The Russian Institute of Economic Analysis (IEA) claims that the Climate Research Unit (CRU) used only one in four available weather stations in Russia over the past fifty years (see here for the IEA report in Russian). CRU uses their selection of stations to generate their global surface trend. According to the president of the IEA, "The IEA authors calculated that the scale of actual warming for the Russian territory in 1877-1998 was probably exaggerated by 0.64°C." He claims that the stations selected by CRU are mostly urban, and that the rural Russian stations show no warming in the past fifty years.
Some climatologists appear to have suspected the Russian stations were being selected based upon their warming trends, and tried to publish academic papers on the subject back in 2004. Unfortunately for them, their papers were reviewed by the director of the very same institute they were criticizing. In this e-mail, we see Phil Jones, the director at that time, telling Michael Mann that he rejected the papers.
The station data we obtained from CRU via NCDC may be incomplete. The trend we obtained ourselves may contain an erroneous 0.64°C rise in Russia, which represents 12% of the world's land mass. We do not know how much pre-selection of stations has been done in other parts of the world. We see no evidence of such selection being performed upon US data.
The US data appears to be subjected to corrections instead of selections. For example, according to a NASA Paper, rural stations in the US show a cooling of −0.05°C from 1990 to 1999, but after NASA applies its correction, rural stations show a warming of 0.35°C. Most of this warming is due to what the authors call the "time of observation debiasing" correction. (We did our best to wade our way through this paper and all its corrections, if we have made a mistake, we welcome correction.)