13

The Olympic bump (II)


 

A libertarian I knew in high school—though I don’t believe he was an avowed libertarian at the time—passes on the abstract of a study that would, in theory, lend some credence to the theory that the Olympics could benefit Stephen Harper’s hopes.

We leverage a natural experiment to explore whether personal happiness unrelated to incumbent performance affects voting behavior: the outcome of the local college football team’s games right before Election Day, an event that government has nothing to do with and for which no government response could possibly be expected. We collected football scores from 1964-2006, as well as county-level election returns for presidential, gubernatorial, and senatorial elections. On average, a pre-election win causes the incumbent to receive about one percentage point more of the vote, with the effect being larger for teams with stronger fan support.


 

The Olympic bump (II)

  1. So if the Conservatives can schedule an election within a 24 hour time window of the Olympics, they could get a one percent blip. Though the timing of any election next year would be entirely dependent on the opposition parties. So…

  2. I'm not yet sure what I think about the theory that happy people usually vote for incumbents, I haven't really read enough to decide one way or another.

    But I'm inclined to value this paper less (vis a vis its usefulness in a Canadian application) for two reasons, stated in no particular order: One, there is an entirely different sport culture here than in the US, and Two, we have a different electoral setup.

  3. So if the Conservatives can schedule an election within a 24 hour time window of the Olympics, they could get a one percent blip. But the timing of any election next year would be entirely dependent on the opposition parties. So…

  4. On average, a pre-election win causes the incumbent to receive about one percentage point more of the vote, with the effect being larger for teams with stronger fan support.

    CAUSES? Total and complete credibility FAIL. Even if there is an unassailable statistically supported association, no scientist worth anything would have placed "causes" in the abstract like that.

    • While you have a point, that correlation does not equal causation, it would be highly unlikely that the election results cause the home team to win a football game that occurs before the election.

      • Exactly Dan.

        The only other reason not to infer causation between X and Y is some Z variable causing both things to happen. While possible that's also unlikely here.

        I see no problem with the abstract.

        • There is also the possibility that Y causes X. It is theoretically implausible but stuff like that does sometimes show up. For instance, if you do a granger causality model of Christmas cards and Christmas, Christmas Cards granger "cause" Christmas.

    • Though you are likely right, there are a lot of ways you can get at causality in a paper like that. Typically, when we think of causality, we think of three things – time order (does X happen before Y), covariation (do the two tend to move in consistent patterns vis-a-vis each other) and non-spuriousness (have we accounted for Z). Regression analysis can tell us about covariation. Because this study has a time dimension, we could use a granger causality model to get a measure of time order. If this study is well constructed, it will account for the most important Z's (you don't need every single variable in your model for an assessment of your independent variable – so long as excluded variables are not related to X and Y, you aren't going to bias your results).

      Though the study almost certainly didn't do this, it could also have done some process tracing – perhaps following individuals in a particular case. From this one could assess whether, qualitatively, there is a strong case for the argument. This would also get around the risks of the ecological fallacy in using county-level data (we don't necessarily know whether or not it is sports fans that are voting for incumbents because we don't have individual data).

      Social science, statistics and other methods are capable of finding much more than mere correlation.

  5. What about the reverse? What happens if the athletes don't do well, or if something goes haywire in the Games itself?

    • That was my first thought, as well.

  6. There is always the risk that something will go wrong with the Olympics, but the most likely outcome is that Canada will bask in the spotlight for a few weeks and Canadian athletes will win lots of medals, causing national morale to soar.

    • Yes, that is true. Am I the only one who was on constant pins and needles during the Calgary games worrying that something would go dreadfully awry? No reflection on Calgary; any Canadian city, I am sure, will get the same reaction from me.

      Anyway, interesting theory. In my heart I feel that there is some merit to it, although my head certainly needs more evidence than that. And I don't want something to go wrong at the Vancouver Olympics, just to prove the opposite of the theory!

  7. On the other hand, talking about it may lead to it cratering.

    I seem to remember Brian Mulroney musing about how the 1992 World Series might help out the Charlottetown Accord in its referendum. It still went down to a resounding defeat.

    (But maybe it would have been worse still…)

Sign in to comment.