NBR points to a newly established government website providing easy access to government data sets. Should prove fodder for potential honours projects for the next several years.
The NBR piece also points to NZ data on top baby names, which I ought to incorporate into my algorithm: anything in the top 100 in the last decade is forbidden. On similar topic, Andrew Gelman wonders why so many folks around the world are called Matthew. Popular names become more popular. Hasn't he heard of the Matthew Effect?
Tampilkan postingan dengan label Andrew Gelman. Tampilkan semua postingan
Tampilkan postingan dengan label Andrew Gelman. Tampilkan semua postingan
Selasa, 03 November 2009
Rabu, 21 Oktober 2009
State-level political polarisation
Andrew Gelman points to some really neat data work by a colleague of his, Boris Shor (jointly with Nolan McCarty) on ideological positions of US State legislators. How does he put them on a common scale? By exploiting that many members of State assemblies go on to Congress where they then get an ADA score. So long as the state legislator's ideology doesn't change in the move from state legislature to Congress (or, more importantly, doesn't exhibit any changes that are specific to any legislators coming from that state), we then have some fixed reference points in each state legislature.
Gelman provides this picture from Shor:
Some things aren't that interesting to me. Of course New York, Connecticut and Maine have both Democrats and Republicans far to the left of the Congressional averages. What's more interesting is the gaps between the parties. In Utah, Indiana, Wisconsin and California (and a couple others), there is no overlap between the Democratic range and the Republican range. In Rhode Island, there's no overlap but only a tiny difference in means. In lots of other states, there's substantial overlap between the two parties' positions. I wonder what drives that kind of state by state variation. Open versus closed primaries? Something else?
Gelman provides this picture from Shor:
Some things aren't that interesting to me. Of course New York, Connecticut and Maine have both Democrats and Republicans far to the left of the Congressional averages. What's more interesting is the gaps between the parties. In Utah, Indiana, Wisconsin and California (and a couple others), there is no overlap between the Democratic range and the Republican range. In Rhode Island, there's no overlap but only a tiny difference in means. In lots of other states, there's substantial overlap between the two parties' positions. I wonder what drives that kind of state by state variation. Open versus closed primaries? Something else?
Langganan:
Postingan (Atom)