Yesterday I posted on a election prediction study from the University of Colorado. In the comments a few readers wondered about the track record of the study from previous presidential elections. This issue is addressed in a LA Times article about the study:
But Bickers said much of the polling thus far means relatively little, with much of the electorate still not focused on the race. The academics said their model focuses on the preeminent issue of the economy. Applied retrospectively, the model predicts the correct winner in every presidential contest going back to 1980, they said.
The model missed by an average of 20 electoral votes when applied to those prior elections, Bickers said. He said the formula employed in the analysis is unusual because it measures the states, rather than trying to predict a popular vote for the entire nation.
Of course past performance is no guarantee of future performance. The number of presidential elections in history is so small that any analysis of them is statistically meaningless. For example, from the UC news release on the study:
Their results show that “the apparent advantage of being a Democratic candidate and holding the White House disappears when the national unemployment rate hits 5.6 percent,” Berry said. The results indicate, according to Bickers, “that the incumbency advantage enjoyed by President Obama, though statistically significant, is not great enough to offset high rates of unemployment currently experienced in many of the states.”
With unemployment rates 8-9%, one could argue this is a relevant point. But on the other hand, this could be the first presidential election where the incumbent advantage doesn’t disappear under these circumstances–maybe because the media bias is so strong this election cycle.
My overall impression of the study hasn’t changed. It’s an encouraging result but no reason to stop aggressively campaigning.