[Previous] [Main Index] [Next]

Sunday, May 10, 2009

RATS OUTPERFORM HUMANS IN DECISION-MAKING --- OR WHAT TO DO WITH FRAUDULENT STATISTICAL MODELING?

Strange Topic As It Stands, No?

Well, note that it captures more or less what the thread at Economist View, Professor Mark Thoma's laudable web site, deals with . . . where prof bug posted his own views on the problems, more technically stated, of data-mining and model fiddling by scholarly researchers ---- and not just in economics. 

Though the topic sounds wonkish, the initial fascinating article linked to by Prof Thoma is easy to follow and documents a study carried out by a physicist who did, in fact, find that rats guessed better the color of the next toss of two balls --- red or green (which rats perceive differently, color-wise, from humans; but see the difference anyway) --- than humans did. 

That Initial Article Is by William Easterly . . .

. . . a justifiably prominent economist, formerly for decades with the World Bank ---- with lots of field experience in Africa and elsewhere --- who specializes in the problems of economic backwardness in developing countries. 

A free-market specialist, he grew skeptical of all the nostrums pushed by the World Bank and other developmental agencies around the globe since WWII --- along with almost the entire bulk of the divergent theoretical work produced by economists --- and, in the end, concluded .that the developed world has really no formula for development that the more backward countries in Africa, the Middle East, parts of Asia, and parts of Latin America can use to overcome their backwardness.  His best book --- which is written with verve, insight, and clarity --- remains one that he published in 2001, called The Elusive Quest for Growth: Economists' Adventures and Misadventures in the Tropics. 

 In it, he sets out a variety of the best-known formulas; shows that none has worked --- other than, at some point, how the more rapidly developing countries in East Asia integrated gradually into the global economy and tried to develop manufacturing industries with multinational help that would export into rich countries; shows further that the major pre-conditions of such steady development require fairly honest politicians and bureaucrats, plus effective political and legal institutions that limit predatory elite behavior and corruption; and have let markets gradually operate.

It's a laudable book, easily read by non-economists --- and full of wisdom, theoretical and practical alike.

Back to the Subject of the Thread

What Easterly's article does is show how the theoretical models produced over the decades by hordes of economists who favor this or that nostrum --- large amounts of foreign aid now back in fashion --- are generally fraudulent . . . based on what statisticians call "data-mining" in order to pluck this or that sub-set of data from an already limited data-set as the basis for running their statistical models on. 

The discussion in the Thoma-thread is generally of high quality ---- the theoretical and especially statistical ones drawing replies mainly, it seems, from knowledgeable scholars.  Prof bug adds his own slanted stuff to the discussion . . . none of it, believe me, particularly technical or hard to follow.

Click here for the thread and the long buggy commentary, posted at 9:06 A.M. on May 10th, 2009.   Oddly --- the reason really for the date and time of the post --- is that right now, a little past noon in Santa Barbara, the buggy post is picked up at Prof Thoma's web site by Firefox and Opera, but not Internet Explorer 8.  Who knows?  Maybe a malevolent E.T. is trying to sabotage the buggy professor's web-blasted ratiocinations from a hidden space ship parked in his neighbor's backyard . . . the bugged out guy back, with his family, in his house in the steep hills right above downtown and somehow earning the enmity of our outer-space visitor.