Tim Lambert (deltoidblog AT gmail.com) is a computer scientist at the University of New South Wales.
The IPA is the Australian version of the CEI, so you don't have to read an article they publish on global warming to know what the conclusions will be. But you do have to read it to find out what pretext will be used to dismiss concerns about warming. In the latest issue of IPA review we find an article by two economists (Sinclair Davidson and Alex Robson) that attempts to spread confusion about the IPCC fourth assessment report.
The article is not online, but most of it is available here.
They start off by taking a leaf from Michael Crichton's book -- they change the vertical scale on the graph of global temperatures by a factor of 20 to make the recent temperature increase look smaller. I used the same technique on the temperature data from the Vostok ice core and ice ages turn into tiny little bumps.
Why do those silly scientists think the climate was different during the last ice age? Using the Crichton/Davidson/Robson method the temperature is not noticeably different.
Damien Eldridge also finds their argument unconvincing.
The second part of Davidson and Robson's article is a bizarre claim that the "IPCC pulls numbers out of thin air":
There are legitimate difficulties with the IPCC's 90 per cent confidence in anthropogenic warming. It is not ludicrous to question what that number means. The IPCC seems to imply that this number results from a scientific process -that it has tested a hypothesis. Indeed, the IPCC tells us its understanding is based "upon large amounts of new and more comprehensive data, more sophisticated analysis of data, improvements in understanding of processes and their simulation in models, and more extensive exploration of uncertainty ranges". If this is what the IPCC has done, it has very weak evidence. Ninety per cent is the weakest acceptable level of confidence in a hypothesis test. It is not clear from the Summary whether the IPCC has, in fact, undertaken such an analysis. It is more likely that it has neither a testable model nor data available for external researchers to replicate such a test. In other words, the IPCC's 90 per cent confidence has emerged from scientists evaluating whether they think their own work is correct.
But look at what the SPM says
This Assessment considers longer and improved records, an expanded range of observations, and improvements in the simulation of many aspects of climate and its variability based on studies since the TAR. It also considers the results of new attribution studies that have evaluated whether observed changes are quantitatively consistent with the expected response to external forcings and inconsistent with alternative physically plausible explanations.
Most of the observed increase in globally averaged temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations12. This is an advance since the TAR's conclusion that "most of the observed warming over the last 50 years is likely to have been due to the increase in greenhouse gas concentrations". Discernible human influences now extend to other aspects of climate, including ocean warming, continental-average temperatures, temperature extremes and wind patterns (see Figure SPM-4 and Table SPM-1). {9.4, 9.5}
FIGURE SPM-4. Comparison of observed continental- and global-scale changes in surface temperature with results simulated by climate models using natural and anthropogenic forcings. Decadal averages of observations are shown for the period 1906-2005 (black line) plotted against the centre of the decade and relative to the corresponding average for 1901-1950. Lines are dashed where spatial coverage is less than 50%. Blue shaded bands show the 5-95% range for 19 simulations from 5 climate models using only the natural forcings due to solar activity and volcanoes. Red shaded bands show the 5-95% range for 58 simulations from 14 climate models using both natural and anthropogenic forcings. {FAQ 9.2, Figure 1}
Despite the SPM including a graph that fills an entire page showing the results of tests of models, Davidson and Robson claim that it is likely that the IPCC doesn't have a testable model. Davidson and Robson are taking denial to another level. They have also confused the IPCC's more than 90% probability (what is the probability that most of recent warming is man-made?) with 90% confidence on a hypothesis test (what is the probability that a model with only natural forcings could produce half as much warming as observed?). The hypothesis tests that the IPCC refers to achieved levels of confidence greater than 95%.
It gets worse. Davidson and Robson also claim:
As an elementary textbook of statistics reminds its readers, with large data sets, confidence intervals have to be increased, so a 90 per cent confidence level would not then be valid -- the hypothesis is falsified.
No, elementary textbooks of statistics do not say that. Mainly because it's not true. A larger data set means that smaller effects will be statistically significant, so it is possibly that an effect could be statistically significant but so small that it is not practically significant. Furthermore, if a test does not achieve a specified confidence level, it does not falsify the hypothesis -- all you can say is that there is insufficient data to draw a conclusion.
Sinclair Davidson is a Professor in the School of Economics, Finance and Marketing at RMIT. Alex Robson is a Lecturer in Economics at ANU. It is worrying that economists at Australia's universities have such a poor understanding of basic statistics.
Email this entry to a friend
View the Technorati Link Cosmos for this entry
Comments
Rabett Run has a take on yet ANOTHER front group. This one is going to play the "well, it's a problem, but we can't do anything about" song and dance.
It's run by Joe D'Aleo, apparently from his house, judged by the registration of the website.
Posted by: Thom | April 12, 2007 2:25 PM
Posted by: Meyrick Kirby | April 12, 2007 2:26 PM
If they are changing the graph scale to run from 0 to 1000K, does that mean they are expect to have to accomodate a big spike in temperature to something like say 800K! This would easily fit on the graph, although the paper it is printed on might catch fire.
Posted by: bigTom | April 12, 2007 2:44 PM
Australian climate-change deniers are getting an assist from the Roman Catholic cardinal who heads the archdiocese of Sydney. Cardinal Pell thinks Gore and others are scaremongers.
Posted by: Zeno | April 12, 2007 5:51 PM
Well yes but the Cardinal also thinks secular democracy is a bad idea.
Posted by: Ian Gould | April 12, 2007 6:12 PM
According to their own publication in 2004, "the IPCC does not conduct new research, monitor climate-related data or recommend policies."
The argument in our IPA Review article is based on what the IPCC says about itself and its own role. They admit that they do not collect data, and that they do not conduct any research or hypothesis tests.
So the basic question we raise still remains unanswered: if they are not monitoring data or conducting new research, where does their 90% figure come from?
Posted by: Alex Robson | April 12, 2007 8:05 PM
This is so sad and frustrating. Whoever is running the IPA should be fired immediately.
I've tried discussing this with Jennifer Marohasy on her blog, but she isn't interested.
The IPA refuses to move past this lame science denial. They put all their resources into challenging the science. This means that their articles discussing the actual means of mitigating climate change are few and far between - no more advanced than the odd alan moran article saying 'every climate policy will hurt the economy, though nuclear power is best' kind of stuff.
The political climate in Australia wrt climate change has progressed markedly in the last year. And now, Australia is in desparate need of some credible, intelligent and informed conservative thinkers and opinion writers to participate in public debate about policy creation. They need to catch up.
In the last few months, the Australian govt has banned incandescent lightbulbs, and the opposition has committed another idle $50mil to photovoltaic power. There is also the spread of energy efficient housing regulation, which though commendable, is to an extent poorly targeted and less than effective - it could use a strong and intelligent critique, with suggestions for alternative approaches.
We need some conservative writers to be critiquing this kind of populist, heavy handed, yet ineffectual fluff policies.
Banning incadescent lightbulbs is a peripheral policy that targets the emissions of individuals at the exclusion of the emissions of the big emitters - power stations. And it curbs individual liberty - a nanny govt policy. (in any case, everyone will switch to halogens, which are worse).
Its politically acceptable to do that, to target green policies at households, but it isn't effective or efficient.
While the IPA is hung up on arguing the science, they have no legitimate voice in participating in discussion of new policy measures to mitigate emissions.
The least bureaucratic, heavy handed means of mitigating climate change that is widely regarded as one of the least cost approaches, is emissions trading. the IPA should be slamming all these inefficient policy measures like banning light bulbs and promote the policy measures that sit best with their free-market pro-individual ethos.
But instead, they keep spinning this anti-science crap, and painting themselves as mining industy shills and extremist anti-science goons. They need to grow up and get new leadership.
Posted by: Steve | April 12, 2007 8:36 PM
That's called a Type III error. Correct answer, wrong question. It's well known that as n tends to infinity the probability of spurious rejection of the null increases. This problem even has a name and a solution. "Lindley's paradox: tiny errors in the null hypothesis are magnified when large data sets are analyzed, leading to false but highly statistically significant results." The point is though, we don't know how data the IPCC have because they don't reveal how they come to their 90% estimate.
Posted by: Sinclair Davidson | April 12, 2007 9:46 PM
Lambert: "The hypothesis tests that the IPCC refers to achieved levels of confidence greater than 95%". I have searched TL's source, the SFP
Instead, we find that "there is high confidence (i.e. 80%) that the rate of sea level rise increased from the 19th C to the 20th C". First, as there is no evidence on what the RATE of sea level rise was before 1900, how can the IPPC so "highly" confident that the RATE has increased since 1900? Second, "the total 20th C rise is [highly confidently] estimated to be 0.17 (0.12 to 0.22) m". When the range is 30% either way, the claimed level of confidence is close to meaningless. If your surgeon was highly confident that a course of aspirins would raise your life expectancy by 2 weeks, +- 30%, it could be time to change to another. However I will cite the IPCC to my Mebourne Cup bookie to get him to adjust my odds by 30% either way.
Posted by: Tim Curtin | April 13, 2007 4:12 AM
IPCC: Red shaded bands show the 5-95% [i.e. up to] range for 58 simulations from 14 climate models using both natural and anthropogenic forcings. {FAQ 9.2, Figure 1)
Time Lambert The hypothesis tests that the IPCC refers to achieved levels of confidence greater than 95%.
Posted by: Tim Curtin | April 13, 2007 4:22 AM
Sinclair, you claimed that this was in any elementary stats book. It isn't. Why don't you provide a cite to an elementary stats book that tells you that you have to increase the significance level with large samples? I agree with Chris from Mixing Memory on Lindley's paradox -- it's not a problem in practice.
In any event it's only a problem with frequentist statistics, and the AR4 report cited plenty of Bayesian attribution studies as well. For instance:
Posted by: Tim Lambert | April 13, 2007 5:53 AM
Steve -
This is what I see as one of the worst effects of AGW denialism - it has effectively meant that all suggested policy solutions have come from what I'd call the environmentalist-left.
Unfortunately, this grouping also tend to be highly suspicious of large industrial enterprises (which is fair enough in some ways..) and therefore favours small scale and/or 'natural' solutions. Hence Wind, Biofuels, Efficiency, and at the most extreme Relocalisation. All of which sound great on paper, lead to very nice pilot projects but stand no chance whatsoever of scaling up to the degree required to actually make a difference.
(And if you don't believe me on the above, feel free to show me an instance where any of the above have put a noticeable dent in the CO2 emissions of a country.)
So denialism of global warming has had the (unintended?) side effect of ensuring that reduction/mitigation policies are largely ineffective.
Posted by: Andrew Dodds | April 13, 2007 6:23 AM
Tim, every stats book should show the confidence bound calculation as being a function of the standard error which is the standard deviation divided by the square root of the number of observations. As the number of obs gets big so the square root get big etc. I am shocked to hear this isn't a problem in practice. In finance data we often have large data sets and this is a problem. The IPCC claim to have a large data set, and if they do then Lindley's paradox might be a problem. Now do they have a lerge data set or don't they? If they do then Lindleys paradox is a problem, if they don't and Lindley isn't a problem then they don't have a large sample. Who do I believe? The IPCC or you? How is the 90% value derived? If they were transparent as they are required to be, then we wouldn't be having this debate.
Posted by: Sinclair Davidson | April 13, 2007 7:31 AM
"We need some conservative writers to be critiquing this kind of populist, heavy handed, yet ineffectual fluff policies."
Good luck with that. It's probably close to a decade ago that I tried to convince my then-employer the Queensland EPA to shift the subside on hot water systems from the end-buyer to the distributor or manufacturer. This would have been more cost-effective and also probably have resulted in a higher uptake. (I'm not going to argeu the benefits of subsides here, just making the point that a $500 drop in the wholesale price will probably lea to a more than $500 drop in the retail price and therefore to more consumption of the subsidised good.)
We quickly got the word from the political level to drop the idea - what mattered was solely those government cheques showing up in voter's mailboxes.
"Banning incadescent lightbulbs is a peripheral policy that targets the emissions of individuals at the exclusion of the emissions of the big emitters - power stations. And it curbs individual liberty - a nanny govt policy. (in any case, everyone will switch to halogens, which are worse)."
Umm Steve if individuals electricity consumption goes done, what do you think will happen to emissions from power stations?
As for people shifting to halogens, all I can say is that from personal observation compact fluros have virtually disappeared from shop shelves around Brisbane since the ban plan was announced. Somebody's buying them.
Posted by: iangoud | April 13, 2007 8:48 AM
"Banning incadescent lightbulbs is a peripheral policy that targets the emissions of individuals at the exclusion of the emissions of the big emitters - power stations. And it curbs individual liberty - a nanny govt policy. (in any case, everyone will switch to halogens, which are worse)."
How I wish that were so, I supply a vital component for halogens and not for incandescents or CFLs. However, from within the light bulb industry that's not what seems to be happening. Before the Australian and EU bans were announced (although after Venezuela and Cuba's) metal halide bulb production was roughly static at some 100 million a year while CFL production this year has ramped up to 1 billion (that's a minimum).
The bans will have very little efect: for the markets were already making the switchover. But at least it will give the politicians something to crow about, even if they didn't actually cause the change.
Posted by: Tim Worstall | April 13, 2007 8:50 AM
Dear Sinclair: What the hell? Large samples give more power, better stability for non-zero effects, and better stability to estimate effect sizes. Consider the IPCC process as a meta-analysis, across measures, methods and sites. Notice how the estimated effects for warming have not changed at the means since the last report, but the confidence intervals have decreased. in any case, it's the effect size that's important. We're not talking about a statistically significant, meaningless change of .03 degrees - we're looking at several degrees, which those climatology folk seem to get exercised about. I'm not a climatologist, but I'd take scientists in their own profession at their word, without strong evidence to the contrary. You wouldn't tolerate that sort of statistical bait and switch nonsense from a left-leaning undergrad, being Conservative doesn't give you a free pass on it either.
Posted by: Stewart | April 13, 2007 8:59 AM
Stewart - I don't disagree with your sentiment. I had expected them to have done a meta-analysis. (Although it turns out the IPCC claim not to do any sort of analysis.) But it's not clear that that has happened. It's not clear what has happened. Now if any student had written that piece I would say "where does the number come from? make it obvious". It is not obvious. Tim keeps on saying I don't understand stasitics. Okay fine. Whatever. But if I don't understand what the IPCC are doing in their Summary for Policymakers, why do we expect policymakers to understand what they have done? The IPCC have a mandate to be open and transparent. It is not too much to hold them to that mandate.
Posted by: Sinclair Davidson | April 13, 2007 9:43 AM
Zeno,
I read Cardinal Pell's article. Here's the best bit:
"We know that enormous climate changes have occurred in world history, e.g. the Ice Ages and Noah's flood,"
Posted by: Tony Jackson | April 13, 2007 10:09 AM
Sinclair, your claim was not that stats books say that large samples give you smaller confidence intervals, but this:
Please provide a cite to an elementary textbook of statistics that states this. Surely this cannot be too hard for you?
Sinclair, in their SUMMARY for Policy Makers they are SUMMARIZING their assessment of the research. It is a SUMMARY. Hence the title. The numbers in the SUMMARY refer you to sections in the full report which go into all the details of the research that they ASSESSED in the 4th ASSESSMENT Report. I know this will bounce off your fact shield.
Posted by: Tim Lambert | April 13, 2007 10:18 AM
alright, so it seems that:
a. global warming is a scientific theory largely supported by the research conducted with the resources and information available to the scientific community at this time
b. many measures taken/proposed to curb global warming have met extensive-and often justified- criticism because they are ineffective
so, assuming global warming to be a valid theory (frankly, if I received a telephone call saying that fourteen thousand people were about to be shot and that I could save them but only if I gave up my boat I'd rather risk losing the boat for nothing than letting 14 thousand people die for no reason- maybe that's just me, in which case debating the legitimacy of global warming is perfectly constructive)what SHOULD be done? I understand the criticisms of many of the suggested measures, but this is a problem that can be largely controlled, so does anyone have any constructive suggestions?
Posted by: Coleen | April 13, 2007 1:14 PM
alright, so it seems that:
a. global warming is a scientific theory largely supported by the research conducted with the resources and information available to the scientific community at this time
b. many measures taken/proposed to curb global warming have met extensive-and often justified- criticism because they are ineffective
so, assuming global warming to be a valid theory (frankly, if I received a telephone call saying that fourteen thousand people were about to be shot and that I could save them but only if I gave up my boat I'd rather risk losing the boat for nothing than letting 14 thousand people die for no reason- maybe that's just me, in which case debating the legitimacy of global warming is perfectly constructive)what SHOULD be done? I understand the criticisms of many of the suggested measures, but this is a problem that can be largely controlled, so does anyone have any constructive suggestions?
Posted by: Coleen | April 13, 2007 1:16 PM
"IPCC: Red shaded bands show the 5-95% [i.e. up to] range for 58 simulations from 14 climate models using both natural and anthropogenic forcings. {FAQ 9.2, Figure 1)
Time Lambert The hypothesis tests that the IPCC refers to achieved levels of confidence greater than 95%."
Ouch, that's a hardcore misunderstanding of statistics.
The 5-95% refers to the range in which anthropogenically caused warming is 90% likely to occur. I.e. there's a 90% chance that anthropogenic forcing is responsible for X amount of warming. This is the bright red band in the figure.
A 95% confidence interval is for all intents and purposes the probability the observed data is due to anthropogenic forcing. I.e. there's a 95% chance that the black line is inside the red band because anthropogenic forcing leads to warming (and therefore a 5% chance that that black line is in there due to random chance).*
*More accurately, a confidence interval refers to the probability of the null hypothesis given the data.
Posted by: LogicallySpeaking | April 13, 2007 3:24 PM
""We know that enormous climate changes have occurred in world history, e.g. the Ice Ages and Noah's flood,""
I knew the Cardinal was in an advanced state of decrepitude but I didn't realise he regarded the ice ages as within the scope of recorded history.
Posted by: Ian Gould | April 13, 2007 5:44 PM
You'll appreciate I haven't checked each of these books myself, but I suspect all of them will have section on hypothesis testing. Most chapters include a discussion of confidence levels, Type I and Type II errors and the like.
But Tim, you digress. Where does the 90% figure come from?
Posted by: Sinclair Davidson | April 13, 2007 7:34 PM
I obviously haven't read AR4, and it appears neither have you (you only seem to be aware of the summary for policymakers). That said, I'm not even sure if the full document has been released yet.
Nonetheless, I'd assume that the 90% is derived from peer-reviewed publications. Something along the lines of 90% of climate models require anthropogenic forcing to within a certain confidence interval. Given the amount of skepticism present on the IPCC panel, I'd assume it's a rather high confidence interval, and not just 95%.
Posted by: LogicallySpeaking | April 13, 2007 7:52 PM
Cardinal Pell isn't worried about global warming because he knows that Heaven is a very hot place, in fact, Heaven is hotter than Hell, as can be deduced from Isaiah 30:26 and Revelations 21:8.
Posted by: Joe | April 13, 2007 9:38 PM
We're talking about the SPM - that has been clear from the beginning. The AR4 report is being released in stages. I am only able to read documents in the public domain.
The operative word here is "assume". The IPCC are required to be open and transparent. We should not have to assume anything. They have reported a number (indeed several numbers) and all I'm asking for is an explanation of where that number came from. (Of course, four of the numbers have no human attribution study at all - so where did those numbers come from, given the IPCC admit they don't come from peer-reviewed publications, or indeed any publication).
Posted by: Sinclair Davidson | April 13, 2007 9:56 PM
Sinclair: I'm 90% certain it was a show of hands
Posted by: chrisl | April 13, 2007 11:28 PM
"We're talking about the SPM - that has been clear from the beginning. The AR4 report is being released in stages. I am only able to read documents in the public domain."
Fair enough, but you can't criticize them for not including all of the details in a summary. It would be equivalent to being a handed the abstract of a paper, told that the full paper will be viewable in a few months, and then proceeding to critique the abstract for not being open and transparent.
"The IPCC are required to be open and transparent. We should not have to assume anything. They have reported a number (indeed several numbers) and all I'm asking for is an explanation of where that number came from."
I agree with you here. Could you give us a general idea of whom you attempted to contact that is affiliated with the IPCC that was unwilling to answer your question(s)? Or what their response was when you posed this question of them?
Posted by: LogicallySpeaking | April 14, 2007 12:34 AM
I am unable to contact the IPCC directly because I don't have ethics approval to collect data from human subjects. I was hoping there documentation would be open and transparent enough to provide an answer to the question. Afterall, that 90% figure is the crux of the whole issue.
Posted by: Sinclair Davidson | April 14, 2007 1:49 AM
"I read Cardinal Pell's article. Here's the best bit:
"We know that enormous climate changes have occurred in world history, e.g. the Ice Ages and Noah's flood,""
But the ice ages didn't really happen because the world was created in c. 4000 BC (along with all the evidence to trick humans into believing that the earth is much older). Hasn't Pell been keeping up with his Bible studies?
Posted by: Chris O'Neill | April 14, 2007 3:03 AM
Logically speaking: You are right, I was merely trying to find TL's source in the SPM. His pretty figs at the beginning of this thread are all very fine, but the red and blue bands are clearly stated to be the outcomes of modelling using various assumptions about solar effects, volcanoes, and CO2. Where are the data inputs for these assumptions and statistical analysis thereof (preferably in Excel format complete with cointegration, ts, ps, etc)? Counting up and then averaging computer model simulations is NOT a valid test of an hypothesis.
Posted by: Tim Curtin | April 14, 2007 3:55 AM
Sinclair, you didn't say that elementary stats book discuss hypothesis testing but that:
If your statement is true, it would be easy for you to provide just one example. Why can't you do this? Is it because you claim is a fabrication?
Posted by: Tim Lambert | April 14, 2007 4:46 AM
Where did the IPCC get their numbers from?
Posted by: Sinclair Davidson | April 14, 2007 5:05 AM
It's explained in the full report, Sinclair.
If your statement about stats texts is true, it would be easy for you to provide just one example. Why can't you do this? Is it because your claim is a fabrication?
Posted by: Tim Lambert | April 14, 2007 5:26 AM
Fabrication in what sense?
Posted by: Sinclair Davidson | April 14, 2007 7:29 AM
TL & SD: One recent standard text (Charles Feinstein and Mark Thomas, the late Charles was quondam editor of the EJ) states that "any confidence interval is directly affected by the size of the sample on which it is based" (p.140)
Posted by: Tim Curtin | April 14, 2007 8:00 AM
Um, where do get the idea that they are assumptions?
Posted by: Abe G | April 14, 2007 8:45 AM
Abe G: if they are not assumptions, why do you need so many models (50+)?
Posted by: Tim Curtin | April 14, 2007 9:15 AM
Tim C,
No, you said they made assumptions about volcanoes, solar and CO2. Furhter, you said this was clearly stated. Where is it clearly stated?
Posted by: Abe G. | April 14, 2007 9:23 AM
Abe: The pics at the beginning of this thread state that the bands show the results of about 76 different simulations. If the data did not change, why so many simulations? So am I right that what we have is about 76 alternative theories, none of which has been tested against the data? One million computer models with the same data but differing theories do not aggregate into a test of the nul hypothesis that anthropogenic effects are innocent of causing climate change. What is your hypothesis? what are your data for testing my nul hypothesis?
Posted by: Tim Curtin | April 14, 2007 10:06 AM
Sinclair: fabrication in the sense that elementary texts do not say that "with large data sets, confidence intervals have to be increased" but rather you and Alex Robson made that part up. It is trivial to prove that you didn't -- just cite one elementary text that says it. How about it?
Posted by: Tim Lambert | April 14, 2007 10:21 AM
What part of "wait for the paper" do you not understand???
Posted by: David Marjanović | April 14, 2007 10:48 AM
So what's the effect size? Are you saying GW is real but trivial and makes no difference, or are you saying you don't care, you just found a handy stick that doesn't actually fit the question?
I see that the various National Academies and other learned bodies all take this very seriously. Don't all of these folks have to do something with statistics at times? Maybe the EC and all these first world academics just have deep-seated Stalinist tendencies?
Sinclair, this is just another variation on 'climate scientists are fools', dressed up with a little more statistical jargon. Now. I understand that economists and string theorists are fighting it out to see who is the smartest, and all other disciplines are populated by fools, but they put in the time and effort to learn their discipline. It's rude to say they are fools without clarifying how.
Posted by: stewart | April 14, 2007 11:38 AM
That's a very agressive stance you're taking Tim. You're leaving no wiggle room for yourself. I'm wondering what you're going to do when I name a book?
I'll give you a clue. While some hardcover versions did come out, it's a big fat blue paperback with a spine that cracks easily.
Posted by: Sinclair Davidson | April 14, 2007 5:58 PM
Stewart, this is about accountability. This is about irritation about climate change hype. The SPM is actually more nuanced than people think and despite drawing on a lot of work the limits of what is known are quickly reached. A number of 90% gets thrown out with no context of meaning. Human attribution numbers get thrown out with the admition there is no human attribution study. They have no brief to manufacture those numbers and they have no brief to hype up their results. As Alex and I say nobody is calling for the environment to be despoiled, and as I have written elsewhere there may be good arguments for a carbon tax indepedent of any climate change. The problem Tim is having is that he has set Alex and I up sterotypical denialists (in his own image of such people). So Tim and I have had this debate here and at Catallaxy for the past week (and an earlier exchange last month) and you'll notice he keeps changing the subject - so right now he wants me to name a textbook in my office bookcase. Rather than concentrate on the issue - how did the IPCC derive their headline number? As David suggests it might be in the main reports (or one of them, at least). This may all be a poor presentation problem and then Alex and my argument will all be a non-issue.
Posted by: Sinclair Davidson | April 14, 2007 6:18 PM
"Umm Steve if individuals electricity consumption goes done, what do you think will happen to emissions from power stations?"
No need to be condescending, I'm not an idiot.
WHen you consider that electricity is about 34% of australia's emissions, residential electricity is maybe 30% of that, and lighting is <10% of residential electricity, and many people are already switching to CFLs or to halogens, and CFLs are not zero power consumption, then multiplying all those fractions out gives a policy that has small potential for GHG reductions, big impact on the general public, and big media posturing. I'd hate to be the low level policy officer fielding phone calls from fridge light bulb manufacturers, people who get headaches from fluro lights, microwave lightbulb manufacturers, torch lightbulb manufacturers, xmas tree light manufacturers, lava light manufacturers etc who don't know WTF to do and how this new lightbulb policy will affect them.
Where was the conservative criticism? Where were the conservatives saying 'let the market decided how to reduce emissions and stop this proliferation of ad-hoc policies aimed at the punters'? They were busy with the climate change denial.
A sensible and conservative IPA policy is not climate denial. It is acceptance of the consensus, and campaigning for long term targets, steady progress, simple and broad policy responses, and anti lotsa bitsy pricey policies supporting light bulb bans and solar power etc. The IPA should be pro- broadly applied emissions trading and anti every other climate mitigation policy.
I'm not conservative, but i do flinch when i hear some of the inane or even ineffectual policy suggestions coming from some corners of the green-left side of politics. We need a constructive and credible yet conservative voice to balance this. The IPA is currently not credible with their denial message.
Posted by: Steve | April 14, 2007 10:14 PM
TimW, when I talk about halogens, i mean the halogen downlights that are currently the norm in most well to do urban homes (3,4, 5+ bulbs per room), not metal halide floodlight kind of gear.
Posted by: Steve | April 14, 2007 10:17 PM
Sinclair, the normal practice when citing a book is to give the author and title. I'm afraid that "big fat blue paperback" isn't a very useful way of citing.
I did not cast you as a stereotypical denialist. Didn't you notice the title of this post? "A new flavour of Global Warming denial". You are doing soemthing different -- denying the existance of the attribution studies the IPCC cites. Even though I posted the abstract and a link to one of them in this thread.
Posted by: Tim Lambert | April 15, 2007 9:56 AM
How does "any confidence interval is directly affected by the size of the sample on which it is based" imply "with large data sets, confidence intervals have to be increased, so a 90 per cent confidence level would not then be valid" ? Particularly when the first statement is true, but that second statement is impossible to interpret in any way that corresponds to actual statistical theory or practice, (or even makes sense on a standalone basis)?
Posted by: z | April 15, 2007 3:33 PM
"As the number of obs gets big so the square root get big etc. I am shocked to hear this isn't a problem in practice. In finance data we often have large data sets and this is a problem. "
Let me get this straight. Are you saying that in the graphic above, the fact that the data follows the pink prediction and not the blue prediction is invalid, because there is (or might be?) too much data??? And that this argument is commonly held to be valid in the world of finance? If so, that would explain one hell of a lot regarding the strange refusal of trickle down economics to stay decently dead, for instance.
Posted by: z | April 15, 2007 3:52 PM
Tim, I'm sorry if if misrepresented your argument. I was taken in by the sentence where you said you don't need to read stuff to know the outcome. Yes, I have accussed the IPCC of not having human attribution studies because they say they don't have the studies. Accuse is too strong a word perhaps, quote would be a better phrase.
Yes, it can happen that having too much data biases standard hypothesis tests. It wouldn't happen unless you have many thousands of data points. I don't know how much data the IPCC have - they say they have a large data set and then report a marginal confidence level of 90%. In that situation my first thought is always "Lindleys paradox? Is it a problem here?" It may not be, but I reckon they never even thought about it. (To be fair, most people don't worry about it because having large datasets is a recently luxury). Now, it may be they don't have dataset large enough for Lindleys paradox to even be a problem, but again we don't know. It's just not good enough. If the IPCC are preaching to the choir then the SPM is fine. But if they are trying to provide a summary of evidence for use by policy makers, they the SPM is poor job. Why is that?
Tim, the book is Judge etal. Intro to practice of econometrics (title from memory) it's out of print now. Another is by Zellner that's an intro to Baysian stats (probably of out of print too). Leamer wrote about this as well in one of his books (probably not an intro, from memory). There are also a few articles aimed at students and practitioners that bring this up. In stock market data this can be a serious problem especially when using high frequency data (or even long periods of faily data). Climate data is not high frequency but there might be several thousand observations in any one dataset if not more(afterall there are millions of years of climate change). It is something that needs to be checked. If not a problem it needs to be footnoted, if a problem it needs to be adjusted for. 90% confidence may not be statistically significant in large sets. Of course we can quibble about the definition of large and the SPM is silent on that.
Thank you, all. This has been a very civilised (and fun) discussion.
Posted by: Sinclair Davidson | April 15, 2007 5:45 PM
Tim & Sinclair
http://www.amazon.com/Practice-Econometrics-Probability-Mathematical-Statistics/dp/0471059382
probably this one. From memory though, there was a book 'Judge & Box' Econometrics, and yes it was blue and massive (lost my copy when I moved house).
Posted by: Valuethinker | April 16, 2007 2:56 AM
Steve
You are too pessimistic on non-carbon energy sources.
Wind capacity, worldwide, is growing at 30% pa. That kind of exponential growth (doubling capacity every 2.5-3 years) means wind power is going to be a significant source of energy. 60GW installed capacity now, but quite credibly 1000GW installed capacity by 2025. (US is about 800GW installed capacity for all forms of electricity generation).
And it's a big business. In Spain, Germany, Denmark, the UK and the US (notably that home of green sentiment, Texas), it's some of the largest power companies in the world that are investing.
Average cost per GW of wind capacity is falling by c. 18% with each doubling of sector size (the normal learning curve effect).
And offshore wind, which is even more promising (because the wind even 3 km offshore blows much more frequently, and stronger), is barely beginning.
In some markets (Ireland) new wind capacity (without subsidy) is competitive with new gas fired plant.
Because we don't tax carbon emissions, wind still requires subsidies to be competitive (but, like nuclear, the capital cost is all up front, the operations & maintenance is a relatively small figure, whereas with a gas-fired combined cycle (CCGT) plant, fuel is 2/3rds of the lifetime cost of plant).
I am all for sensible macro policies (like carbon taxation).
Where there is a role for 'regulatory' solutions is in areas where:
That's a 40%, guaranteed, annual return that consumers are turning down.
Interestingly, when it comes to energy conservation incentives, industry functions on similar payback rules (2-3 years to justify) which is a far higher (and safer) return that venture capital demands.
See all the papers on 'hyperbolic discounting' by Laibson et al, as well as the work of Kahneman, Tversky and Thaler and Meir Statman ('Prospect Theory' and 'Behavioural Finance') to see why this might be so.
situations of limited information when consumers don't know what the lifecycle energy costs of what they are buying is
split incentive problems eg in commercial lets, the tenant pays the energy cost, but the landlord specifies the building (typically at lowest cost). The tenant won't be around long enough to reap the benefits of more efficient heating, ventilation, air con etc, and the landlord doesn't care
'white roofs' - this one has been studied to death at Berkeley's Energy lab. Painting the roofs of the houses and buildings of the southern 2/3rds (roughly) of the USA white can save up to 10% of all air conditioning bills (the offset in terms of lost winter insolation is about 1%). Much of the benefit of this action actually accrues to your neighbours (who don't get the heat spilled out from your air con) and people driving by ditto. So it won't happen if we don't collectively mandate it (benefit to me isn't worth the hassle).
Whilst electricity consumption per head in the US since 1980 has risen by 40%, in California it has risen not at all, despite California being the 5th (?) richest state in the Union. This is the consequence of tough energy efficiency standards and a model of what the rest of us can achieve.
Posted by: Valuethinker | April 16, 2007 3:20 AM
Valuethinker, somehow "Theory and Practice of Econometrics (Probability & Mathematical Statistics S.)" doesn't sound like an elementary text to me.
Apparently the publishers don't think so either: "This broadly based graduate-level textbook covers the major models and statistical tools currently used in the practice of econometrics."
I'm sure Sinclair had something else in mind.
Posted by: Ian Gould | April 16, 2007 8:23 AM
Posted by: Eli Rabett | April 16, 2007 9:29 AM
Sinclair, the IPCC SPM says they have attribution studies. I already quoted the abstract and linked to one. Care to comment on it? Oh right, you're not going to because then you would have to admit that exists. Why don't you just simplify things and deny the existance of the IPCC while you are at it?
I guess that is also why you haven't looked at the graphs in my post that show attribution studies using the last 100 years of data. That's not a lot of data, and the only reason why can detect significant effects is because the effects of anthro ghgs stick out like a sore thumb.
And no, the SPM did not do a bad job. The fact that you