2 stat classes huh? bahaha, you should probably get your money back. Look it goes like this. You make assumptions on temps based off growth of trees. More heat, longer growing season, more growth. No actual temperature data, just assumptions on tree growth directly correlated to temperature, with no real factoring in of other factors related to tree growth. Then you go through those sames cherry pick the ones that fit your desired results and use those in the model. (this did actually happen)
Or you can take samples of actual measurements from "official" weather stations. Because every one of those weather stations use the exact same thermometer. And every thermometer was tested using the exact same stringent quality standards. None of those thermometers were ever moved not did we change the method of gathering the data in those 100 years. Lets face it, Joe blow could look at one of those thermometers and say 100 degrees and then Jim blow could look at it and say 101 degrees. I've seen and witnessed modern weather NOAA approved weather readings... There is only the hope that weatherman 1900 was meticulous in his recordings and measurements. But then we take that data they recorded, and say without a shadow of a doubt that oh no, the world has warmed 2 degrees, now we're going to enact policies that cost trillions of dollars, triple the cost of gas. yes, that makes perfect sense.
So to put it in a nut shell. There is no way that 2 degrees is within any sort of margin of error. No matter how you cook the data. Even if you can assume that it fits within a 95% probability curve or 90% or whatever. Because well, the data cannot be verified as good...