Broback's Blog

A favorite of mine from Fooled by Randomness:

“It is a mistake to use, as journalists and some economists do, statistics without logic, but the reverse doses not hold: It is not a mistake to use logic without statistics.”

I frequently see conclusions from those who lean on tortured data sets in order to reach preferred policy conclusions riddled with internal inconsistencies. Several of these show up in the many documented “Krugman Kontradictions.”

Not sure these two things can be separated. It appears to me that internal consistency is a critical element of robust statistical analysis.

A personal example. For decades, a good friend and I have done simultaneous eye rolls whenever we hear people say things like “I can’t sell my house” (or car, boat, concert ticket etc. etc…) In the decades we’ve hung out, it’s come up scores of times. We agree that if you “can’t sell” something, you are charging too much. In other words, demand curves slope downwards.

Sadly, this same friend will enthusiastically reference Steve Keen and his notion that demand curves can be any shape. And in any particular instance where it’s required, he can make (or find) the data appear to sing that song. Of course the contrary data exists too. One can make the data look any way one wants. The difference is the logic.