Wednesday, October 22, 2008

Electoral College Update

Jay's latest update to our Electoral College forecast (right panel) switches West Virginia from Obama to McCain. To be honest, I was a bit squeamish about having it in the Obama column but that's what the model dictated, until today. As it stands right now, North Carolina is teetering and could switch to Obama with a few more polls showing him ahead, and West Virginia could move back to Obama if the polls there reverse course. Other than that, though, I don't see many sates as likely to switch between now and election day.

1 comment:

Anonymous said...

I've had a couple people ask about West Virginia. Two things led to it showing up in the Obama column over the last couple of weeks. The first is the fact Clinton had done very well in WV in '92 and '96 (57.77% and 58.34% of the 2PPV, respectively), so the average Democratic share of the 2PPV over the last four elections was 51.60%. That, coupled with McCain's small lead in the September polls, made the model give McCain a small edge in the state in the September forecast.

Then came the American Research Group poll that was released on the 9th. It showed that Obama had opened up an 8 point lead in WV, contrary to all the September polls which had been leaning towards McCain. It was the first October poll for WV so it became the initial baseline for the state in the October model. As more and more polls came out, we began to see the forecast slowly creep back over to McCain. In the last five days, three polls came out that showed that McCain had a lead anywhere from 6 to 12 points. The Rasmussen poll on the 20th had McCain leading Obama by 9 points and that was finally enough to cancel out the ARG poll and put WV back over into his column... barely (53.8 win probability). The forecast continues to move WV more firmly over to McCain. The latest run (10/22) includes the CNN/Opinion Research Corporation poll which also has McCain up by 9. The impact of that new poll is that McCain's projected win probability for WV jumped to 66.1 overnight.

The thing to keep in mind is that the poll variable in our model is based on the average of the polls for the month. That means that it is likely to have some issues with stability in the first part of the month. The further we get into the month and as more polls come out, the more the errors in each of those polls cancel each other out. In this instance, it just happened to be the case that the first poll out of the blocks for WV this month appears to have been the outlier and it has taken a couple weeks for the model to catch up.

As Tom points out, we're seeing the opposite situation right now with North Carolina. Lately the polls there have shown Obama opening up a lead there and that has been reflected in the forecast. The September forecast had projected a McCain win probability for North Carolina of 83%, owing in large part to the fact that NC has traditionally gone Republican and that Obama had not yet really started his surge there.

The October 22 forecast now puts McCain's NC win probability at just 51%. As it stands right now, NC is the only state that the model considers a tossup. If any more polls come out showing Obama in the lead there, it will likely flip the forecast.