Tag Archives: Political Science

On Thursday, I will be presenting at the 2013 Southern Political Science Association Conference. My paper is entitled “The Invisible Fist: How Potential Power Coerces Concessions.” You can download a copy here. In it, I show that traditional explanations for nuclear proliferation are insufficient, as they do not appreciate bargaining’s role in incentivizing states to not join the nuclear club.

For my presentation, click here.

Here is a generic criticism that has been, is, and will be levied at forecasting models like Fivethirtyeight and Votamatic:

The forecasting models fail to account for x, y, and z. But x, y, and z are fundamentally important! Therefore, we should not use the forecasting models.

Fallacious! I think we can all agree that, for various reasons, being able to predict the outcome of elections is important. We cannot just stop forecasting tomorrow. Given that, the question is a matter of what methodology we use to predict outcomes.

In that light, the above criticism fails to highlight the real question. Rather than asking “are forecasting models perfect?” we should be asking “are forecasting models better than the alternative?” In other words, we should treat what we currently have (talking heads on Fox News, CNN, and MSNBC) as the null hypothesis and the forecasting models as the alternative hypothesis. And that being the case, the forecasting models beat the hell out of political punditry.

Yet, the full criticism we often hear is this:

The forecasting models fail to account for x, y, and z. But x, y, and z are fundamentally important! Therefore, we should not use the forecasting models and instead keep pretending my inane rants actually have meaning.

Of course, the political pundit’s inane rants have absolutely no meaning. The pundit is quick to criticize what he does not like but then gives himself a free pass. However, not only does his punditry fail to account for x, y, and z, it is also completely made up horse manure, often fabricated for the sake of ratings. (Or page views…coughunskewedpollscough…)

Now, we should not take forecasting models completely off the hook. They have problems, and their creators are the first to admit that. But, as with anything else in life, we need to ask ourselves whether this devil is better than the other devil. And personally, I’d rather have Nate Silver’s pitchfork pointed at me than Joe Scarborough’s.

Within a few hours (hopefully), we will know who won Ohio, and this election season will mercifully come to a close. All of the sophisticated forecasters agree: Obama is going to win Ohio and therefore the White House. (See Nate Silver and Drew Linzer.)

However, there is one unsophisticated forecaster who is convinced everyone has it wrong. His name is Unskewed Polls Guy. He is convinced that Romney is going to win. Why? Well, Unskewed Polls Guy’s methodology is that we should take all the state polls and arbitrarily add 5% or so to Romney’s total. Or something. Why? Well…uhh…the polls are skewed toward Obama. Or something. Hence, Unskewed Poll Guy unskews the polls by throwing votes Romney’s way. Or something.

As the last paragraph illustrated, I have absolutely no respect for Unskewed Polls Guy’s methodology. To say he is pulling numbers out of his…ahem…would be gentle. Hell, up until today, he had Oregon as a state Romney could potentially win. Oregon. Oregon? OREGON.

But my post title is not sarcastic: Unskewed Polls Guy is a true genius. Here’s why. First, regardless of the election results tonight, Unskewed Polls Guy is already a winner. His website gets a ton of traffic, half from crazy conservatives who believe math is a false paradigm and think he truly is unskewing polls and half from sane people (both liberal and conservative) who find his lunacy to be highly entertaining. The inaccuracy of his forecasting has brought him a substantial bounty–precisely due to how horrible his forecasting is. Brilliant!

But that’s not the real genius of Unskwewed Polls Guy’s ignorant plot. Suppose the other forecasting models do get it “wrong”–that is, the outcome of the election is far away from the mean prediction. This is well-within the realm of possibility. Silver’s model gives Romney a 9% chance to win. That is substantial, and you generally need 95% confidence to get published in social science. Put differently, social scientists would a Romney victory odd but not altogether shocking.

Of course, if this election cycle has taught us anything, it is that the media has absolutely no clue how the forecasting models work. 24/7 cable news has been embarrassing for a while now, but this has gone to a whole new level. If Romney wins, they will treat Unskewed Polls Guy as a prophetic god. He will be the hottest media commodity for the next two weeks. He will get a book deal. He will make a large sum of money. And why? Because he is a complete idiot, has no clue what he is doing or why he is doing it, and just happened to accidentally hit a miracle. If it has happened with an octopus, the frenzy will certainly happen with a human.

And that’s the real genius. If Obama wins, Unskewed Polls Guy takes home a lot of advertising revenue from the past few weeks, and we all forget about him. If Romney wins, Unskewed Polls Guy hits the jackpot. Consequently, if Unskewed Polls Guy is trolling us all, my hat is off to him. If I had thought about this a year ago, I might have been Unskewed Polls Guy.

Going forward, forecasters have awkward incentives. Fivethirtyeight has pretty much cornered the market. You would think that a forecaster should just maximize his chances of being right. But that pretty much means copying Nate Silver. So, to become famous, you really ought to make ridiculously crazy predictions, hope nature randomly makes you right, and reap the short-term rewards. Long-term, you are screwed. But who cares if you’ve already received your book advance?

A couple weeks ago, I read an interesting article about how illegal immigrants can sway Electoral College votes. As it turns out, the Constitution bases electoral votes off of population counts from the census, which in turn must count all people–citizens, legal immigrants, and illegal immigrants–living in the states. Thus, even though only citizens can vote, states with larger numbers of immigrants receive a disproportionate number of electoral votes.

This has obvious electoral consequences. In particular, California gets hammered. California currently has 55 electoral votes but would drop to 50 if the census only counted citizens. On the whole, traditionally red states tend to gain from the alternative method of counting.

Based on this, the article makes the following causal claim:

If President Obama wins reelection by three or four Electoral College votes next month, the reason may be simple: noncitizens, mostly immigrants, who don’t have the right to vote.

But are immigrants really causing Obama to win reelection? Well, yes, but in a very narrow sense of causation. Presidential campaigns, if nothing else, are extremely strategic. The candidates receive a set of rules and base their strategies off the rules. In such a strategic world, you cannot change the rules and hold strategies as being constant, since the strategies are a function of the rules.

To better understand the relationship, consider the following game. All you have to do is pick A or B. If you pick A, I give you $1. If you pick B, I kill you.

Obviously, you are going to pick A. But that does not mean you prefer A. You just prefer the outcome associated with A. If I were to flip the rules on you and say A leads to your death and B pays you $1, you will suddenly really like B and really hate A.

So now imagine we switched the rules of the election to only grant electoral votes based on citizen population counts. Suddenly, some of Obama’s electoral maps are no longer winning strategies. Will Obama naively continue to pursue those strategies? Certainly not–not anymore than you would continue to select A after I switched A from rewarding $1 to killing you. Obama would spend more time and money in new “must-win” states. Romney would likely follow suit. Obama would probably campaign on different issues. Romney would as well. From here, it is not immediately clear who would win the election, since some red states might switch to blue states due to the new policy offerings and vice versa.

In fact, it is not even clear to me whether Romney would have been nominated or if Obama would have been elected in 2008 under the different set of rules. Indeed, the only clear implication of switching the rules is that both candidates would have selected more conservative policy positions. But this is exactly what you should expect when the median (electoral) voter shifts to the right.

Overall, the original article reflects a common problem with our understanding of causation. James Fearon’s article on counterfactuals is a good reference here. When making a counterfactual argument, his guideline is that a premise A only causes B if in the absence of A we have an absence of B. In other words, when making causal arguments, we must give equal weight to the counterfactual story. What does a world without A look like? If we could still reasonably find that B persists in the absence of A, it is hard to claim that A causes B.

This process isn’t too difficult when players are nonstrategic. But throw in strategic players and you really have to do a lot of work. In fact, game theory has a whole process of calculating such changes called comparative statics. Comparative statics are known for producing brutally counterintuitive results. Below is an example with soccer penalty kicks. Strikers actually aim to their weaker side more often than their stronger side. Weird, right?

Absent learning a lot of game theory and learning how to calculate comparative statics, I suppose the moral of the story is to be very careful when making counterfactual claims and to seriously consider how the entire strategic interaction might change if you alter one of the inputs. Apologies for the somewhat unsatisfying conclusion.