Friday, October 09, 2009

Hockey Stick

I am not a climate physicist, but some years of experience in another field have given me a little intuition about when something is amiss. I may not have the expertise to say "You made thus and such an error" but I can say that something doesn't look right. The famous global warming hockey stick graph set off my glitch alarm.

It looks like I was right to be suspicious--the scientist has been cagey about releasing data, and apparently his tree core sample was hand-picked from a larger set of cores with no explanation of why the rest were excluded. (When all are included, the graph shows no spike.)

You always have to clearly explain why data sets are omitted, and demonstrate that omitting them doesn't bias the results. It seems that half the time we spend estimating the uncertainties in our measurements and half the rest of the time trying to understand the biases.

No comments: