I smite thee! ...The Diary of Nod

Rain-free July 4th Slightly Associated With Voting Republican

Nick Carbone at Time Magazine wrote an article, "Do July 4th Parades Turn Kids into Republicans?" The title sounds pretty sensationalist and reading the article fits the bill.

And the stats from Harvard back up the claim. When children were exposed at a young age to the Americana fanfare associated with the Fourth of July, they were more likely to vote Republican and make campaign contributions to the party.
...
[K]ids who attended one rain-free July 4th celebration before the age of 18 were four percent more likely to vote Republican before the age of 40. And whether or not they went to the polls, young Independence Day revelers were more likely to identify with the right wing by 2 percent.

All I could think is, "There's no way a Harvard paper said that." Well, I was right, but only barely. Really, I underestimated what the Harvard paper DID say.

See, they couldn't actually measure who attended July 4th celebrations in childhood and how they voted as adults; instead, they use rain as a proxy for attending a celebration and the birth year and birth place as proxy for where a voter spent their childhood.

From section 1, "Introduction" (p2):

We thus use the absence of rain as a proxy for participation in holiday celebrations on Fourth of July.

I know, I know, I've gotta be full of shit, right? There's NO way a Harvard paper by an associate professor at the Harvard Kennedy School wrote that. I've GOT to be full of shit. Or not.

From section 3, "Methodology" (p8):

The ideal experiment to estimate the effects of Fourth of July would be to allocate participation in he event randomly to some individuals and not to others, and then compare preferences and be- havior across the two groups. In the absence of such evidence, and because we cannot audit actual participation nor control for unobserved factors likely to motivate those who join the festivities, it s difficult to measure the impact of celebrating Fourth of July.
The key innovation of this study is to exploit random day-to-day variation in precipitation to estimate the effect of Fourth of July celebrations.

From section 3.1, "Data" (p8):

We rely on information from two sets of data. The data on rainfall comes from NOOA National Data Centers (NNDC). The data on political preferences and political behavior is taken from the ANES.
...
In a majority of the surveys, the data contains information on the county of residence. A limitation is that we lack information of the county of birth or county of residence in childhood. When investigating the long-term effect of Fourth of July, we would like to measure rainfall for an individual during childhood. Since we only observe the county of residence at the time of the survey, we match at the county of residence, leading to a measurement error problem. However, in most surveys, data is available on the region of birth.

And after all of these assumptions and maybes, what they find is a slight favoring to vote for Republicans; moreover, they claim causation instead of mere correlation.

Finding causation is very difficult (unless you're at Harvard, it seems), because you have to prove that This And Only This could have caused the result.

From section 4.2, "Voting Behavior" (p13):

The results above show that Fourth of July celebrations in childhood cause individuals' political preferences to permanently shift toward the Republicans.

I can only sum up the paper as, wipe your tush with it, it's already full of shit. It's 40 pages long but the bullshit stares you straight in the face.

Labels: Andreas Madestam, David Yanagizawa-Drott, harvard, nick carbone, time magazine

posted by Lachlan @ 10:55 PM