Contrary to expectations, I ended up not skipping any of my twice-weekly column appearances as a result of my recent visit to Toronto (gotta occasionally put in face time with that gay/Zionist/abortion-loving mainstream-media mafia, you know). In fact, I'm not quite sure in retrospect how I failed to miss one. I have the uneasy sense that I've been the victim of some shell game by my editors. At any rate, let's run down what you might have missed. One week ago I commented on monobrowed Scots cherub Susan Boyle; in Tuesday's paper I contributed to our comment series celebrating the virtues of print (though the attentive will note that wood pulp plays no necessary role in my vision of the newspaper, as outlined therein); and in today's paper I look at the latest new piece of significant research into ADHD drugs and note the increasing cognitive dissonance surrounding them.
(A point I would have liked to have had space to make about the Berkeley ADHD study: the p-value for the observed effect on math scores in medicated children was a barely bar-clearing .04. Which means that even accepting all the assumptions in the model, and ignoring the possible existence of non-published studies that led to null findings, such a result could be due to chance—if there were no real effect—one time in 25. The gains on the reading side were much more significant, but I have to think a researcher who puts such a core result in a study destined to be hyped must still have to take a very deep breath first.)
Comments (3)
Know why I hate the Twitter? Because journalists love it, and seem unconstrained by it. So why have I spent half a century reading the other 675 words of all those pieces? Just so they could pay their mortgages and brace their children's teeth?
Posted by Crid [cridcridatgmail] | May 2, 2009 4:23 AM
Posted on May 2, 2009 04:23
A p-value of 0.04 is bar-clearing only in the sense that 0.05 the conventional wisdom for acceptability.
A real, bar-clearing p-value depends on the balance of errors that you fear: it would be unacceptably high if I'm sending someone to the electric chair, but way too low if I'm trading foreign exchange futures. I make no assertion about this particular study, but I only want to say that acceptable p-values have not been brought down from the mountaintop by Moses.
Posted by NewAlgier | May 15, 2009 12:26 PM
Posted on May 15, 2009 12:26
JX1nTn hi! mi site is http://norffg.com
see you!
Posted by marvell | May 27, 2009 4:21 PM
Posted on May 27, 2009 16:21