Freakonomics at work

I have been taking a class called Econometrics this semester. We learn how to do statistical analysis on any data that has numbers or can have numbers put to it. Most of the book “Freakonomics” uses econometrics, at least in part, to arrive at the conclusions they have. I’m working on a project that is somewhat akin to that. Basically what I’m doing using data on the price of baseball cards and testing to see if there is racial bias in the price for a baseball card. The price of a baseball card is the result of the demand for that certain card. The demand for a card is usually based on how successful that player was. So using the career statistics for every baseball player in a card set I’m testing to see if there are black baseball players that have high career stats but a low priced baseball card. It should be interesting to see what the final results are. Perhaps I won’t find anything. Perhaps I will. Who knows.

On a related note, I had a professor last year who was involved in a project with Justin Wolfers to test for racial bias in foul calls by referees in the NBA. They used box score data from the 1993-94 season all the way to the 2003-04 season to test for any racial bias in foul calls. The media went up in arms, the NBA criticized it and individual basketball players took their own shots. When you look through the comments from the media and then actually read the study you realize how oblivious some people really are. The comments from NBA players were amusing. But none of them read the study, just heard about it. Charles Barkley was my favorite. He said that of course black players receive more fouls than white players; there are more of them in the NBA. Then he said "The thing that bothered me most is these guys didn't even go to the game. They looked at box scores." I don’t think that in order to crunch numbers, you have to go to every single game for 10 years just to get the “feel” for it.

I think I found something that people who report on studies get wrong often. When you read statistical studies you will find the words “significant” and “not significant” attached to the numbers reported. Real world significance and statistical significance are to different things. Significance in statistics means that the data is consistent enough to confidently say that there is an effect. Let’s say we find that the average price of a specific bag at a normal store is $30. Now the average price of the same bag at stores by the beach is $50. So we can say that beach stores sell bags for $20 more. Now what if the price from each store looked like this: Store 1: $10, Store 2: $70, Store 3:$20, Store 4: $30, Store 5: $90, Store 6: $80. The average is still $50, but the prices vary so much that it isn’t very significant. In the case of the NBA study a player receives 0.18 fewer fouls from referees of his own race than another race. This is during 48 minutes played. The results are “significant” meaning that the variation is small enough to not just pass it off as random variation. However, to a basketball player that means that assuming he plays 48 minutes in every game and that all 3 referees were a different race for all those games, he would get about 1 foul call race every 5 games. Who is going to notice that? No wonder the basketball players haven’t noticed any bias. For all practical purposes, there isn’t much. The study shines more of a light on the fact that on split second decisions, the subconscious part of our brain can have an effect.

Well this is getting long and it’s probably extremely boring to everyone except me. And maybe my Uncle Ken… and Craig… and perhaps Danielle… Sorry to everyone else!