In his 1954 book “How to lie with statistics,” Darrell Huff devotes a chapter to the “Gee-Whiz” graph. He shows that by truncating figures and altering the proportions between axis that it is possible to create a misleading impression without falsifying any data.
For people who would like you to belive that climate change isn’t a big problem this approach can sometimes be too tempting to avoid. A prominent think tank has a figure similar to the following on their homepage.
I have used data from the UEA Climatic Research Unit to reproduce the figure. There doesn’t seem to have been much in the way of global warming over the last decade, right? But if we want to see if “Global Warming has Halted,” we can use the same data source to take a look.
Borrowing an idea from Tamino, in the below figure I have plotted global mean yearly temperatures from 1975 up to 2000, using the same data source as for the figure above. I have also plotted a trend line (0.0174 °C∙yr-1) estimated by linear regression and dashed lines representing ± twice the standard deviation of the residuals. The idea is that aproximately 95% of the yearly values should fall between the two dashed lines. (Actually, to get things completely right, I need to account for autocorrelation of the residuals, but it shouldn’t make too much difference for our purposes.) By extending the dashed lines forward in time to the present we should be able to see if warming has halted or not.
If when the data points for 2001 – 2009 are included they sit within the two dashed lines, then claims that global warming has halted would seem to be premature. In fact as the below figure shows, the global mean yearly temperatures for 2001 – 2009 are right where we would expect them to be if global warming were continuing without pause.