Questioning Climate Statistics

There is quite a discussion going on over at Watts Up With That regarding a paper that appeared in Nature. My sister has already commented on the study. The paper claims to have used statistical methods to demonstrate that the temperature of Antarctica has been rising for several decades. The discussion going on now at Watts Up With That is about the statistical methods used. Knowing enough of the math and statistics to understand much of the discussion, it's tempting to delve in further and see for myself if the study is sound. The study in question uses something called Regularized Expectation Maximization. A fundamental paper in the climate area on this is

Schneider, Tapio. "Analysis of Incomplete Climate Data: Estimation of Mean Values and Covariance Matrices and Imputation of Missing Values" Journal of Climate, Vol 14, 2001, pp 853-871.

From the abstract:

In contrast to the conventional EM algorithm, the regularized EM algorithm is applicable to sets of climate data, in which the number of variables typically exceeds the sample size.

More variables than data samples? Are they serious?

I salute the folks at Watts Up With That for questioning this study. But do we really need to go into the statistical detail here? Isn't it prima facie ridiculous to have more variables than data points in your analysis? The abstract tells us that this is common in the climate analysis field.

The more I look into climate science, the more skeptical I become.

Ann says: I should think that grade schoolers could figure this one out. When were we all taught that it takes 2 points to make a straight line? These guys must be geniuses; for them, it only takes one.