SD3 Wrote:
Jan 20, 2013 9:50 AM
You don't have to sample at every possible location to compute an average with a high probability. Unless you the temperature effect of climate change (or lack thereof) is distributed in a non-random way, there is no statistical reason why measuring temperatures over time at, say, 5000 randomly chosen locations wouldn't yield a highly confident average temperature timeline. The real point that you should be making is not that the statistic is doubtful (it's not), but that it is erroneous to infer major changes in climate (that we know from history happen over thousands of years) from a sample statistic of a hundred years. It is like saying that the insulation in my house is getting poorer because the temp fell between 8 and 9am.