These past few weekends, I’ve been thinking quite a bit about historic elections: 1896, the birth of the economically left Democratic Party; 1912, 1924, 1948, 1968 and 1992, the races with formidable 3rd party candidates; and of course, the Reagan landslides of the 80’s, which the Right-Wing blogosphere has been desperately trying to compare with the Romney-Obama race.
Perhaps the weirdest election I’ve paid any attention to was the 1952 election. It wasn’t particularly close and wasn’t really a blowout of epic proportions. Eisenhower was a war hero, while Adlai Stevenson, although a notable and well-respected figure in American history, was simply not up to the task of being a nominee for a major political party.
However, there are two things which make this election worth mentioning: the first was that it ended 20 straight years of near-complete dominance by the Democratic Party; the second was that the election was projected by a computer. Considering the first factor, the fact that UNIVAC was able to accurately project the election so early was even more impressive. So impressive that the UNIVAC operators and CBS Executives refused to air the original results.
Because they don’t want to discourage people on the West Coast from voting, the Networks no longer air computer projections of elections on a national level (Though perhaps they should just not allow results to be released to the media until all polls have closed, it’s a bit absurd to allow a huge chunk of results to be released before many have even voted). I’m wondering though, if they dusted off a UNIVAC I and utilized the same controls (with different data of course, to reflect political realignment) used to make the initial 1952 projection, would UNIVAC be able to make an accurate projection? I don’t see how it wouldn’t. I think it just goes to show that the political field was one of the first to make use of big data in such an accurate manner. One could even argue that Big Data was used in an analog form with the rapid growth of public opinion polling in the 1940’s, though in the case of 1948, it wasn’t necessarily all that accurate.