Pedro Magalhaes writes:
A quick note on your 538.com post on “All Politics is Local?”
In a paper I [Magalhaes] presented at the last APSA meeting. we address this issue in presidential elections. We use wavelet analysis – originally employed in physics and geology – to estimate the spectral characteristics through time of the presidential election returns at the national and the state level, and develop a tool to estimate the dissimilarity between those electoral cycles. We find:
1. That, historically, the South and the Deep South form two clusters where the cyclicality of presidential electoral returns has been significantly different from that prevailing in the remaining set of “core” states.
2. That, since the 1950s, dissimilarity in electoral cycles between the South and both the national cycle and that which prevails in the “core” states has diminished for most states.
3. That the group formed by Alabama, Georgia, Mississipi and South Carolina continue to display an ebb and flow of electoral changes that seems to be quite different from that we find elsewhere.
I don’t have a chance to look at this right now but I thought I’d pass it on to all of you.