Even social scientists can think like pundits, unfortunately

I regularly read the Orgtheory blog which has interesting perspectives from sociologists. Today I saw this, from Sean Safford:

I [Safford] actually hold to the idea that the winning candidate for President is always the one who has a clearer view of the challenges and opportunities facing the country and articulates a viable roadmap for how to navigate them.

I disagree entirely. I think Safford’s view is naive and it is based on the idea that the outcome of the election is determined by the candidates and the campaign. In contrast, I believe the political science research that says that economic conditions are the most important factor determining the election outcome. You could go through election after election and be hard-pressed to make the case that the winning candidate had a clearer view of the challenges etc. Sure, you can find some examples: arguably Reagan had a clearer view than Carter, and Obama had a clearer view than McCain, and . . . ummmm, maybe that’s about it. Or maybe not. You could make a pretty good argument for either candidate in most elections, from 1948 onward.

Also Safford should really really watch out about that “always.” In the postwar period, there have been three elections that were essentially ties: 1960, 1968, and 2000. Even if you want to make the case (a case that I completely disagree with) that presidential elections are won by candidates expressing a clearer view and a more viable roadmap, still, you can’t hope to think that this will work every time, not given that some elections are basically coin flips.

The above fallacies—the idea that elections are determined by candidates and campaigns, and the idea that there is some key by which the election outcome can be known deterministically—appear a lot in political journalism, and my colleagues and I spend a bit of time at the sister blog explaining why they’re wrong. I don’t usually see academic researchers making these errors, though.

I looked up Safford and he’s done a lot of qualitative work on labor and social networks. This is important stuff, and I expect that if I started opining on the effects of labor union strategies, I’d be about as confused as Safford is when writing about electoral politics. So I’d like to emphasize that I’m not trying to slam the guy for making this mistake. We all make mistakes, and what’s a blog for, if not to put some of our more casual speculations out for general criticism. (In contrast, I was more annoyed a few years ago when political theorist David Runciman had the BBC as his platform for spreading pundit-level errors about U.S. public opinion.) So, no hard feelings, it’s just interesting to see an academic make a mistake that I usually associate with pundits.

18 thoughts on “Even social scientists can think like pundits, unfortunately

  1. Andrew

    Since you believe the political science research that says

    – economic conditions are the most important factor determining the election outcome

    I wonder what you believe will determine the outcome of the current presidential election (and what that outcome will be).

    Just asking, no answer expected.

    • Ethan:

      I wrote something on this in a recent news article so haven’t posted on blog. Once the article appears somewhere, I can link to it from here. The short answer is that recent economic growth has been steady but not great, suggesting a small benefit to the incumbent party’s candidate but with much uncertainty.

      Regarding the question of what will determine this particular election: It is certainly possible that non-economic factors can make a big difference this time. One reason it was easy for me to respond so forcefully to Safford’s post was that he was making a claim about past elections, a claim which in aggregate I think is just wrong. If he had just said that in 2016 he thought the winning candidate would be the one who “has a clearer view of the challenges” etc., then I could’ve responded that, given historical patterns, I was doubtful of his claim, but I wouldn’t have been able to say he was necessarily wrong.

  2. While it’s true that past US presidential election results have been, to a considerable extent, predictable based on economic performance prior to the election, there is still a lot of variation unexplained by said economic performance, right? And isn’t it possible that some or much of that variation may be explained by candidate qualities or campaign strategies? That particular efforts to measure such things have not shown strong relationships may speak more to the difficulty of measuring such things than to their irrelevance. Though of course the strong effect of economic performance suggests that pundits typically overestimate the importance of such things.

    • Anon:

      Yes, I think that some of the variation must be explainable by candidate and campaign factors. Safford’s statement was particularly weak and pundit-like in that (a) he was claiming a deterministic relationship, (b) he was making an assertion with zero evidence, and (c) I don’t think it’s even close to correct, just considering past elections. It’s not clear at all to me how you’d decide, prospectively or retrospectively, “who has a clearer view of the challenges and opportunities facing the country and articulates a viable roadmap for how to navigate them” in each election. But if there were a way to do this, it’s hard for me to imagine this lining up with the winner in each election. As noted in my above post, in one or two cases (Reagan and Obama), I can see it, but in other cases, no, I don’t.

      Had Safford written something like, “Political scientists have found the economy to be the most important predictor of presidential elections, but, based on the economy, 2016 is predicted to be a close race, and so other factors will come into play. And my [Safford’s] guess it that this election will be determined by who, Clinton or Trump, has a clearer view of the challenges and opportunities facing the country and articulates a viable roadmap for how to navigate them”—had he written this, I’d say, sure, maybe so, I wouldn’t quite put it that way myself but it’s an opinion.

      It’s tough, though, because people often like to think and express themselves deterministically, and people often like to impute nonexistent historical evidence for their speculations. That’s what bugged me enough for me to post on this.

      • Variation is also determined by the secular trend of non-white composition of the electorate (not within any one election but over elections). In 1968 the economy was great (3% unemployment, 4.97% GDP growth–http://www.multpl.com/us-real-gdp-growth-rate/table/by-year), but there was a draft (wars don’t seem to matter–W won reelection in 2004 with two wars going on). I think more useful than the economic paradigm (which is important to be sure) is whether the campaign was winnable with the candidates who are running or whether a different candidate might have won. For example, in 1988, a different candidate (Cuomo?) might have won for the Democrats, and in 2016, given how unpopular Clinton is, Cruz is kicking himself for not attacking Trump early and often. 1972, 1980, 1984, 2008 it wouldn’t have mattered much (IMHO) over who the candidate was.

    • Erikson and Wlezien are good on this. The economy explains 50% of the variation in some of their analyses. Wlezien has observed around sixty percent or slightly more in recent work. Wars and other major “salient” events impact.

  3. Elections are infrequent. When there are few events, it is easy to fit them into your theory retrospectively. I have read that the taller candidate always wins, that the candidate with the shorter name wins, and that the election is determined by whether a team from the old NFL wins. And who can forget watching where a free roaming cow drops a pie? We should have a betting market, and every commentor should be required to disclose what percentage of their annual income they have bet on their prediction. I would pay more attention to someone who predicts that Candidate X will beat Candidate Y if they bet 10% of their annual income.

  4. Isn’t the bit about who holds “clearer views” grossly subjective? Sure, the candidate with clearer views wins, depending on how you decide who had the clearer views.

    Isn’t that statement tautologically true & with little scientific content because it isn’t falsifiable?

    • Rahul:

      Yes, the thing about clearer views is grossly subjective, but I can imagine it being measured in some way. For example, you could assemble a bunch of historians and journalists who are expert in recent American history, give them a list of all the major presidential candidates in the pasts 70 years, and have them rate each candidate on a series of characteristics, including, say, likability, apparent competence, aggressiveness, openness to new ideas . . . a bunch of things, and include something like “clarity on the challenges and opportunities facing the country” as one of the things on the list. I’m not saying it’s worth the effort to do this, but it could be done.

      • Right, but till we do that experiment how do we know Safford is wrong?

        i.e. Without even knowing which candidate your hypothetical experts panel would rate as the clearer thinker how can we say the clearer thinker didn’t actually win every election?

        It all sounds very hypothetical to me.

        • Rahul:

          1. Burden is on the person (Safford) who proposed the idea. It seems pretty clear that he wasn’t thinking through the historical cases; he was just guessing at a pattern without any real reflection.

          2. If the clearer thinker did win every election, it would just be luck, given the three tied elections.

          My point is, yes, it is all hypothetical, and my problem is that Safford is doing the pundit thing of saying it’s correct, when all he has is a hunch, and not even that. A social scientist should know better.

        • I just think it’s one of those loose, ill-defined statements with very little falsifiable content to waste our time on.

        • Re #2: Maybe your hypothetical panel would be tied deciding who was the clearer thinker in those three elections?

          Hey, it’s all hypothetical after all. One can imagine crazy things!

  5. Isn’t it really more of a performative belief, i.e. “I believe this because I want to believe that my fellow Americans elect the best candidate [in some sense]”? I’m reminded of that place where Kant says that we should believe that the wicked are punished, not because it may be objectively true but because it shapes our in a certain way.

  6. You came to this conclusion based on models estimated on 15 observations? With incredible amount of confounders to boot? That’s some amazing filtering right there.

Leave a Reply

Your email address will not be published. Required fields are marked *