Yes, Our Press Corps Is That Bad
February 15, 2010I’ve seen the claim that Phil Jones has “admitted” that there has been no climate change since 1995 in several places, but I was startled tonight when my mild-mannered father — a scientifically-inclined, but otherwise disinterested and innocent dentist in quasi-rural Virginia — raised the question with me. “What’s the deal with one of the supposed great figures in climate science saying that there is no more climate change?” he asked.
Sigh.
Evidently, many in the press don’t bother much to parse the language of statisticians, or less charitably, some interpreters of science are willing to go to extraordinary lengths to bend a statement to suit their purposes. Real Climate, I think, nicely disassembles the bullshit. Here’s a quick quote of their analysis:
What Jones actually said is that, while the globe has nominally warmed since 1995, it is difficult to establish the statistical significance of that warming given the short nature of the time interval (1995-present) involved. The warming trend consequently doesn’t quite achieve statistical significance. But it is extremely difficult to establish a statistically significant trend over a time interval as short as 15 years–a point we have made countless times at RealClimate. It is also worth noting that the CRU record indicates slightly less warming than other global temperature estimates such as the GISS record.
Amazing how handily some statements can be twisted.
Plainly, this should serve not as an indictment of Jones or climate science, but of the reporters of climate science, of the mouthpieces of disinformation and obfuscation. It is also a warning about accuracy and nuance. Jones was being mighty nuanced in that interview. Such nuance yields gaping holes for sophists to exploit and for most people to peer through.
As a dentist, my father is actually reasonably familiar with statistical significance, but the nuance of Jones’s comment was lost on him. He just didn’t want to bother to dig much deeper, so he let it slide and went with the press’s narrative. For reasons that escape me, it’s an easy narrative to swallow, particularly if one is already inclined that way.
LATE ADDITION (moved up from comments):
The question posed to Jones, as it happens, is a loaded (or complex) question. There have been many such questions in recent months.
Here’s a parallel:
“Do you agree that from January 2009 until January 2010 there has been no statistically-significant global warming?”
That’s a question with a damned if you do, damned if you don’t response. “Yes, I agree” is accurate, but it’s accurate only because the scale is too small and it gives a misleading impression about climate science. “No, I don’t agree” is inaccurate, because it is wrong about statistical significance, and it too gives a misleading impression about climate science. Even to say in blanket fashion: “Yes, but only just” or “Yes, but only because your statistical significance is harder to demonstrate across shorter timespans” which is pretty close to what Jones said, gives a misleading impression.
I think it’s too much to expect that someone can identify some of these fallacies on the fly. I actually traffic in fallacies on a daily basis, and it can be very hard for me to identify them and call them by name.
Cruel Mistress 
Funny how the “Press” has their shoe on the other foot now. The AGW’ers have been very used to a compliant press that has swallowed, hook-line-sinker, most everything said by AGW’ers for years, ignoring most arguements against (and no the “science is not settled”, nor should it be). Now the “press” is acting like sharks with fresh meat and you are surprised? No need to “chapter-verse” this. AGW “used” the press, perhaps for the greater good, but bottom line – the press doesn’t give a damn. I doubt any inaccuracies that were pro-AGW bothered you much…..
We don’t know each other, so I doubt your doubts about what worried me. Plus, I challenge your view that the “press,” scare-quoted or not, has been in the pocket of the climate scientists. That’s not congruent with my experience at all.
Compliant is my operative word for how the press has acted over the past decade to the AGW put forth by some. Now they are being more critical in their articles and perhaps investigating things a little more. My “hook-line-sinker” comment doesn’t translate to “in the pocket of climate scientists”, rather the press didn’t question enough. Now they are and as usual the pendulum might swing the other way for a while. That’s how things usually happen. You’re right, I don’t know you and shouldn’t presume to know what bothers you, but I would guess you’re a pretty decent guy….
Ben:
I think RealClimate is also spinning what Jones actually said. You probably don’t see it as clearly as I do because you are inclined towards their view.
Here is the actual question and answer from the BBC interview:
[BBC question:]
B – Do you agree that from 1995 to the present there has been no statistically-significant global warming
[Jones answer:]
Yes, but only just. 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. Achieving statistical significance in scientific terms is much more likely for longer periods, and much less likely for shorter periods.
So the answer was “Yes – but only just.” The rest of his answer is trying to diminish the importance of his yes answer to the question.
RealClimate’s version is more what they wish Jones had said, and how they spin what he said – not what he actually said. The most important part of Jones answer which they omit is the “YES” portion of Jones answer. They only leave in the diminishment portion of the answer.
I would have to say that your bias is showing a little when you uncritically accept RealClimate’s view as “nicely disassembles the bullshit” – when in fact they are piling it higher and higher.
That’s certainly possible. That’s what I appreciate about your comments, Rick, so thanks for the alternative perspective.
I actually read the BBC interview a day before I read Real Climate’s take on it, and I was pleased that Real Climate gave pretty much close to the same interpretation that I had. I think the “Yes, but only just” comment has to be taken in the context of a math and science geek answering questions as near as truthfully as possible.
This, as it happens, is a loaded (or complex) question. There’ve been a lot of them in recent months.
Here’s a parallel:
“Do you agree that from January 2009 until January 2010 there has been no statistically-significant global warming?”
That’s a question with a damned if you do, damned if you don’t response. “Yes, I agree” is accurate, but it’s accurate only because the scale is too small and it gives a misleading impression about climate science. “No, I don’t agree” is inaccurate, because it is wrong about statistical significance, and it too gives a misleading impression about climate science. Even to say in blanket fashion: “Yes, but only just” or “Yes, but only because your statistical significance is harder to demonstrate across shorter timespans” which is pretty close to what Jones said, gives a misleading impression.
I think it’s too much to expect that someone can identify some of these fallacies on the fly. I actually traffic in fallacies on a daily basis, and it can be very hard for me to identify them and call them by name.
“So the answer was “Yes – but only just.” The rest of his answer is trying to diminish the importance of his yes answer to the question.”
Well, that’s because it’s not particularly important. “but only just” clearly means he gets a confidence level just shy of 95%. 93%, 94% even 90% is a long way from saying “there’s no trend”. Nor does statistical significance guarantee that there’s a trend. WMO says to look at thirty years for trend significance, due to the intrinsic noisiness of the system.
The conventional use of the 95% confidence as a marker of statistical significance isn’t intrinsic to statistics. In other words, it doesn’t fall out of theoretical statistics, or any fundamental property of statistics.
Here’s a good discussion, by a professional statistician, of the history of the adoption of the 95% confidence level as a threshold for statistical significance.
It’s well worth a read.
It’s a convenient rule of thumb that’s been adopted for practical purposes. A trend that falls “just shy of”, but doesn’t meet that level of confidence is still very, very suggestive of there being a trend.
And … don’t forget that it’s just as likely that the trend is twice the observed value of +0.12C/decade as it is that there’s no trend at all …
Note that Jones is also saying that the trend WOULD be significant for starting values before 1995.
Ben:
What I find so interesting about the “no statistical significance” issue is how it has changed over the last few years.
The longer the flat or slightly cooling period lasts, the longer the “no statistical significance” period has become.
Of course I have my own set of biases – but it seems to me that we have gone from – 8 years is to short a period to now 15 years it to short a period.
30 years is the gold standard for trends to be statistically significantly (at least from my reading of RealClimate) – so if it should occur that at some point in the future the trend is flat or slightly cooling, over a 30 year period – that will be a game changer in the climate change debate.
Personally, I think this is quite likely to occur within the next 10 to 15 years.
My view is that the models are not very accurate and I await the passage of enough time to get decent data over about a 60 year period (to encompass both a warming and cooling period based on ocean currents) – which measured from 1978 (the start of satellite data) means we won’t have decent data to analyze until 2038.
All the current models seem to be analyzing trends over just the warming half of the cycle, and the longer we get into the cooling half – the more the models diverge from the trend.
So I think a lot of the fighting we have is just based on two short of an actual decent global temperature data set. At least that is the way it looks to me.
The other half of the fight is based on the validity of creating historical data sets based not on actual temperatures, but on proxies for actual temperatures.
So I say lets wait for actual decent temperature data over a long enough period to be statistically significant – which to my way of thinking is 2038 or later.
But is the scale of the question too small? A lot of people are trying hard to spin this but in the Phil Jones Q&A he discussed temperature for different periods. He broke down three warming periods in the instrumental temperature record.
1860-1880 21 years
1910-1940 31 years
1975-1998 24 years
By the standard some want to apply two of those three warming periods aren’t long enough either, including the late 20th century warming that some attributed to AGW. Somehow 24 years is enough to establish that AGW is taking place from 1975 to 1998 but 1995 to 2009 isn’t enough to establish no statistically significant warming?
I also think the idea of the “loaded question” is amusing. The BBC interviewed Jones in writing through the UEA press office and neither Jones nor the UEA press office objected. Jones was able to reply in detail with caveats to deal with the question.
A lot of people don’t want the question of no statistically-significant global warming since 1995 answered. I’ve even seen some blogs claiming some newspaper lied while showing the GISS temperature graph to rebut Phil Jones answer.
Ben your parallel question on a single year of global warming is hardly a parallel to a 15 year span.
The fact that a question is loaded doesn’t have anything to do with how much time a respondent has to reply. In many cases, no answer could adequately address the question.
And the parallel example on the single year of global warming is to illustrate the point about statistical significance (which was obviously Jones’s point), not to make a claim about the strength of 15 years of data over one year of data.
Ben – the point about the written Q&A which went through the UEA press office is that Jones could have refused to answer if he felt it was a loaded question. He didn’t answer a number a questions that were posed to him. In fact it’s a legitimate question and Jones answered with the appropriate caveats.
Where should the cutoff be in how many years is valid to look for a warming/cooling trend? As I pointed out the warming attributed to AGW is a span of 24 years while the question posed was for a period of 15 years. Is the cutoff somewhere in between? Why?
I understand your example but I disagree that it is in any way a parallel. We’ve been hearing for years that single years don’t make a trend in regard to no warming or cooling. That’s obvious, but in this case the question posed a period of 15 years and Jones answer with appropriate caveats was that there has not been statistically significant warming over that period.
How significant that is in the overall picture is a good question but it’s a bit silly to claim it was a loaded question.
By the way, the actual Q&A from Jones is here in case you haven’t seen it.
http://news.bbc.co.uk/2/hi/science/nature/8511670.stm
How Jones feels about whether it is a loaded question is also not at issue. The question is loaded, whether Jones feels it is or not. And, incidentally, refusing to answer just plays into its loadedness.
Jones may have answered with the appropriate caveats, but he is still taken out of context, and his meaning is parsed according to a specific agenda.
The cutoff should be established by appeal to the best explanation for the full set of data, and not a selected subset of data.
I read the full Q&A a few days ago.
We know the question was loaded because the interviewer (Harrabin?) stated that he was posing questions giving to him by “sceptics”.
Why 15 years? Why not 14 or 16, or 10 or 20? The question was loaded, probably not by accident.
The key point, as Phil noted, is that a trend that’s strongly significant over thirty years will, as it’s analyzed over shorter and shorter periods, become less signifcant and then, when it gets short enough, stop being significant. As Ben notes, by the time it gets down to a single year, even a very sharp change will not be significant.
All of that said, I hope Phil now realizes that the best thing to do in response to such a question is to not answer it other than to explain why it’s poorly framed.
[…] Mistress Being Human on a Harsh Planet « Yes, Our Press Corps Is That Bad False Concretism February 16, 2010 Among the many tropes circulating around the IPCC is […]
“A lot of people don’t want the question of no statistically-significant global warming since 1995 answered.”
It’s a meaningless question. Given that the WMO has adopted 30 years as a convention (decades ago), if we were to find a statistically significant trend in 15 years that would be a bit of a surprise.
30 years is too conservative, i.e. we have data on the noise in the system, so for any particular signal the number of years needed to show significance at a given level can be computed.
15 years doesn’t do it for HadCRUT. It does for GISTEMP. 16 years does it for both. The fact that adding a year, or using a different temperature product, makes the trend significant should be ignored?
It’s a loaded question because it “coincidently” happens to be the longest period of time for which the answer is “no statistical significance”.
This tool doesn’t give us significance information, but it shows that the trend 1994-present is greater than for 1995-present.
The higher trend combined with the longer time frame bumps significance up above the 95% confidence level. Just less than 95% confidence level, to just more than 95% confidence level isn’t, in the real world, particularly interesting.
Given that he took questions from the skeptic community, I think it’s clear why 1995 was chosen.
Ben –
I am adamantly against the warmist agenda, but I agree that the question posed is meaningless and has been spun to political advantage. However, you are missing the overwhelming context that surrounds the question; namely, that over this 15 year time frame we have been told ad nauseum that global temperature was not just rising, but that the rate of rise was accelerating. This was all clearly untrue. So, the data over the last 15 years is meaningless in regard to whether global warming is occuring or not or whether models are correct or not. But it is relevant in regard to whether the warming proponents tell the truth
The interesting factoid about warming since 1995 that everybody is ignoring is that it has stayed hot. When we had a heat wave where I live last year, it got about 5 degrees hotter per day for a number of days and then the peak heatwave daytime high was the same for two days.
Nobody said that the heat wave was over when that second peak day just stayed at 106 and didn’t continue the trend to 111. The heat wave was over when the temperature went back down to near what has been considered normal.