July 21, 2005
Defending Mann, Bradley & Hughes
One of the defenses that has been used for Mann, Bradley and Hughes is that Ross McKitrick has made mathematical errors in the past. Brad DeLong has used this defense. But we can't trust DeLong either on anything, based on this type of argument. After all, DeLong doesn't understand the concept of Confidence Intervals (if you don't believe me, go ask Tim Lambert1).
This type of defense is rather dumb, in my view. Can somebody make a mathematical error? Sure. Are we to believe that Tim Lambert, Brad DeLong and others have never ever made dumb mathematical errors (oh wait we already know DeLong has made a classic one)? Unless the answer is yes, we can simply discount everything that Lambert, et. al. have to say in defense of the Hockey Team.
Update: Looks like Tim Lambert also failed his statistics qualifying exam. According to Tim Lambert,
- On the one hand we can't make probabilistic statements about confidence intervals.
- On the other hand we can make probabilistics statements about confidence intervals.
You know, this is really great. This is the kind of thing that Lambert will focus on in his attacks on guys like John Lott. Sweet, sweet irony.
_____
1Also according to Tim Lambert we can assume that Brad DeLong failed his qualifying exam in econometrics.
Learn to read.
Posted by: Tim Lambert on July 21, 2005 09:08 AMWhat? Is this just mean spirited bashing or do you really have a point?
Posted by: simon on July 21, 2005 06:48 PMWhat? Is this just mean spirited bashing or do you really have a point?
Posted by: simon on July 21, 2005 06:49 PMWhat? Is this just mean spirited bashing or do you really have a point?
Posted by: simon on July 21, 2005 06:49 PMTim,
Sorry, you are wrong. You clearly think Ujeio got it right and he makes probabilistic statements about CIs, which you rightfully took Tim Worstall to task. Your bias and inconsistency and intellectual dishonest are all showing.
I'm sure you are used to being wrong.
Posted by: Steve on July 21, 2005 10:13 PMOh for heavens sakes. Not all probabilistic statements are the same.
Posted by: Tim Lambert on July 22, 2005 03:21 AMROFL, good one, Steve.
Posted by: SPQR on July 22, 2005 11:42 AMTim,
If we can make probabilistic statements about the "middle part" of the CI then we should also be able to make probabilistic statements about the entire thing. You know we can't and you glossed over/ignored that part of Ujeio comment. Why I don't know, but I suspect that like Lott, you did so because it supported a political viewpoint you liked.
Posted by: Steve on July 22, 2005 09:09 PMYou can make probabilistic statements and I never said you couldn't. The problem with Worstall's atatement was not that he made a probabilistic statement but that he said that the probability was 95% when there was no good reason to believe that. But that was not the reason why I said that he had failed statistics. That was because he claimed that a relative risk of less than three was not statistically significant. That's not true -- it's one of the scams that Milloy promotes.
Posted by: Tim Lambert on July 23, 2005 03:42 AMReally? From you Lancet post,
No, a 95% confidence interval does not mean that there is a 95% chance that the true value is in the interval.
The only probabilistic statement that can be made about confidence intervals are trivial (0 or 1).
Posted by: Steve on July 23, 2005 07:48 AMLet me also note Tim that you are free to make any probabilistic statements about confidence intervals you want. Of course, in doing so you leave the realm of Frequentist statistics (i.e. you are jettisoning the same machinery that gave you the confidence interval in the first place). Methodologically it is quite suspect. But of course, if you do this then you really can't take Worstall to task for his Frequentist blunder either. After all, his (subjective) probability assessment is just as good as yours--at least until you tell us why yours should is better (namely showing that your subjective probability estimate has better predictive power).
As the other commenter (at Instapundit) notes, with an interval of [8,0000,194,000] we'd not reject the hypothesis if the number were 8,000, 101,000, or 194,000. These are all consistent with the hypothesis and according the Frequentist methods none is more or less likely than the others. Frequentist methods provide no comment on which number inside the interval is any more or less likely.
So go ahead and feel free to construct a pdf over your confidence interval. Why you want to do this and use Frequentist methods over Bayesian is beyond me. Methodologically you are on rather shakey ground.
Posted by: Steve on July 23, 2005 08:25 AMOh please, using Bayes theorem is not methodologically suspect.
Posted by: Tim Lambert on July 24, 2005 08:25 AMLearn to read, I didn't say it was. What I said was that combining Frequentist and Bayesian methods into some sort of ad-hoc mish-mash amalgam is suspect.
Posted by: Steve on July 24, 2005 12:30 PMSteve: After all, DeLong doesn't understand the concept of Confidence Intervals (if you don't believe me, go ask Tim Lambert).
No need to confirm this hypothesis with Tim, Steve. It's easy for me to believe DeLong doesn't know jack about confidence intervals. The guy's a dolt (not to mention, a comment-deleting control freak).
I have never noticed this particular deficit on Brad's blog. But I haven't read it in months because it's too boring. And given the statistical illiteracy among economics professors in the blogosphere (Levitt, the boyz@gmu (especially Bryan Caplan), etc.) it wouldn't surprise me if Delong didn't understand confidence intervals.
Although, now that I think about it, my sense was that Steve Verdon was confused about confidence intervals when the topic came up here on his blog.
Hmm. I'm not 2 sure who 2 choose if I have to pick Verdon or DeLong. It's a tossup, that's for sure.
I'm gonna go with Verdon here, just to try something new and different.
Posted by: deb on July 25, 2005 01:17 PM