At our weekly group meetings, one member from each half of the group gives a 20-minute progress report on his or her research. This week, one of the plots shown during the Space talk contained data that showed a pronounced gradient change at a particular point in time. The plot didn't include error bars, however.
Atmos professor: "So what are the error bars?"
Speaker: "About 10 degrees at most."
Atmos professor: "Oh. Because I was going to suggest if they were big enough, you just do a linear regression on the whole data set and ignore the gradient change."
Space professor: "Is that how you guys discovered climate change?"