Wednesday May 31, 2006
Ok, so lets review. In elections for Australia’s (federal) House of Representatives, voters are required to fully enumerate their preferences: i.e., rank order the K candidates, from 1 to K (thats the short version; see more detail here). In the states of Queensland and New South Wales, state legislative elections use “optional preferential” voting: in this system, you may fully enumerate preferences, or simply put a “1″ next to your 1st preference, and leave it at that. The Queensland Labor Party has used this to great effect: asking voters to “just vote 1″. This proposal, if voters adopt it, turns the election into a simple plurality election, rather than situation you get under full preferential voting where a candidate requires a majority of votes (before or after the transfer of preferences) to be elected. What does this have to do with anything…?
David Freedman was down yesterday for a talk in our MAPSS series. His points were pretty simple: (1) experiments are superior to observational studies if you want to estimate causal effects; (2) the default position should be to analyze experiments as experiments (i.e., simple comparison of means), rather than jamming in covariates and even worse, interactions between covariates and treatment status in regression type models. He also laid out a template for writing up a study that uses experimental data: Table 1, demonstrate that randomization worked, with differences of means/proportions on covariates plausibly related to the response by treatment status; Table 2, estimates/inference for treatment effects; then and only then might you start fritzing around with covariate-adjustment via modeling.
Friday May 26, 2006
Jas Sekhon at Berkeley has been talking about the Steve Jobs reality distortion zone for a while, pointing out that OS/X has some significant speed disadvantages relative to linux. Sure. He’s posted a nice discussion on how R on OS/X and just about anything else on OS/X will run slower relative to linux, even on the same Intel-based processors, due to some nagging issues in the way the OS/X kernel handles memory allocation/de-allocation. Apparently he got the engineers at Apple to chime in on this one, helping him get R on OS/X running faster but still kept coming up against some drawbacks due to the malloc used in OS/X. Jas rightly suggests that perhaps we build R against an optimized version of malloc that seems to work faster. We’ll see…
so on the United Mileage Plus website you can bid your miles in auctions for all kinds of goodies. Get this: the current bid on some time in a United simulator is approx 250,000 miles, while fighter pilot for a day is currently attracting bids of just over 100,000 miles, and a visit to United’s San Francisco maintenance facility, just over 20,000. Revealed preferences…!
Wednesday May 24, 2006
laughed out loud at this…
Tuesday May 23, 2006
Malcolm Mackerras has been on this one for a while: that Bennelong is increasingly a marginal seat and if it looks sufficiently dodgy, Howard will resign, rather than risk the ignominy of losing his seat. Malcolm’s latest serve on this appeared in Crikey on May 22, 2006:
Tuesday May 16, 2006
Stanford PoliSci is fortunate to get a few PhD admits from the armed forces every year. A recent grad is currently directing the Combating Terrorism Center at West Point, and sent through the following.
This could be really interesting, depending on how it is set-up, and the nature of university/private-foudnation mix that gets it going, emphasis given to scholarship etc. The Australian Prime Minister, John Howard writes:
Monday May 15, 2006
these are a little mellowed out. but why not?
more homage (49MB) to Cathy Horyn of the New York Times.
and something downtempo… (32MB)
A very nice piece by Peter Brent (of mumble fame, and now, enrolling the people) is in the most recent issue of the Australian Journal of Political Science. Bottom line: the Australian ballot was not the world’s first secret ballot, just the world’s best version of the secret ballot, in large part because it standardized voting procedures (the voter is handed a generic ballot, not color-coded by candidate or anything like, the voter retires to a private space, marks the ballot, deposits it, such that no one can verify the choices made by any particular voter).