Last night I was up in Santa Fe listening to a talk by Jim Hansen, author of Storms of My Grandchildren and one of the primary researchers in the field of climate change and its relationship with humans.
I had never been as conscious or critical of the facts presented by scientists as I was during that talk, and I believe a very large part of it was due to How to Lie With Statistics and what we have been learning here. For instance, although many of his temperature (global temp. change since the beginning of the Cenozoic era, sea temp. change since 1950 etc.) graphs and data appeared to have a clearly positive correlation (i.e. as CO2 emissions increase, temperature also increases), Dr. Hansen did not provide a single P-value and thus I wondered how significant the conclusions he drew were... Perhaps it was because the average listener does not understand what a p-value is and he didn't want to waste time explaining it, but nonetheless it made me wonder...
Mika-
ReplyDeleteI definitely can relate to your experience. I just recently went to a talk myself. While listening to last week's chemistry department seminar, I also became concerned by the numbers we were provided. The seminar speaker was discussing his findings on solar cell efficiencies. I was confused why he didn't include uncertainties in his final measurements. The efficiencies he was recording were all quite small (1-2.5%). With one combination of components, he saw something closer to 5% efficiency. Was this increase significant? If you take everything as an absolute percentage (which we know is dangerous), then yes, it is quite significant. But how much was this varying??? Is the variance larger than the increase?