More GIGO. How statistical inference requires necessary conditions with respect to population, e.g, random sampling a population with a normal distribution. If these conditions are not met, inference is not valid and the reasoning fails the test. If the conclusion happens to be true, it is coincidentally and not shown by the model.
Clint Ballinger gets a shout out, too.
Lars P. Syll | Professor, Malmo University
Thanks Tom (and Lars of course). This issue it such a problem for the social sciences it is simply ignored. Needs attention.
ReplyDeleteLars' article is now over at the Real-World Economics Review here Econometrics and the art of putting the rabbit in the hat
ReplyDelete________
Clint Ballinger: On good urbanism, sane economics, & problems in the social sciences