The rule of thumb was probably derived from scientific literature. It's very common in the scientific world to express temperature differences in Kelvin -- the official SI unit for temperature and temperature intervals -- even when using degrees Celsius for measurements. Part of the reason is because the Celsius scale is not zero based (not an absolute scale), but it's really a matter of custom/tradition than anything else.
E.g., OAT measurement of 15 degrees C, +/- 3 K.