Skip to content

Sides and Vavreck on the 2012 election

Political scientists John Sides (one of my cobloggers at the sister spot) and Lynn Vavreck are writing a book on the 2012 election.

I’ve seen what John has posted on this so far and it looks very reasonable to me. The paradox is that “reasonable” and “scholarly” are not as exciting as “soap opera.”

Nonetheless, I recommend the book to those of you who are interested in politics. My advice: if you want a gripping, edge-of-the-seat story with larger-than-life characters and fly-on-the-wall dialogue, pick up a good novel. If you want to understand the election, pick up Sides and Vavreck. (Also read Red State Blue State for some historical background!) Ultimately, I think you’ll get more enjoyment from your favorite novel plus Sides/Vavreck than you would from reading some dramatic journalistic treatment of so-called pivotal moments during the campaign.

In the future (and, to some extent, even now), I hope and expect that journalistic treatments of elections will be informed by political science understanding. It should be possible to go back and forth between politics, personality, and policy. Personality matters—not because voters care about charisma etc., but because politics is not just about winning elections, it’s also about the one-on-one interactions needed, first to climb the greasy pole and, second, to make policy once you’re at the top.


  1. Jonathan says:

    The Gamble by Tom Ricks is good too.

  2. A. Zarkov says:

    In this two-minute video (excerpted from a longer program) Feynman tells us that social science is not science. “They [the social scientists] don’t do science.” “They follow the forms.” “No laws.”

    I think Feynman’s observations especially apply to political science. Where are the laws? What are the invariants? From what I see, political science is a collection of stories with some observational statistics thrown in to make it look a little technical. I don’t think we can validly predict the outcome of the election right now from polling. Perhaps in a few weeks. We have a lot of disconnects. Hibbs “Bread and Peace Model” political science model predicts a Romney win. Nate Silver, using mainly polling data, predicts a strong win for Obama. How can we take this stuff seriously?

    Just for fun I took the Real Clear Politics electoral map and put in my personal guesses for what “solid,” “probable,” “leaning,” “tossup” means in terms of a probability. Then I used the method of generating functions to get the distribution of the electoral vote if the election were held now. It took me a 30 seconds to write a one-line code in Mathematica to do the calculation. I got the same answers as Nate Silver on his website for the win probability and the final electoral vote count. Nate does not seem to know about the method of generating functions, so he uses simulation. I have a hard time taking him seriously too.

    • Andrew says:


      1. Feel free to use the term “social studies” rather than “social science.” Feynman is certainly not the only person to point out that, in some ways, social science is much more difficult than physics.

      2. Whether or not social studies is science, I think much of value comes from surveys and social statistics. The unemployment rate and other things are estimated from surveys. We know about public opinion from surveys. Our knowledge is imperfect but I think is better than guesswork. During World War II the government needed Fenyman and his friends to develop the A-bomb, but social studies tools helped them figure out how many bombers etc. to build. Social studies is used by business marketers all the time—we can debate about whether this is a social benefit, but it’s certainly viewed as a private benefit for the people who conduct the studies.

      3. I like simulation, it helps me understand what I’m doing, it frees me to think less about algebra and more about whatever I am studying. Another strength of simulation as a method is that it allows less mathematically-trained people to do useful research. That’s great that you know generating functions, but a lot of people don’t, and it’s good for them to be able to solve applied problems too.

  3. A.. Zarkov says:


    1. Feynman was saying something more than the problems we find in the sciences are harder than physics. His comments were in the nature of a warning. By imitating the methods of hard sciences, the social sciences gain a kind of false authority. Economics provides a good example. The economists invented an ersatz Nobel Prize– “Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel” to make it seem that the field belonged to the hard sciences. This false authority can lead to expensive experiments, such as the 2009 ARA– an $831 billion gamble.

    2. Surveys give us measurements, and that’s good. But you need a theory to do something valid with the measurements. As for business marketing, how well does that work in practice? GM certainly blew it with the Volt. They will never sell enough Volts to earn a profit because the electric car will never be anything more than a niche market. And for very good reasons having to do with basic physics and chemistry. Nissan too has fallen into the trap. Perhaps these companies never did any market research. I don’t know. I only the the product is doomed.

    3. EE majors get generating functions in their Junior year– at least I did. Statistics majors should get the Z Transform in first year graduate school. But you would no more about this than me. It’s not a hard subject. The power series you get in elementary calculus should provide all the needed background.

    I like stimulation too, but I don’t like doing more work than I have to.

    • revo11 says:

      I’m not a social scientist, so I don’t have a vested interest in defending the profession.

      There may turn out to be few or no universal invariants for these kinds of complex phenomena (whether it’s social science, epidemiology, economics), but what’s the alternative? I’m fine with arguing over definitions and calling social science something else, but people will still be interested in things like monitoring the impact of a policy (business or government) decision, medical treatment, or various environmental effects on the population regardless of what you call the field. The efficacy of a particular medication doesn’t teach you some sort of universal law, but I’m pretty sure it’s still important knowledge.

      At least with quantitative models you can debate their correctness more precisely. A false sense of confidence in quantitative models is a genuine sociological issue, but it would be wrong to force researchers to be less precise about their hypotheses, assumptions and predictions just because the general population needs to be better educated regarding their interpretation.

  4. Mayo says:

    “cobloggers” reminds me of one of my son’s favorite rhyme: “Cobloggler, Cobloggler, mend my shoe, Get it done by half past two”(with slight spelling change).