Or why I love the internet. Iowahawk has a very good walk-through of regression analysis as applied to climate data – with a detailed how-to for using real NOAA data and creating/replicating some of the temperature graphs in various research papers. He makes the point that the famous “hockey stick” global warming graph is the result of a plausible choice for proxy variables (tree ring data, ice cores, etc). And also makes the point that other reasonable choices for proxy variables (bristlecone pine tree ring data, for example) yield very different graphs for global temperatures. One of the surprising things is that the proxy variable/principal component choices by the CRU team makes the Medieval Warming Period vanish, but that including bristlecone pine tree ring data as a principal component yields temperature graphs that keep the MWP.
I haven’t done the downloading and analysis myself yet, but I intend to. I will share anything interesting that I find.
Later: I can confirm that Iowahawk’s how-to does work and yields the graph described. This is fun, but not science – since I don’t know enough to think critically about what are wise or unwise choices for principal components.