting in a U.S. election for the first time in their lives – even in states where the outcome is not expected to be close – because of their strong negative feelings about Bush and his record. To me it is a clear indicator of increased turnout caused by anti-Bush sentiment.

The Democrats are targeting increased voter registration and turnout among younger voters and minorities. See, for example, this discussion of increased registrations in New Mexico. For the Republicans a key strategy has been to increase turnout among evangelical Christians. Bush strategist Karl Rove has calculated there is a pool of 4 million voters the Republicans could tap through turnout efforts – voters he thinks Bush should have had last time. I have no way of assessing whether the Republican efforts are going to be successful.

8. In conclusion let me note one of the biggest imponderables in polling in the U.S. this year. Polling firms can’t reach cell phone users and it’s a problem for them because many younger people rely exclusively on cell phones. There is no hard evidence that the cell phone problem has yet had a distorting impact on poll results but given the growth in the phenomenon, it seems to me that it is only a matter of time. Is this the year?

No comments: