Several papers have used correlations between outgoing radiation and surface temperature to estimate climate sensitivity via a linearized energy balance equation. A crucial assumption in such studies is the use of global averages to relate radiation and temperature data that are actually functions of space and time. Here it is shown that this assumption is important to understanding why transient and equilibrium climate feedbacks may differ. There are important limitations to derivations of climate sensitivity from correlations between radiation and temperature data. In particular, because of heat transport between regions, obtaining the equilibrium temperature change by multiplying the forcing by a climate sensitivity is valid only for a global domain. The analysis of Lindzen and Choi (2009) erroneously applies global concepts to a limited region. I also show that a simple, point-by-point regression of outgoing radiation against surface temperature gives better slope estimates than the interval method used by Lindzen and Choi (2009).