Tuesday, May 01, 2007

correlation and causation

This comment was originally posted as a response to a discussion point on ClimateAudit.
The thread focused on the alleged discrepancy in the Great Global Warming Swindle documentary. The producer acknowledges the error and will correct it prior to the release of Swindle on DVD. Meanwhile, many, many proponents of anthropogenic global warming (AGW) have jumped on this error as "proof" of the program's entire irrelevancy. The discussion at ClimateAudit lauds the scrutiny aimed at Swindle but queries why those self-same guardians of scientific integrity do not similarly deconstruct the IPCC report or Gore's Inconvenient Truth, or acknowledge the errors both have when others do establish them.

It is within this thread and discussion that Francois Ouelette posted this comment:


  • Francois Ouellette says:
  • May 1st, 2007 at 7:33 am
  • I urge everyone to pay attention to Friis-Christensen’s statement:
  • For a physicist a break down of a correlation where you would expect one is just as – or sometimes even more - informative as a good correlation when it comes to the ultimate goal, which is to understand the physics. Climatologists are more concerned whether the observations fit their preconceived model and prefer to describe solar activity by one single parameter.
  • The breakdown of the correlation between sunspots and temperature is used ad nauseam as an argument against a larger solar role in 20th century warming. Read RealClimate for example. Yet, as Steve points out, the same argument does not seem to hold for the divergence problem, or for the simple fact that CO2 and temperature do not correlate during the period. Thomoas Kuhn has described how, when faced with “anomalies”, scientists tend to refer to ad-hoc hypothesis to salvage a theory. In this case it’s either aerosols, or CO2 fertilization, or what else. But the anomalies are real, and they may indicate a more serious flaw in the theory. The same applies to any solar theory. There is a breakdown in correlation after the late 1980’s. Which just goes to show that, as Friis-Christensen aptly points out, correlations are but a clue to an underlying physical mechanism. One must then elucidate such mechanism before proceeding further, and, for example, including the mechanism in climate models.
  • There could be many physical reasons why the global temperature stops correlating well with the solar cycle at the end of the 80’s. GHG’s are one of them. Other human influences on climate are possible, as repeatedly pointed out by Roger Pielke Sr. It could also be that the mechanism linking the sun and the climate is just not linear, and past a certain level of solar activity (which is what happened lately), we enter a different regime. There could be a feedback or threshold effect that we’re not aware of. I have a paper somewhere (too lazy to look for it this morning) that has analyzed the correlation between the sun and the climate using wavelets, and they do see a change in the pattern at the end of the 20th century. There is still a correlation, but it seems to change phase or something. Phase changes have been noted by others whenever there was a big volcano eruption, like the Pinatubo in 1991.
  • So there is plenty of “juice” left in the solar-climate link research program. Whether that program will be allowed to proceed or will be censored by the pontiffs of AGW is an important question. Do we let scientific progress be conditional on political and ideological rectitude? Many argue that we are in a “post-normal” science era, where anything goes, as long as it suits a particular point of view. That would indeed lead us to disaster.
  • OT (but not quite): I’ve just seen the movie “The lives of others”, a German movie about how everybody was spied on in east Germany. All in the name of the good socialist state. Having known many former east-Germans (especially scientists), I know that this was just too real. Is that what we want?
Every so often, you read something so concise and accurate in its tone that it would be an injustice to paraphrase or selectively quote its meaning. That's how I feel about the above comment.

Scrutiny in science is good. Data disclosure and archiving should be the norm. Errors are something to learn from not hide from. False allegiance to dubious hypothetical constructs and post-normal science are a recipe for totalitarianism.

As another post states:
  • Grosverson says: May 1st, 2007 at 9:14 am
  • Let’s see, where have I seen this type of double standard before? Oh yes, in religion and politics. If catastrophic AGW is a religion or political movement, this double standard would all make sense.
  • This type of double standard in criticizing the faithful vs. criticizing heretics is common in religious hierarchies or political autocracies. It’s certainly not science.