Also, a quick note: I'm over at Greg's place today, so blogging here might be erratic.
On to the good stuff:
1. Andrew Gelman has a note worth reading on where the biases are when (some) political scientists consider reform. I mostly don't agree with him (and I'll try to give a real response if I get a chance)...but it's well worth reading the other side, so I recommend it.
2. Sarah Kliff on where the doctors aren't.
3. Brendan Nyhan, on why we shouldn't pay too much attention to small demographic splits within horse race polls.
4. You may have seem something last week about GOP (possible) misbehavior in the Michigan legislature. Kevin Drum had the same reaction as I did -- sounds plausible, but what was this really all about? He looked into it, and tentatively concluded:
So in some sense this is similar to Republican abuse of the filibuster in the U.S. Senate. It's not outrageous in the sense that Republicans are doing something unprecedented and anti-democratic. It's outrageous in the sense that they're playing Calvinball: breaking a norm of governance by taking a tradition that was used occasionally in the past and turning it into a routine part of party politics.
5. And awesome campaign posters.
Of course you shouldn't pay any attention to demographic breakdowns of horse race polls.
ReplyDeleteHorse race polls measure statistical noise to begin with. There is no statistical validity whatsoever in watching percentages of 300 voter samples go up and down a few points day to day. In other words, they don't even measure the overall race.
The reality is that most political polltaking is about giving the media and academics something to talk about and justifying the salaries of overpaid political consultants. It's psuedo-science.