Thursday, April 29, 2004
Random Question: Have you ever wondered why there were so many TV shows in the fifties and sixties that featured a widowed father and his one or three sons? Fury, Bonanza, My Three Sons, The Rifleman, The Andy Griffith Show, Courtship of Eddie's Father.
And then there were variations on the theme like The Beverly Hillbillies (widower and daughter) and Family Affair (uncle with niece and nephew).
Was there a single show back then that featured just a mother and child (or children), which now, as then, was the more common single parent type household? The first one I can think of is One Day at a Time, which came a bit later, and featured a divorced (egads!) mother and two daughters. I bet someone out there has written a dissertation on exactly this issue.
This is the kind of weird thing I think about when I've been doing too much grading.
Thursday, April 22, 2004
Thursday, April 15, 2004
Monday, April 05, 2004
The Irrelevance of Winning the Popular Vote: I've been planning on writing some posts about the Electoral College, that gloriously quirky institution the founders whipped up for our endless amusement. This editorial in yesterday's Post gives me a good starting off point. In it political scientist Thomas Schaller argues that the 2004 election might produce another situation where the winner of the national popular vote loses the electoral college vote, and thus the election. Schaller's twist is that it might be Bush who ironically loses the election despite carrying a plurality of the popular vote. (By the way, this is what lots of pundits thought would happen in 2000.)
Schaller refers to this type of outcome as a "misfired" election, and goes on to predict that, "Two misfires in a row would cause enough of an uproar to prompt a genuine national conversation about whether the Electoral College system is still the best way to elect a president." Perhaps, but I want to make a point here that always seems to get missed:
Winning the popular vote is completely and totally irrelevant. Candidates are not trying to win the popular vote, they are trying to win the electoral vote, and they makes choices accordingly.
Consider football. Let's imagine that after the Super Bowl fans of the losing team -- that is the team that scored fewer points -- nonetheless asserted that their team "truly" won because they produced more yards on offense. Would people buy that argument? No. First they would say that the game is about points, not yards. Then they would go on and say that, besides, since the game is about points and not yards, that the winning team probably made all kinds of decisions that reduced its yardage total in exchange for creating or protecting a points lead, e.g., kicking field goals, running instead of passing to run down the clock. If both teams' goal was to produce the most yardage then the game would have been played very differently.
This analogy applies almost exactly to this business of misfired elections, and the continued snide statements made by Gore supporters that somehow Bush "losing" the popular vote is relevant. (Note that I am not talking about Florida here -- which is a wholly separate issue.) If the goal in 2000 was to win the popular vote then both candidates would have done all kinds of things differently. They would have traveled differently, focused on different issues, and spent their money differently, all in an attempt to win the most votes nationally, rather than the most votes on a state by state basis. Under the Electoral College candidates expend little resources on a state that they know they will win or lose easily. But under a popular vote every vote counts equally. Thus, for example, Bush might have spent far more resources on California, Texas, and New York.
Here's the punchline. If we went back to 2000, but ran the election this time based just on the popular vote, we do not know who would have won.
There are some good reasons to change or scrap the Electoral College. The occasional "misfire" of this sort is a not one of them.
Monday, March 29, 2004
Reductio ad absurdum: NFL owners are -- once again -- considering adding two more teams to the playoffs, thus bringing the total 14 out of 32. Here are a couple of supportive quotes:
Why don't we just skip ahead and create a 32 team playoff format? That way there will be excitement in all 32 cities and Erickson doesn't have to worry about losing his job, at least not because he didn't make the playoffs. NFL commissioner Tagliabue's fondest dream seems to be to have an NFL season where all the teams finish the regular season 8-8. Here's a parity design he'll love:
All teams play a 10 game regular season. The regular season provides the necessary data for seeding, but seeding matchups pit the highest seeds against the highest seeds and the lowest seeds against the lowest seeds as seen below.
A. 1st seed vs 2d seed
B. 3rd seed vs 4th seed
C. 5th seed vs 6th seed
D. 7th seed vs 8th seed
E. 9th seed vs 10th seed
F. 11th seed vs 12th seed
G. 13th seed vs 14th seed
H. 15th seed vs 16th seed
In the second round the winner of Matchup A faces the winner of Matchup B, and so on. The AFC would have the same setup and the whole thing ends up with a Janet Jackson-free Super Bowl. Why have parity based on last year's results? If you are going to level the playing the field then by golly let's do it for the teams that really need it, the ones who stink this year. Of course this sets up totally perverse incentives where teams actually want to do poorly during the regular season, but so what? Nobody cares about the regular season anyway.
It's not because Pennsylvania is a swing state, Bush goes there so often because he is trying to figure what's so great about the Liberty Bell: Sunday's Washington Post brings us this:
Several things. First, I hope the actual study is a little more sophisticated than suggested by this article. Thirty-nine percent of the trips went to swing states, where swing state is defined as those states with a 6 point or less margin. Okay. Here are the 2000 swing states:
Florida, New Mexico, Wisconsin, Iowa, Oregon, New Hampshire, Minnesota, Missouri, Ohio, Nevada, Tennessee, Pennsylvania, Michigan, Maine, Arkansas, Washington.
That is 16 out 50 states or 32% of the states. In other words, 39% of Bush's trips went to 32% of the states. Perhaps taking Texas and Virginia out of the mix or, better yet, a reasonable multivariate analysis might show something a little more impressive. But the evidence cited in this newspaper article is lame. (I'll see if I can get a copy of the actual research article.)
Despite this data, it is certainly not a new argument -- at least among political scientists, maybe we don't send out enough press releases -- that presidents use travel for political purposes. But Clinton was hardly the first to do this, it goes back to, oh, James Monroe. Why do you think John Kennedy was in Texas in November, 1963? The Republicans during the Clinton years were just good at getting myopic journalists to make something old and routine sound new and scandalous. Every trip a president makes has political and presidential elements to it and those do raise ethical questions since taxpayers are footing the travel and security bill. This travel might include a tendency to go to swing states -- though that's not proven in the above data -- but it also means going to states to raise money -- New York might be a good place -- and to campaign for others.
By the way, it is not just domestic travel. Brace and Hinckley demonstrated a while back that there are some distinct political patterns to the timing of a president's foreign travel.
Four for Four: And just because it never hurts to beat a dead horse a little more: do some simple bracket rearranging and the Final Four could just have easily included Cincinnatti or perhaps even Gonzaga. (Or play the tournament a few hundred times and maybe even "First Seeds" Stanford and Kentucky will get in once or twice.
Now if I were in office pool -- which of course I am not -- but if I were it would turn out that while I am -- or would be -- the only person to get all final four teams correct, I will still lose the pool because I did (or would do, if I were in a pool) rather poorly at predicting the second round. Go figure.
Wednesday, March 24, 2004
The Pork Politics of Homeland Security: Time magazine this week includes a fascinating article by Amanda Ripley on the allocation of homeland security dollars. Here are some pertinent bits:
Read the article, because it has some nice nuance, but the bottom line is that money is being spread out not because of any direct assessment of risk but because of the nature of Congress, especially the U.S. Senate. California, which has two high risk areas in LA and San Francisco, has as much representation in the Senate as Wyoming, which has no high risk areas. But it's natural for a small state senator, who is hired and fired by the citizens of his or her state, not of the entire nation, to want just as much money as everyone else. After all, Casper, WY or Burlington, VT or Santa Fe, NM might be victimized by terrorism. It takes a majority for a spending bill to pass and small state senators make up a larger proportion of the senate than do large state senators.
Blaming Congress for this is a bit like blaming a scorpion for its sting. Plus bureaucracies, highly attuned to the desires of members of Congress, will tend to conspire in the same distribution game. The one actor that has the ability to overcome these natural parochial tendencies is the president. But, of course, it is easy for the president to fall prey to his own parochialism. After all, the high risk areas we are talking about are not exactly areas of substantial Bush support.
Monday, March 22, 2004
Bush's Approval: I was playing around with approval data while putting together a talk. I thought I would compare Bush's approval with that of other presidents during the same point of their administrations, i.e., March of the fourth year. Here's the breakdown going back to Eisenhower (which is more or less the beginning of time for public opinion analysis):
Bush 2.0: 50
Bush 1.0: 42
Everyone above Bush 2.0 was re-elected the following November (counting the Johnson case as a re-election). Everyone tied or below Bush lost. One shouldn't read too much into this data. It's only a bunch snapshots, eight months from the election, and each have an error term. I think the real lesson here is that the Bush is right on the edge and he needs the employment picture to improve a bit.
Here is each president's approval the following October (sans Bush 2.0, of course):
Bush 1.0: 34
For reasons I'll get into later, I think the states vote uniformly enough now that Bush will win easily if his approval gets up around 55. He'll probably squeak by at 52.