This weekend saw the truth finally finishing with her boots, when Phil Jones, former head of CRU, admitted to the BBC that there has been no statistically-significant global warming.

I also calculated the trend for the period 1995 to 2009. This trend (0.12C per decade) is positive, but not significant at the 95% significance level. The positive trend is quite close to the significance level.There is much goodness to be had from that Q&A, among others Jones' admission that the Medieval Warm Period was probably real. This would imply that the warming of the nineties is hence not human-induced. (see also WUWT here (original reporting) and here (commentary).

But as someone with a bit of statistical schooling, the quote given above proves to me the decrepit state of climate 'science' (and it infuriates me no end in the process).

Proper statistical hypothesis testing means formulating a null-hypothesis (H0) and an alternative hypothesis (Ha). The H0, for statistically technical reasons, is usually framed as a negative, while the Ha is expressed positively. As an example: If one would like to test the differences in IQ between a sample group of men and one of women, the H0 would be 'There is no difference in IQ between men and women', while the alternative Ha would be 'There is a difference between the IQ of men and women'.

The way hypothesis testing works is showing that H0 cannot hold and therefore Ha, by inference, must be true. In fact, statistical hypothesis testing is only able to reject (or not) the null-hypothesis. In science, a threshold of 95% significance is used, meaning there is a probability of 5% or less that H0 is true, given the data. Or, in slightly more technical terms (but the importance if this is considerable): It means there is a probability of 5% of observing Ha when if fact H0 is true.

In terms of the trend in global temperatures, the H0 must have been 'There is no trend in global temperatures between 1995 and 2009', while the Ha would have been 'There is a positive trend in global temperatures between 1995 and 2009'. The admission by Phil Jones that that he wasn't even able to reject the H0 hypothesis (let alone demonstrating that we humans are responsible) is a HUGE admission.

It doesn't matter a damn if he saw a positive trend, or that it was 'quite close to the significance level'. The simple fact is that the trend he observed was NOT significant, was not conclusive proof that H0 should be rejected. By his own admission he was not able to show that H0 does not hold. Ergo H0 does still hold, ergo there is no trend, positive or otherwise, to global temperatures between 1995 and 2009. Had he been a scrupulous, conscientious scientist he would have reported the fact and that would have been the end of the matter. But instead he chose to ignore his own calculations and latch on to the UN induced panic for ideological reasons and/or precious grant money.

This admission alone should be reason enough to cast Phil Jones (and all those like him) outside, where there is much wailing and gnashing of teeth. Phil Jones is a charlatan, as are all those like him.

Also reporting:

EURef

Jim Hoft

## 6 reacties:

Falsifying the nul hypothesis does not automatically verify the alternative hypothesis. In a complex scenario, there many well be several tenable hypotheses. Scientific method would require gereration of as many plausible hypothesis as are indicated by the preliminary evidence - and experience - each then to be "attacked". As a general principle, the one that best survives the attacks is not the truth, but the closest approximation to the truth ... until new evidence emerges, either to falsify or modify it.

Where you are dealiing with multiple hypotheses, and possibly multi-factoral causation, crude statisitical analysis very rarely works. My supervising Prof used to say to me, "if you need statistical methods to demonstrate a relationship, there isn't one!". He was not entirely joking.

Hi Richard, thanks for stopping by.

Falsifying the nul hypothesis does not automatically verify the alternative hypothesis. In a complex scenario, there many well be several tenable hypotheses.True, of course. But the point in this story is that by his own admission Jones was not able to reject H0. So, whatever alternatives he had, there was no reason for him to assume those more correct.

Scientific method would require gereration of as many plausible hypothesis as are indicated by the preliminary evidence - and experience - each then to be "attacked". As a general principle, the one that best survives the attacks is not the truth, but the closest approximation to the truthWell, as a general guideline this would be a fine principle. However, assuming the trend Jones was the best fit on the data (minimal error variance) any alternative he could come up with would be even less significant. Had Jones not been bothered by the usual null-hypothesis he could have presented the trend as the best fit and hence 'true'.

To Jones's credit, he did exercise minimum of statistical decency, in that he tested against an properly formulated null-hypothesis. But after all the filtering, deletion of stations and 'adjusting' of records Jones was still not able to reject the H0. That should have been the end of the matter right there: There was no proof of warming in the period 1995-2009. That he subsequently could not reject H0, but proceeded to ignore his own results and preach the gospel of Gore anyway is wholly unacceptable. He should lose his job and reputation over this.

Climate science was never science anyway. It was politically driven. and if a hypothesis was posited that politics/taxation was the main driver, then that would be the best hypothesis going.

As for the science, if it requires statistics then there is no science.

The truth can topple the grandest of lies, as Dr North has ably demonstrated.

The truth can topple the grandest of lies, as Dr North has ably demonstrated.

The truth can topple the grandest of lies, as Dr North has ably demonstrated.

Een reactie posten