More from the 'Wrong Files' - Josh 224

The Met Office (along with everyone else, it seems) has much to say about Doug Keenan's post (simplified by our host here) and can be found on their blog here and website here. 

Cartoons by Josh

Click image for larger version

Met Office responds to Keenan

Doug Keenan's article about statistical significance in the temperature records seems to have had a response from the Met Office.

The text is here and there is a blog post here.

Met insignificance

This is an ultrasimplified version of Doug Keenan's post this morning.

The Met Office has consistently said that the temperature rise since 1850 is too large to have been caused by natural causes. Questioning from Lord Donoughue elicited the information that they came to this conclusion by modelling temperatures as a straight-line trend (global warming) plus some noise to represent normal short-term variability.

However, would a model in which temperatures went up and down at random on longer timescales, but without any long-term trend at all, be a better match for the real temperature data? Doug Keenan has come up with just such a "temperature line wiggling up and down at random" model and it is indeed a much better match to the data than the "gradual warming plus a bit of random variation" model used by the Met Office. In fact it is up to a thousand times better.

In essence then, the temperature data looks more like a line wiggling up and down at random than one that has an impetus towards higher temperatures.* That being the case, the rises in temperature over the last two centuries and over the last decades of the twentieth century, look like nothing untoward. The global warming signal has not been detected in the temperature records.

*Here I'm only referring to the two models assessed. This is not to say there isn't another model with impetus to higher temperatures which wouldn't be a better match than Doug's model. It's just nobody has put such a )third model forward yet. (H/T JK in the comments)

Met Office admits claims of significant temperature rise untenable

This is a guest post by Doug Keenan.

It has been widely claimed that the increase in global temperatures since the late 1800s is too large to be reasonably attributed to natural random variation. Moreover, that claim is arguably the biggest reason for concern about global warming. The basis for the claim has recently been discussed in the UK Parliament. It turns out that the claim has no basis, and scientists at the Met Office have been trying to cover that up.

The Parliamentary Question that started this was put by Lord Donoughue on 8 November 2012. The Question is as follows.

To ask Her Majesty’s Government … whether they consider a rise in global temperature of 0.8 degrees Celsius since 1880 to be significant. [HL3050]

Click to read more ...

No let-up for the Met Office

Doug Keenan writes:

A new session of parliament began yesterday, and already parliamentary questions about the statistical analyses of Chief Scientist Slingo have been tabled in both houses.

In the House of Lords:

Lord Donoughue to ask Her Majesty’s Government, further to the Written Answers by Baroness Verma on 14 January (WA 110), 5 February (WA 31–2), 21 March (WA 170–1), and by Lord Newby on 23 April (WA 359), whether they will give their numerical assessment of the probability in relation to global temperatures of a linear trend with first-order autoregressive noise, as used by the Met Office, compared with a driftless third-order autoregressive integrated model and ensure that that numerical assessment is published in the Official Report; and if not, why not. [HL62]

(Background posts include “Questions to ministers” and “Advisers advise politicians to look in the peer-reviewed literature”.)

In the House of Commons:

Graham Stringer (Blackley and Broughton): To ask the Secretary of State for Business, Innovation and Skills, with reference to the Answer of 15 April 2013, Official Report, column 261W, on climate change, what statistical models were used in any analyses done to calculate significances. [153909]

(Background posts include “Not answering the question” and “More from the Beddington FOI”.)

Whether to trust statistics

Betsey Stevenson and Justin Wolfers, writing at Bloomberg's website, consider some rules of thumb for helping the layman decide whether they should trust someone's statistical analysis or not. Here's the first of them:

Focus on how robust a finding is, meaning that different ways of looking at the evidence point to the same conclusion. Do the same patterns repeat in many data sets, in different countries, industries or eras? Are the findings fragile, changing as one makes small changes in how phenomena are measured, and do the results depend on whether particularly influential observations are included? Thanks to Moore’s Law of increasing computing power, it has never been easier or cheaper to assess, test and retest an interesting finding. If the author hasn’t made a convincing case, then don’t be convinced.

It's hard not to recall the case of the Hockey Stick and its reliance on the bristlecones. And all the other paleoclimate studies that are said to support the Mannian stick, and which rely on bristlecones too.

Climate Dialogue on long-term persistence

The Dutch site Climate Dialogue has launched a debate on long-term persistence in climate records. It features Rasmus Benestad, Demetris Koutsoyiannis and Armin Bunde.

This could be interesting.

Data in the Raw - Josh 217

With several questions from MPs recently, see here, here and here,  on the statistical analysis supporting the Met Office's claims about recent warming, it is probably time for the Met Office to do some revealing of evidence. Julia Slingo holds up the relevant papers on the subject.

Cartoons by Josh

Update on Apr 24, 2013 by Josh

H/t Anthony for the 'Gut instinct' quote and story here.

More Met Office and statistics

There is more evidence of parliamentarians' attention being focused upon the Met Office's statements climate change. This question to ministers in the Department of Business Information and Skills (the department that sponsors the Met office) has been tabled for a reply next week. 

Graham Stringer (Blackley and Broughton): To ask the Secretary of State for Business, Innovation and Skills, whether the claim that (a) every year since 1998 has been significantly warmer than the temperatures you would expect if there was no warming and (b) for the last three decades the rate of temperature increase is significant made by the Met Office in a climate science briefing sent to the Chief Scientific Adviser on 8 February 2010 was supported by any statistical time-series analysis.

The document referred is discussed and can be download here.

Questions to ministers

Updated on Apr 9, 2013 by Bishop Hill

This is a guest post by Doug Keenan

Questions relating to the work of the Met Office on global warming are being put in the UK parliament, and the Met Office is refusing to answer them. Parliamentary Questions have a history going back centuries. Giving answers, or giving a valid reason for not answering, is required. The stand-off is yet to be resolved.

The Parliamentary Question that started this was put by Lord Donoughue on 8 November 2012. The Question is as follows.

To ask Her Majesty’s Government … whether they consider a rise in global temperature of 0.8 degrees Celsius since 1880 to be significant. [HL3050]

Click to read more ...

Briggs on statistics

Wonderful lecture by Matt Briggs on the absurdities of statistics as applied in the social sciences.

And climatology.

IPCC statistics ruled illegal

Bayesian statistics, the approach favoured by the IPCC in its assessments of the world's climate, has been ruled illegal by the Appeal Court in London. As the judge explained in a case revolving around possible causes of a fire:

Sometimes the "balance of probability" standard is expressed mathematically as "50 + % probability", but this can carry with it a danger of pseudo-mathematics, as the argument in this case demonstrated. When judging whether a case for believing that an event was caused in a particular way is stronger that the case for not so believing, the process is not scientific (although it may obviously include evaluation of scientific evidence) and to express the probability of some event having happened in percentage terms is illusory.

David Spiegelhalter notes that "[to] assign probabilities to events that have already occurred, but where we are ignorant of the result, forms the basis for the Bayesian view of probability". That being the case, one wonders whether this opens up the possibility of legal challenges to the IPCC assessment reports.

For once, however, I find myself on the IPCC's side. I imagine a higher court will set the ruling aside.

Batting back at Beenstock

The Beenstock, Reingewertz, and Paldor paper that did the rounds before Christmas applied some whizzy statistical methods to the temperature and forcing trends and found that they were statistically independent of each other. There's a useful discussion of it here at David Stockwell's site. As Stockwell notes, this is not a debunking of AGW, but rather of the use of linear regression to "demonstrate" that something unusual is happening in the temperature records.

To that extent, Beenstock's paper should be entirely uncontroversial. I'm not sure that anyone really thinks linear regression is a suitable approach to apply to temperature records. Nevertheless there has been a rapid rebuttal posted to the journal in the shape of an article by two Oxford academics, D. F. Hendry and F. Pretis:

We demonstrate major flaws in the statistical analysis of Beenstock et al. (2012), discrediting their initial claims as to the different degrees of integrability of CO2 and temperature.

A new hockey team paper

There are lots of familiar names behind this new paper in JAMS - Rutherford, Mann, Wahl and Ammann. It seems that their infilling RegEm methodology has received some criticism. Apparently though, it "doesn't matter".

Smerdon et al., (2010) report two errors in the climate model grid data used in previous pseudoproxy-based climate reconstruction experiments that do not impact the main conclusions of those works (Mann et al., 2005; 2007a). The errors did not occur in subsequent works (Mann et al., 2009, Rutherford et al., 2010, and Schmidt et al., 2011) and therefore have no impact on the results presented therein. Results presented here for the CSM model using multiple pseudoproxy noise realizations show that the quantitative differences between the incorect and corrected results are within the expected variability of the noise realizations. It should also be made clear that the climate reconstruction method used in Smerdon et al. (2010) to illustrate the nature of the errors, RegEM-Ridge, is known to produce climate reconstructions with considerable variance loss and has been superseded by RegEM-TTLS in Mann et al. (2007) and subsequent works.

Uniform priors and the IPCC

Last week, I posted about a comment Nic Lewis had written at RealClimate. In that comment, Lewis had spent some time discussing a study by Aldrin et al, and noted that its findings were distorted by the use of a uniform (or "flat" prior). Although Gavin Schmidt did not respond directly to this point, one commenter pushed the question of the validity of the uniform prior approach a little further.

Graeme:

I thought James Annan had demonstrated that using a uniform prior was bad practise. That would tend to spread the tails of the distribution such that the mean is higher than the other measures of central tendency. So is it justified in this paper?

This elicited a response from a statistician called Steve Jewson (a glance at whose website suggests he is just the man you'd want to give you advice in this area):

Click to read more ...