FUN WITH STATS
Matthew Yglesias makes an interesting argument about a statistic he found in the LA Times:
In every election since 1960, the party in the White House lost when the unemployment rate deteriorated during the first half of the year. If the rate improved, the party in the White House won.
Says Matt,
But how sure are we that it isn’t a coincidence. 1960 was a long time ago, but we’re talking about a sample of just 11 elections. You wouldn’t do a public opinion survey of eleven randomly selected people and believe that there was any statistical validity to your results.
Well, when your entire universe is eleven, a sample of eleven is quite good indeed!
Matt’s point is a good one, though. The problem isn’t just that the number of elections is too small and the number of variables that could reasonably be expected to impact the outcomes too large to draw many hard conclusions. Analysis is further complicated by rather significant changes in the nature of the contests.
For example, Matt points out that 1960 seems an arbitrary starting point for this “study.” But, as we know, 1960 was the first real “television election.” We haven’t elected a short, bald, or bearded president since then. The major ideological battles, campaign finance rules, and the nominating process have all undergone numerous and significant changes over the years.
Also, while it certainly doesn’t prove anything, I would also note that the political parties don’t seem to believe in economic determinancy. If they did, one would think they would always rally behind their most ideologically pure candidate rather than trying to find an amiable centrist. If the election is simply a referendum on the economy, then ideology is irrelevant and there’s no point in selecting a compromise candidate.
James writes: “We haven’t elected a short, bald, or bearded president since then.”
John responds: “Damnit!”