Why can’t political analysts predict anything?


The study of politics is a lot of things, but what it isn’t is predictive — yet, I suppose.

When people hear I have education in politics, I inevitably get the: “So, who’s going to win in the X election?” Of course, this is right after the “are you a democrat or republican” question.

To which I would reply (to the election question): “Well, it depends.”

There are a whole lot of complexities when trying to understand and/or predict politics and political development.

Let me run through all this, because it’s a lot:

What is political analysis good for?

Here’s the first thing to know about politics: Making point predictions is risky, and pretty difficult.

Physics can tell you how fast an object falls, astronomy can tell you how far away a star is (and it just proved Einstein’s theory about wrinkles in time!), biology can tell you how many cells are in your blood, and genetics can tell you which traits you got from who. These sciences tell you about stuff, and it can predict how that stuff will change over time.

Political science, on the other hand, cannot predict who will win each race all the time. International politics cannot tell you where the next war will be, or which dictator will get overthrown next. … But the discipline does have segments that are trying to do all that with quantitative analysis — which I personally have problems with.

Here’s an example of how difficult it would be to accurately predict who wins any given election, and just some of the things you would have to look at:

First, you need to know each candidates personal situation. Is there anything that could alter their performance in debates or appearances? Maybe one of their kids is sick, perhaps they personally have the flu, they have a medical condition nobody knows about, they got into a fight with their spouse the day of an important speech or rally. At a personal level, lots of things can affect personal performance.

Second, we need to know more about each individual’s campaign structure and effectiveness. How well-organized are their campaigns? We all know they have a lot of money, but that’s only half of it. If you don’t have a quality infrastructure, dedicated volunteers and interns, effective communications, and successful voter drives, chances are that will definitely affect a candidate’s standing. How effective are the tactics used? Most effectively, face-to-face interaction with voters is a time-tested way to increase votes — so which campaign is doing that the best at that? How many people are volunteering?

Third, what about other external variables? Aside from how effective a candidate’s campaign is, there are the events they go to. How successful were those? Was there any energy? Lots of media? No media? Is there support from the upper-echelons of each community? Have the strategists correctly identified their demographics? Also, are they reaching out across the demographic lines enough? Does the candidate have an effective campaign strategist or consultant?

Fourth, what about donations? And donations are only part of the complexities. A lot of money indicates support; however it doesn’t translate into votes necessarily — ask Jeb Bush.

Fifth, there’s other external stuff. Like polls.

… ehhh …. Polls.

These things are tricky, because they CAN be accurate IF there is a genuinely random sample, and there’s enough data. Consider this: The FCC says you can’t randomly call cell phone numbers. Thus polls use LAN lines, and you know who has LAN lines? Old people who are more often on the conservative side. So already your sample is biased.

Also: How many people answered the questions accurately? You can round up people in person to take these polls, but I know there are LOTS of people who check C down the line, collect the $10 the polling agency is offering, and leave. Also, I’ve never met ANYONE who was actually a part of one of these polls, have you? So how far is the reach? Then there’s the questions, and the wording — pollsters really need to critically think about the questions and how to extort genuine answers before the polls are worth anything. The age-old advice of “crap in, crap out” is alive with polls.

… I have lots of problems with polling. But, regardless, it’s an attempt by political scientists to get enough data to be able to claim empirical evidence in order to provide an idea of who COULD win; and, at times, they do help lend credit to predictions. But I would have the question: Do polls predict outcomes, or are polls responsible for outcomes? … Does the existence of polling create a self-fulfilling prophecy of sorts?

Anyway, there are lots of complexities in political analysis like that. International studies has is own set of highly complex variables to sift through.

The point is, making predictions in politics is always a gamble, because politics is always changing, dependent on random chance to some extent, and really difficult to grasp. But, the good news is that when there’s only 2 candidates, you have a 50/50 shot at a prediction. But generally, political scientists stay away from point predictions like that.

So the question: What is political analysis good for?

Explaining complex stuff in the present.

As a whole, political study is firmly rooted in present tense. It’s also historical. It can use insights from the past, explain the present, and offer advice moving forward. What it cannot do, however, is predict, because it isn’t a predictive science (though Nate Silver is pretty successful, but I think he uses all sorts of complex Bayesian statistics, which are rarely taught in school).

So why is it political science called a science? Because you MUST make an analysis based on empirical facts and data, rather than relying on something else. It’s scientific in the fact that political scientists create theories on why things occur in the world, and others test those assumptions.


You’re telling me that political analysis can’t predict the future!?

Not at all. So, sorry all of those out there who thought that if you study politics long enough that you can predict developments. You simply cannot.

In fact, there are markets built around predicting politics precisely because the people running the site know that it’s impossible to predict. Go to predictit.org if you’re a gambler, and see how accurate you can be.

Notes on good analysis:

— Making a good political argument takes a lot of thought. It requires the person to make sense of the world.

If you look at the painting at the top of this article, it’s an abstract painting by Joan Miro. It makes no sense, unless the viewer makes sense out of it. Until the viewer creates some sense of understanding about it, it’s meaningless and random. The same goes for politics.

— A good analysis picks good variables to focus on.

If you want to discuss politics in a coherent manner, choose and parse down what matters from what doesn’t. I’ve heard some people talk about North Korea one minute, then switch to healthcare the next. I have no idea where the thoughts are going, and it’s really difficult to keep up, because they aren’t well thought out ideas. Instead, a lot of people ramble about politics.

Pick what matters from what doesn’t, and, if you’re really interested, you need to research it beyond what the media gives you and make sense of it (in a coherent and logical way, otherwise it’s randomness).

Examples of stuff I hear, a lot: “Obamacare leads to high insurance rates,” or “taxing wealthy people will bring in more tax revenue.” What I don’t hear, however, is why. … I have no idea why. Someone, explain it to me.

Also, people have a tendency to radically simplify the world. A lot of people talk to me as if the federal government is one coherent entity, which it, in fact, is not — like at all. So, to you people saying the “government” or the “feds” want to take or give something from/to us — that doesn’t make sense, and is a radical simplification of the world.

At the end of the day, examine what you’re saying, and say it aloud, preferably. Chances are it sounds worse out loud than it does in your head.

… Where are we at with the current presidential election?

To be honest, it’s a bit of a haze right now. Nobody on either side has gained any momentum.

This particular campaign is really up-in-the-air. Nobody can seem to get a grasp on what the people want. And to complicate matters further, always keep in mind that there are 50 states in the US, and several territories that vote, and each locality has a different character.

But, I’m starting to track the data, and I’ll be updating as the campaign develops — because I operate on empirical data too!

So all you political nerds out there, hold on. I’ll soon have actual numbers/data to report on this one. Also, I predicted about a month or two ago who would win the presidential nominations for each party — and, so you know, I’ll stand behind it for better or worse!






2 thoughts on “Why can’t political analysts predict anything?

  1. Pingback: What is going on in Syria? | Political Ideas and Education

  2. Pingback: Benghazi: What I found after digging through Hillary’s email | Political Ideas and Education

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s