Comments on: Chartmanship — A Picture that Sparked A Thousand Words (of Criticism)

By: Steve

Steve — Fri, 22 Feb 2008 03:26:56 +0000

It seems to me, that RWs plan is to distract everyone from the actual point.
Create a big argument, and muddy the waters.

Go on RW please reply to me, so I can ignore you.

And please, lets remember "the debate is over" ROFL

Can we get back to talking about science now please.
Great work guys.

By: RW

RW — Wed, 13 Feb 2008 10:13:21 +0000

I’ll discuss my arguments with Raven with Raven, and not you, thanks.

By: Adirian

Adirian — Tue, 12 Feb 2008 22:42:40 +0000

RW, in case I hadn’t made it painfully obvious enough, I wasn’t interested in the argument between you and I, I was interested in the argument between you and Raven. I already told you you could assume you had won our argument, because I really didn’t and don’t care to have that argument.

By: RW

RW — Tue, 12 Feb 2008 16:12:07 +0000

Wow Isaac, that’s a really insightful contribution to the debate! What a clever fellow you must be, to post spam masquerading as moronic comments like that.

By: Isaac Crawford

Isaac Crawford — Tue, 12 Feb 2008 14:01:19 +0000

Wow RW, you’re an idiot. In the future, if you don’t want to look like an idiot, try answering some questions…

Isaac Crawford
Blogging in Yemen
http://www.isaharr.com

By: RW

RW — Tue, 12 Feb 2008 08:04:45 +0000

If you don’t care to provide a link to evidence that you’re quoting, then I have to assume you’re just making it up.

By: Adirian

Adirian — Mon, 11 Feb 2008 23:19:38 +0000

The individual playing at trying to argue semantics is accusing me of being an obnoxious twerp? How delightfully entertaining. I’m glad to see you’ve realized you can’t win that particular game with me, however, and have moved on to demanding evidence. (Incidentally, regarding semantic arguments – if “comparison” doesn’t describe a statistical test, I also failed entirely to describe an actual statistical test. An actual test description would involve more math than I care to write out. I implied a statistical test using a great many words, an implication which is much simpler stated as a “comparison,” where the context implies a statistical comparison, rather than some other form of comparison.)

And no, I don’t really care to provide a link. Hunting through research papers doesn’t sound like a good use of my evening right now, particularly for a test that would take much less time to reproduce in Excel than to locate, or at least would once you’ve tabularized the data. So you can claim victory over me, if you actually care to, but you’ve still effectively lost your argument with Raven.

Incidentally, it would be a wasted effort anyways – if the model is good, the predictions shouldn’t bear too much similarity to reality, for the simple and expedient reason that the predictions of greenhouse gas levels in any of the three scenarios Hansen produced simply don’t match up to reality, in each case at least one of the major three diverging significantly from prediction. Garbage in, garbage out, no matter how good the model. A retrofit of the original temperature model with modern GhG data would be a much better indicator of its accuracy, and to my knowledge, that hasn’t been done.

By: RW

RW — Mon, 11 Feb 2008 18:12:45 +0000

If you want to have a sensible discussion, you might try not being an obnoxious twerp.

Now you've admitted that 'comparing' is not an adequate description of a statistical test, and you've had a go at describing an actual statistical test. Then, though, you claim that Hansen's model failed this test, and failed miserably - care to provide a link to the paper or papers in which this test has been applied?

By: Adirian

Adirian — Mon, 11 Feb 2008 17:24:23 +0000

It is not the comparison that is the statistics, the comparison is what the statistics achieve. The test is how that comparison measures up. In the case of averages, it is a very simple, and very lossy, statistic that is used. In the case of regional distribution, it is a very complex but much less lossy approach. The comparison is then to simply see where in the distribution of the projection, which is usually but not always a normal distribution, reality falls. In the case of regional comparisons, you have hundreds of comparisons, to see how well the model predicted behavior – and you then do an additional statistic to see whether the distribution of those distributions fell into an arbitrary region of expected values. (The usual arbitrary region is the 95% range, although precise sciences use much higher numbers.) The reason you do the second step is because outliers are expected, and what you are actually interested in is how well the model performed as a whole, rather than how it performed in, say, Detroit. Hansen’s model failed this test, and failed miserably – the regional distributions of temperature were wildly off of his predictions, and it merely happens that the average of them – a single data point out of thousands – aligned.

Comparing the averages would be skipping the entire intermediary step of statistics, and, instead of seeing whether the regional distributions were approximately normal (or otherwise as distribution may be expected), you condense the distributions to a single data point, and compare THAT. As Raven accurately put it, an ice age in the Sahara and boiling temperatures in the Arctic would pass that test just as well – that is, it would perform PRECISELY AS WELL FOR THIS TEST AS HANSEN’S MODEL – because it is a lossy comparison. He wasn’t saying that it would be just as accurate a representation of reality.

If you want to argue with anyone about statistics, you might invest some time in a course.

By: RW

RW — Mon, 11 Feb 2008 10:35:35 +0000

Adirian – you’re obviously stupid, and you’re rude as well. ‘Compare’ is a statistical test? Which statistics, exactly, are you describing when you say ‘comparing’ is a statistical test?