It would be cool if I could tie all those together, but this is actually several thoughts in a row.
First up, we all know that Clinton won Texas, right? Well, only sorta. But sorta not. She definitely won a narrow victory in the Texas primary, 51% to 47% and therefore won 4 more delegates than Obama. That's a victory, no problem there. But, there was also a caucus that night, which Obama happened to win. The caucus only divies up about half of the delegates of the primary, but, here's the kicker, Obama won the caucuses 56% to 44%, twice the margin of the primary. If you take half the delegates but win by twice as much, that comes out even. Clinton won the popular vote, but did not win (maybe a 1 delegate take for Clinton or Obama) the delegate vote, and the delegate vote is what decides the nomination. Despite this, the perception is and will remain that "Cinton won Texas".
Second up, if you are interested in politics, the NY Times has a great election results page that I've lost far too much time to already. Here it is.
Once you go into a state, you can also see both county by county results and exit polling data. I wanted to focus on the polls to talk about causation, which relates to my dissertation.
There's tons of fascinating stuff in these polls, and I spent some time comparing Texas to Ohio yesterday. One result had to to do with the questions: "do you think gender is important" and "do you think race is important". The majority of Democratic voters in both states usually said neither was important, with people who thought it was important maxing out at 20%. Of that 20%, some results are predictable, others surprising. For instance, in both states, if you thought gender was important, about 60% voted for Clinton. Makes sense. If you did not think gender was important, the voters split evenly 50/50.
In Texas, both if you thought race was important and if you thought it was unimportant, the voters also split 50/50. However, in Ohio, if you thought race was important, 60% of voters voted for... Clinton, not Obama. The not important people still came in at 50/50.
How do you interpret this little statistic of race-importance equalling pro-Clinton votes? One way to interpret it is that thinking race is important caused people to vote for Clinton. Race is important was a reason to vote for the white candidate. This is certainly a real possibility, but it's not the only one. Instead of thinking about race as a cause of a Clinton vote, they could also be common effects of some other cause, let's call it unknown cause X. In this possibility, Cause X both causes one to think race is important and causes one to vote for Clinton. Sometimes it's helpful to think of little diagrams to follow these things (but I can't figure out how to do so in this blog entry). In scenario 1, you have a box for "race is important" and a box for "vote Clinton" and a causal arrow leads from the first to the second. In the second scenario, there are three boxes. One box is for Cause X and it has an arrow going to the other two boxes, like a Y shape. In this scenario 2, thinking about race had nothing to do with the voting pattern. If you could erase all beliefs about race from the voters heads, they would still vote the same way (for Clinton) because Cause X is making them do it. Cause X also happens, however, to have some effect on beliefs about race importance.
Does this make any sense? Let me see if I can give an everyday example.
Let's say that 60% of drivers who wear winter coats in their car act irritably to their passenger. One causal model is that wearing coats makes people act irritably. Another possibility is that when it's cold in the car, drivers put on coats and act irritably. The coat isn't causing the irritable behavior, the cold is. In fact, if you removed the coat, they would act just as irritably as before -- or maybe more so.