Very freakonomic (and I mean that in the best sense of the word).
3 thoughts on “U-Haul statistics”
A commenter there points out that it might just be the periodic difference between Augusts and Octobers, not a secular difference between 2005 and 2010.
I'd also say U-Haul might not have been the only game in town. Maybe they had a rival in Vegas in 2005, so that instead of detecting a greater willingness for angelenos to move to Vegas in those days, they were only detecting a greater willingness for angelenos to hire U-Haul to get them there.
"In one sense, this analysis is very Freakonomics-y… In another very important sense, however, there is something unFreakonomics-y about it. If we want to know about American migration patterns, there are easy ways to use census data to do it. In general, I [Steven Levitt] am not in favor of doing things the hard way, just for the sake of it being hard."
Good catch! But if you read on, you'll see that Levitt does actually think the U-Haul analysis is potentially a good idea.
From a statistical point of view, the point is that the U-Haul data should be used in addition, not instead of, the census data. To the extent the two data sources disagree, maybe something can be learned. (Maybe what is learned will have more to do with U-Haul than with migration patterns, but, hey, U-Haul is important too!)
A commenter there points out that it might just be the periodic difference between Augusts and Octobers, not a secular difference between 2005 and 2010.
I'd also say U-Haul might not have been the only game in town. Maybe they had a rival in Vegas in 2005, so that instead of detecting a greater willingness for angelenos to move to Vegas in those days, they were only detecting a greater willingness for angelenos to hire U-Haul to get them there.
"In one sense, this analysis is very Freakonomics-y… In another very important sense, however, there is something unFreakonomics-y about it. If we want to know about American migration patterns, there are easy ways to use census data to do it. In general, I [Steven Levitt] am not in favor of doing things the hard way, just for the sake of it being hard."
—
http://freakonomics.blogs.nytimes.com/2005/09/08/…
Alex:
Good catch! But if you read on, you'll see that Levitt does actually think the U-Haul analysis is potentially a good idea.
From a statistical point of view, the point is that the U-Haul data should be used in addition, not instead of, the census data. To the extent the two data sources disagree, maybe something can be learned. (Maybe what is learned will have more to do with U-Haul than with migration patterns, but, hey, U-Haul is important too!)