Email: Bickers and Berry Model: Isn’t it a new model that was applied retroactively?

Hi Mr. Malstrom,

 

Just stumbled on your blog a few days ago and am enjoying your political posts.  On Bickers and Berry – from what you have written, it seems that you are suggesting that Bickers and Berry have been right every single election going back to 1980. 

 

I do not believe this to be the case.   My understanding is that the model itself is actually new, but was retroactively applied to “predict” prior elections.  So the authors did not predict Reagan’s victory in 1980 – they “predicted” it in 2012 by feeding their model economic data from 1980.  Ditto for all successive elections.

 

Here is the relevant paragraph from the widely published news article after the model was updated (to still show a Romney win):

The state-by-state economic data used in their model have been available since 1980. When these data were applied retroactively to each election year, the model correctly classifies all presidential election winners, including the two years when independent candidates ran strongly: 1980 and 1992. It also correctly estimates the outcome in 2000, when Al Gore won the popular vote but George W. Bush won the election through the Electoral College.

The model is certainly impressive – but it has yet to correctly predict a *future* election, as I understand it.

 

If you are interested, I can send you a PDF of Bickers and Berry’s journal article from the October 2012 issue of PS: Political Science & Politics, where the authors present the model and their findings.  Just let me know – I didn’t want to send it with this email since your spam filter might boot it.

 

Keep up the great work.  I still don’t really understand how you came to call it “Kermit Day,” but your analysis is insightful and tracks with most of what I’m reading/hearing/seeing about what’s going to go down next Tuesday.
_______________________________________I was under the impression that the quote you mentioned was due to the recent modifications to the model for 2012 which caused them to reapply the model retroactively. I do distinctly remember in 2000 of a University of Colorado forecast that had Bush and Gore split the popular and electoral vote. Perhaps that was someone else’s forecast or model. In the TV interviews with Bickers and Berry, the interviewers speak of it as forecasting races in the past.But here is the rub, if the model was created and applied retroactively, why isn’t that part of the criticism?I’m finding out that Bickers and Berry have an extremely high reputation among the other political scientists. Karl Rove quotes a statistical analysis of Bickers and Berry in 2011 as an example. I’m getting the impression they’re fairly famous. In an televised interview, Bickers mentions the forecast of the model is very important for their reputation. I thought he was referring to the reputation of the model’s performance. Perhaps he is referring to his and his colleague’s reputations.

Consider the tone of this New Republic article written about the Bickers and Berry model.  The writer sounds angry about the model and says it doesn’t matter as there are many other models. This is true. However, all those other models aren’t published in peer reviewed science journals.

Bickers and Berry must have sterling reputations as scientists as no one seems to be attacking them. What I hear instead is that ‘economic indicators aren’t relevant’. When this model came out, Nate Silver immediately went on TV and started dismissing economic indicators and said that things like how people ‘felt’ (such as having a beer with the incumbent) are more relevant. I’m sure Herbert Hoover thought the same thing.

We’ll see what happens in the next few days.