An example: how people calculate the "average temperature of the Earth"
via http://arbat.livejournal.com/855100.html#cutid11

The average temperature was extrapolated with precision 0.01 degree, from data that covers only 12% of the surface of the Earth, including data from some weather station that were "adjusted" so that the results show October 2015 to be "hottest October ever".
Data from satellites (covering all surface) show that there was no especial warming this October.
http://realclimatescience.com/2015/11/record-crushing-fraud-from-noaa-and-nasa-ahead-of-paris/


In my previous post, I estimated that the average temperature of the Earth surface is only defined up to 2 degree uncertainty. By choosing one or another averaging procedure, one can obtain any result within the 2 degree uncertainty. If the data is "adjusted", of course any result is possible.