Friday
Jun102011
by 
Bishop Hill
Bishop Hill Jones: post 1995 warming "significant"
Jun 10, 2011 Climate: Jones Climate: Surface Phil Jones has announced that post-1995 warming is now "significant", with new data changing the picture he had previously reported to Roger Harrabin. The news comes via Richard Black, in one of those "we don't want anyone disputing this, so we've switched commenting off" articles.
By widespread convention, scientists use a minimum threshold of 95% to assess whether a trend is likely to be down to an underlying cause, rather than emerging by chance.
If a trend meets the 95% threshold, it basically means that the odds of it being down to chance are less than one in 20.
Last year's analysis, which went to 2009, did not reach this threshold; but adding data for 2010 takes it over the line.
I wonder what he makes of the Koutsoyiannis paper then?
Reader Comments (86)
Don't have time to read everything Jones said, but as far as his statement to the BBC some months back goes, he's probably right.
Temps have risen slightly over the past 15 years and it's now a long enough period to be labeled statistically significant.
Context is king, of course... and I doubt if I'd get that from Mr. Black, which is why I'm not chasing the link. But Jones was clearly trying to say that months ago and twisted his own words up enough to cause a controversy. Hasn't he had enough grief?
It is now June 2011 . How come it has taken Jones and co. 5+mths to analyse data which must have been known at least 4mths ago ? The work rate at the CRU is amazingly slow. The only conclusion is that there people out there ,with plenty at stake ,who are getting very worried for some reason.
1995-2010. What combination of Spanish does that represent?
La-Nina to La Nina?
La Nina to El Nino?
El Nino to La Nina?
El Nino to El Nino?
Just asking.
Tom Fuller, appreciate your point, but if 2010 was predictably a warm year due to the pacific thingy wotsit, and 2011 is going to be a cold year due to the pacific thingy wotsit, where is the signature of CO2?
If 2011 cancels out the rise recorded in 2010, what does Jones post in a years time? Ooh er just a leettle bit more, or ooh er just a leettle bit less?
As it is likely to be less, our BBC idiot has probably now condemned Jones to further ridicule, whilst thinking he was helping the cause.
DOH!
So, where do we complain about lack of balance in Mr Black's report?
Clearly, he is reporting the work of a partisan political lobbyist, which shouldn't be done without a balancing view.
Dead Dog Bounce
You can complain to the BBC, but only if you like wasting your time and banging your head against a brick wall. It is a fact well known within the BBC that the BBC is always right and can do no wrong.
http://www.bbc.co.uk/search/?q=BBC%20Complaints
I'm still wondering where the BBC science bias enquiry by Professor Steve Jones has got to. It has taken well over a year to gather information and was due to report in April. It appears the BBC would like us to forget all about it, while the enquirers, who are, of course, independent, appear unconcerned at the delay.
Jones and the BBC have to reproduce the calculations Jones did in order to convince. People should not simply take anyone's word, especially Jones with his track record. If they can't or are unwilling then it can be argued that both the BBC and Jones are attemping to mislead.
So Douglas, if I'm reading this right, even under the assumption recommended in the IPCC 2007 report for trend analysis - viz that the underlying process is AR1 - significance fails.
Given that (1) the AR1 assumption is quite untenable given the size of partial autocorrelations at lags greater than one interval, and (2) that any more realistic assumption, requiring differencing, substantially reduces the significance of the trend, this is desperate stuff indeed.
@Ross
Probably, he found only now a sympathetic journalist willing to listen.
Damned by faint raise. Applicable to both Jones & Black.
Is that the first R code pasted into a comment at Bishop Hill? That calls for some bubbly.
Anyhow, Nick Hallam summarises the position well. An infinitesimally small storm in a very ordinary green teacup - namely our state broadcaster armed with the latest press releases from climate consensus central. Yes, when is Steve Jones going to produce his findings?
Only a climate scientist could reach the top of a hill and state that this couldn't possibly be the top of the hill because he was previously going up.
@Dead Dog Bounce
I e-mailed both Richard Black and Roger Harrabin about the statistical error (including the R code and a link to the IPCC Appendix with the statistical method) at 17:30 Friday.
@Nicholas Hallam
Yes, that's it.
Mr Keenan
Thank you and please let us know about any responses you receive.
Tom Fuller says
"Hasn't he had enough grief?"
We are talking about someone who appears to have broken the law, behaved unprofessionally, seems to have been happy at the death of someone, appeared to have tried to pervert the peer review process, helped gang up on people who questioned his work, and misled the public about the science - he is still in his well paid job and still has his gold plated pension.
In essence my answer is NO!
Doug,
The object of the press release is to create the impression it creates, not to convey the information it contains.
By the time they 'respond' to the emails, if at all, the work will be done. My suggestion would be that this gets published quickly as a note in a journal because it deserves a wider public hearing.
Manchester United have been trying to sign Phil Jones. According to rumours, it's something to do with his creative skills and ability to drag positive results out of nothing.
HadCrut has more than 1 temperature product. If we look at the one Douglas Keenan analyzed, the trend isn't statistically significant even if we assume the residuals are white noise. If we look at http://www.cru.uea.ac.uk/cru/data/temperature/hadcrut3vgl.txt, the trend is not significant using the garden variety correction for AR1 commonly used in climate science. (i.e. correct the number of degrees of freedom using (1-R)/(1+R) where R is the lag 1 autocorrelation coefficient as done in the multi-author Santer paper on tropical tropospheric temperatures.)
Anyone can do either analysis with EXCEL. (I used R.)
I think Phil Jones needs to clarify and explain what data set he used and what method and assumptions he made. Because, right now, it looks like he may have committed a statistics and PR blunder.
Whether the trend since 1995 is statistically significant is not scientifically important: it's a short period. But if Jones can't explain the basis for his claim of statistical significance, the PR fallout for Jones is likely to be bad.
I don't have Richard Black's email so I submitted a comment at http://news.bbc.co.uk/newswatch/ifs/hi/newsid_4000000/newsid_4000500/4000569.stm.
Douglas, do the decent thing and pass on Richard's email to Lucia. And thanks, the both of you.
@lucia
The data set is specified in the BBC story: HadCRUT3. Indeed, that is the only plausible choice. Regarding the statistical method, yes Jones got even the IPCC method wrong—that was the point that I was making.
Hey ZedsDead [at Jun 10, 2011 at 4:19 PM] said:
"Strikes me, once again, that you are critical only of things which don't agree with you, and unquestioningly swallow whole, anything which can be made to seem to support you.
The exact opposite of scepticism."
We're not being paid to do science to justify turning the world economy on its head.
We are here for own amusement and our own edification, if possible.
One would think a sense of proportion and a recognition of the perverse entitlement going with AGW Orthodoxy would require a certain degree of latitude towards the unpaid skeptics.
But no, I see. The Paid are Privileged in your world for All Good Reasons Holy - not the UNpaid McIntyres, Bill & William Gray's, the Andrew Montforts, and myself - and many more.
Go piss off.
Yes. But I'm giving Jones the benefit of the doubt and suggesting he meant "HadCrut3V" or the BBC missed the "V" in their report. For this reason, I looked at both. Neither show statistically significant warming though. But if Jones used HadCRUT3, his blunder is extremely large because it doesn't show statistically signficant warming even if you treat the residuals as <I>white. Yes, I agree with you. And I used the simplified method that implements their statistical model can be applied using EXCEL. This method is only required if we use HadCRUT3V-- since HadCRUT3 doesn't even show significance if the residuals are white.
Several reasonable points in this thread but fundamentally Don Pablo (and one or two others) are offering the killer, namely that you can't go on re-analysing data using the simple p<0.05 (95% confidence, 2 standard deviation) criterion. There are other statistical tests that can be applied in that situation but they are tougher to arrive at a significant result.
JeffId applied the method of Quennouille to monthly HadCRut3V data and that does give Jones basis for claiming significance. The assumptions are Ar1 (as in the IPCC.) So this may be a matter of Jones being very unclear what he did. Maybe he'll clarify.
http://noconsensus.wordpress.com/2011/06/11/experts/#more-11732
a few highly numerate folks wonder what Mrs Jones's little boy is talking about
Lucia,
"Whether the trend since 1995 is statistically significant is not scientifically important: it's a short period. But if Jones can't explain the basis for his claim of statistical significance, the PR fallout for Jones is likely to be bad."
I think such a failure would be more significant than merely bad PR. It would reflect very badly on the quality and competence to do science for him and for his staff. Given that so much of the alarmist argument rests on an appeal to authority (specifically the authority of Jones, et al), it does a lot more damage than just a little bad PR. Of course, this wouldn't be his first step down that path. It could be argued that he's already done more than enough to inflict serious damage to his standing as a scientist.
stan--
Yes. This is going to further harm Jones reputation. But I think reputation management is an aspect of public relations. That's what I mean by PR fall out. This is going to be discussed for the next few months and people are going to be observing that Jones doesn't know how to compute statistical significance for warming.
Recipe for sceptic soup
Step 1) Pick a non-issue: the significance of warming since 1995
Step 2) Prompt Phil Jones to calculate that statistic
Step 3) Spread the news that Phil Jones says no significant warming since 1995, turning the non-issue into an issue.
Step 4) Watch as the media lead with headlines such as "no warming since 1995"
Step 5) Wait until warming since 1995 is now significant and Phil Jones announces such.
Step 6) Complain that Phil Jones is discussing a non-issue. Complain that he's discussing this and not something else. Complain that Phil Jones isn't being clear. Complain that Phil Jones is the devil incarnate. Complain that this is just going to harm Phil Jones' reputation (rather than the reputation of those who pushed the non-issue into an issue in the first place! ha!).
Step 7) Keep complaining and trying to find reasons to blame phil jones
That's quite an imagination you demonstrate, bob. Perhaps it would be beneficial for you to expend similar effort in researching how the events have actually unfolded.
Bob
Do you not understand that the warming is not significant and that therefore Jones is either incompetent or a liar?
Zed, Hengist - and now Cedric the Great and bob. Truly we are spoiled. Seriously, it is one measure of success. I hope you enjoy your visits.
Sorry to be a bit late to this exchange, but in case anyone's left in here, how about a nice graph?
HADCRUT3VGL, GISTEMP, UAH, RSS 1995 - present. Common 1981 - 2010 baseline; trend.
Decadal trend 1995 – present (degrees C):
HADCRUT3VGL 0.01 (0.0094)
GISTEMP 0.02 (0.0157)
UAH 0.01 (0.0127)
RSS 0.01 (0.0065)
Hardly the stuff of nightmares, at least not yet ;-)
Apologies: above is annual trend.
Decadal trend 1995 – present (degrees C):
HADCRUT3VGL 0.1 (0.0094 per annum)
GISTEMP 0.2 (0.0157)
UAH 0.1 (0.0127)
RSS 0.1 (0.0065)
.bob says:
Recipe for sceptic soup
Step 1) Pick a non-issue: the significance of warming since 1995....
True, bob, no empirical "divergence" has any significance to a pseudo-scientific Propaganda Operation.
Your "recipe" simply does not contain that ingredient.