Wednesday, December 26, 2012

"Omitted variable fraud"

"Omitted variable fraud." I like the term. This concisely describes the behavior I have been observing for as long as I have been paying attention to the issue, ever since Gavin Schmidt took on Jerry Pournelle with the repeated assertion that direct solar forcing failed to explain the temperature change in NASA's models, and that CO2 was responsible. He never acknowledged the gigantic leap of faith he was taking in believing his models over the historical data.

Anyway, I quote the money paragraph first and then some of the background.

http://wattsupwiththat.com/2012/02/22/omitted-variable-fraud-vast-evidence-for-solar-climate-driver-rates-one-oblique-sentence-in-ar5/

My training is in economics where we are very familiar with what statisticians call "the omitted variable problem" (or when it is intentional, "omitted variable fraud"). Whenever an explanatory variable is omitted from a statistical analysis, its explanatory power gets misattributed to any correlated variables that are included. This problem is manifest at the very highest level of AR5, and is built into each step of its analysis.

Introduction to the "omitted variable fraud" critique, continued

For the 1750-2010 period examined, two variables correlate strongly with the observed warming (and hence with each other). Solar magnetic activity and atmospheric CO2 were both trending upwards over the period, and both stepped up to much higher levels over the second half of the 20th century. These two correlations with temperature change give rise to the two main competing theories of 20th century warming. Was it driven by rapidly increasing human release of CO2, or by the 80 year "grand maximum" of solar activity that began in the early 1920′s? ("Grand minima and maxima of solar activity: new observational constraints," Usoskin et al. 2007.)

The empirical evidence in favor of the solar explanation is overwhelming. Dozens of peer-reviewed studies have found a very high degree of correlation (.5 to .8) between solar-magnetic activity and global temperature going back many thousands of years (Bond 2001, Neff 2001, Shaviv 2003, Usoskin 2005, and many others listed below). In other words, solar activity "explains," in the statistical sense, 50 to 80% of past temperature change.

Such a high degree of correlation over such long time periods implies causality, which can only go one way. Global temperature cannot be driving solar activity, so there must be some mechanism by which solar activity is driving or modulating global temperature change. The high degree of correlation also suggests that solar activity is the primarydriver of global temperature on every time scale studied (which is pretty much every time scale but the Milankovitch cycle).

In contrast, records of CO2 and temperature reveal no discernable warming effect of CO2. There is a correlation between atmospheric CO2 and temperature, but with CO2 changes following temperature changes by an average of about 800 years (Caillon 2003), indicating that it is temperature change that is driving atmospheric CO2 change (as it should, since warming oceans are able to hold less CO2). This does not rule out the possibility that CO2 also drives temperature, and in theory a doubling of CO2 should cause about a 1 degree increase in temperature before any feedback effects are accounted, but feedbacks could be negative (dampening rather than amplifying temperature forcings), so there no reason, just from what we know about the greenhouse mechanism, that CO2 has to be a significant player. The one thing we can say is that whatever the warming effect of CO2, it is not detectable in the raw CO2 vs. temperature data.

This is in glaring contrast to solar activity, which lights up like a neon sign in the raw data. Literally dozens of studies finding .5 to .8 degrees of correlation with temperature. 

-Max

--
Hahahahaaaa!!! That is ME laughing at YOU, cruel world.
    -Jordan Rixon

I could not love thee, dear, so much,
Loved I not Honour more.