Saturday, March 31, 2012

Not So Bright for a Liberal Prodigy: Dr. Drew Dismantles Nate Silver's Attack on Political Science Election Forecast Models

As a student of charismatic leadership, I'm usually skeptical about the so-called "prodigies" so admired by the leftwing media outlets like the New York Times.  Consequently, I like to think that I was the political scientist who first got under Nate Silver's skin in terms of pulling apart his arrogant predictions of electoral results based on last minute polling data.  I suggested that it really wasn't very interesting to predict elections results from polling data since all we were doing was assuming that the voters would not do much to change their minds between now and the upcoming election. Silver, by the way, rose to fame as a predictor of baseball outcomes.

Nevertheless, young guys like Nate Silver and Ezra Klein have been elevated by the liberal media because of their style of writing has a sort of pseudo-science quality about it which - I suppose - is the only plausible justification for having a young white guy writing for a liberal, extremist entity.  When you look into their credentials, for example, there is not to much to be impressed by their records.  Klien only earned a B.A. in political science at University of California - Irvine.  In 2000, Silver graduated with Honors with an A.B. degree in economics from the University of Chicago. Only in the crazy world of leftwing journalism is either one considered an expert.

As far as I'm concerned both of them are still undergraduates when it comes to understanding politics, economics or sociology.  As such, I enjoy poking holes in their pretentious, arrogant takes on contemporary politics.  As a political scientist and an eyewitness to young Obama's ideological extremism, I'm particularly irritated that folks like Silver and Klein do not seem to appreciate what is obvious to most of us: Obama's extraordinary ties to radical Marxist socialist thought. 

As part of my larger effort to bring these "prodigies" to Earth, I left a comment for Silver months ago that pointed out that political scientists could predict presidential elections far in advance simply by looking at variables like the economy or the success or failure of a war.  In reaction to my critique, Silver has launched a full scale attack against the very idea of political science and the usefulness of making predictions far in advance.  In one of his most recent articles, for example, the title of his article makes the absurd claim that political scient models based on ‘fundamentals’ have "failed" at predicting presidential elections.  This headline, of course, is absurd when you consider, as John Sides writes:

Of the 58 separate predictions that Nate tabulates, 85 percent of them correctly identified the winner — even though most forecasts were made two months or more before the election and even though few of these forecasts actually incorporated trial heat polls from the campaign.


Sides, by the way, is a political scientist who uses these forecasting models himself and sees them as ways of testing overall theory. The larger issue, for me, is that Silver's comments reflect the odd anti-intellectual bias of the left-wing media.  As far as I can tell, the most interesting, powerful and predictive work in political science is coming from conservatives like the late James Q. Wilson or Charles Murray.  Personally, I'm sort of proud that my modest contribution to political science - my child labor, child welfare theory - does such a great job of predicting when and where child welfare programs arose in the U.S. and around the world.

All in all, I'm skeptical of guys like Silver who exaggerate their skill sets to the extent that they imagine that their comments "sit at the intersection of political science and the mainstream media." As an award-winning political scientist, I think common sense alone suggests that a fellow writing for the NYT would be anxious to look for anything to back up his fellow liberals' hope that the horrible "Obama economy" will not cause Obama's defeat in 2012. As a real, published political scientist I have the following comments:

1. A Real Political Scientist Doesn't Play Games with the Dependent Variable: It is silly for him to change the dependent variable to one that the authors of these models don't hold to simply because Silver says, "I’ve always found this choice to be counterintuitive." Silver's bizarre new dependent variable, however, helps him make his case because, as he writes, this choice "increases the error rate of the models so that they are about twice as large."

2. A Real Political Scientist Pays Attention to The Comparative Political Science Literature: If the connection between the economy and election results was as tenuous as Silver suggests, then the comparative date should support Silver's thesis.

3. A Real Political Scientist Understands that the Data is the Variable: In truth, there are no special, unmeasurable things when it comes to political science.  In the end, the data says what the data says. The bottom line is that the economy matters a great deal when it comes to predicting presidential election because of what we know about voters, not what we know about linear regression models. This is bad news for Obama and his supporters.  Those of us who really understand economics and politics predicted as early as January 2009 that the stimulus package would fail, that the economy would remain poor longer than Obama expected, and that this would doom his chances for a second term.  Thankfully, one of the benefits of age is that you get better at political science, not worse.

Being a prodigy is great, but - in the end - there is no substitute for long years of specialized study in your field of expertise.

John C. Drew, Ph.D. is an award-winning political scientist.

2 comments:

Unknown said...

67%. Out of 58 models, 39 correctly predicted the winner. 39/58 is 67% not 85%.

So in 5 elections, including such historically close races as Clinton/Dole and the true nail-biter of Obama-McCain, the political science models managed to score better than random chance, but just fall short of the standard set by Mani the psychic parakeet(71%).

Truly a win for political science

marketfailure said...

In like manner just because you have a PhD is not enough for us to conclude that you are smarter than Nate Silver. I know I have my own colleagues. But unlike them you do not even hide behind the fact that you are a right-winger and have a clear very clear biased agenda so why would you be more credible?
I'm confused...I thought the left were mostly smug intellectuals and then they supposedly have "Anti-intellectual" bias? huh? Doesn't your right wing friends like Santorum demonize the academicians, the intellectuals as rabid biased left-wingers? make up your mind! There are no unmeasurable terms? I'm a little confused...we assume that if we cannot get a variable or even a proxy to represent theory then it is thrown in the error term. And if theoretically we can justify that this empirically unmeasurable variable should be a significant independent variable then the expected value of the error term is not zero.
Stimulus package would fail? what bubble are you living in? If you consider it a failure a larger school of thought have it as being not big enough to solve the worst crisis since the great depression thanks to the man who really never won any election - twice. So its not the education, its the age? which is which? - JM PHD in economics

Most Popular Posts