in a previous post you said that the political scientist with most credibility where Bickers and Berry, and opposed their prediction to the predictions of other people (like Nate Silver, who I didn’t know). This seems a good moment to put these predictions to test. Will you talk about this after the results come out?
Bickers and Berry would be very interesting to go into. I don’t fully understand their model. I know they primarily use economic data but also throw in national variables unlike other models.
They had updated the data to the model, and it flipped New Mexico to Romney. On their report, which goes into more detail, it had New Mexico just bordering on 50%. That is going to be a close state. However, it was also a close state in 2004 if I remember correctly.
Interestingly, since I first posted the results of that computer model, the Romney Campaign has written off Nevada and both campaigns are targeting Wisconsin, Minnesota, and Pennsylvania. The one state that doesn’t make sense to me, in the context of the results shown, is Iowa. Wouldn’t Iowa flip before MN or PA? However, I do hear the economic situation is much better in Iowa than in the rest of the nation.
I’m not going to bother commenting on Nate Silver. Nate Silver isn’t a pollster, he is a mid level statistician currently in stardom due to 2008. The entire political science profession is going to gore this guy after the election (no matter who wins). It’s not going to look pretty. I expect Nate Silver will just say that the results are what the statistics said they would be. Just because a guy is favored to win doesn’t mean the other guy can’t win. However, the goring is going to come due to Nate Silver’s stardom not being about that but about allowing people to believe he’s a ‘soothsayer’ as he was given a 250k grand advance on a book deal and the New York Times made him a columnist. He’s been on TV recently trying to sell his book which many people take as a sign that this is his last chance to get money before the sky falls on him.
The political scientists’ complaints against Nate Silver are complex. The biggest issue, I think, is that he is creating the impression there is only one model of analysis and that good science means less skepticism. Actually, good science is skepticism. The complaints are also directed at the Nate Silver ‘fans’ where their behavior is shoving the ‘statistical probabilities’ in people’s faces, dismissing any other outcome, and declare it ‘math’. Social science has limits.
What I dislike about Nate Silver is how the Obama campaign fed him internal polling which he used but publicly presented the results as his own analysis. I also highly dislike people who make money using other people’s works. Silver is taking other people’s polls, analyzes them, and then declares this statistical sampling his own ‘analysis’.
Whether Obama wins or loses, Silver is going to be tarred and feathered. I’d rather talk about the Bickles and Berry model and see what they got wrong in that map… if they get anything wrong.
You might ask, “Why are the colors wrong?” They’re not wrong. Republican used to be represented as blue and Democrat as red. In 2000, all the networks switched it around. Why? I don’t know.
I really do like how the model is doing political soothsaying based nearly entirely on economic data. I’m a big believer that economics is political destiny. If you want to change the politics of a region, change its economics. This is why the southern states changed because their economic system modernized due to World War 2. After World War 2, those states have been trending more and more Republican but that’s a different post for a different time.
I am curious how the model survived its early days. Imagine you make a computer model and it creates the 1980 election map. That would be hard to believe especially when all the polls said it was ‘tied’. When Bickers and Berry saw the 1984 map, I’m surprised they didn’t throw away the model on the spot. I would have.
At the University of Colorado, they currently have the widest split in opinion. Of the thirteen studies done, they have five (including this model) going to Romney, five going to Obama, and three going to undecided. After the election, all the political scientists are going to be gorging on the election data, doing all sorts of analysis, and it will be fun to watch them.
I went on youtube to see what I could find of any Bickers and Berry interviews. Here is what I found:
This one appears to be a local Colorado interview.
This one appears to be a clip from O’Reilly’s show. I know people are going to have a problem with it being O’Reilly and Fox, but I’m not seeing these guys interviewed anywhere else. Annoyingly, the video ends abruptly. I want to hear these guys explain their model.
In the final video, Bickers is interviewed again on Fox but the video is extremely poor. It is taken of the TV (ugh) and the camera misses half of Bicker’s face. Since we’re only watching talking heads, I figure Bickers audio of him explaining his model is good enough to show here.