As I said, the results depend on the assumptions. So let's make some.
To a first order, you can safely assume that a SS signal looks like bandlimited white noise with a reasonably flat spectrum. (IS-95 CDMA has a little ripple in its spectrum due to the sharp FIR filtering after spreading. We call the composite signal a "Bart's Head" because of its resemblance to the top of Bart Simpson's head when viewed on a spectrum analyzer. The edges are very sharp.)
You can assume that the SS power level will be substantially lower than the equivalent narrowband emission for the same user path. This is due to automatic power control and to the use of power-efficient forward error correction. Most local FM links run with substantial margin, especially in the amateur service, and this tends to skew people's expectations of the power levels that are necessary. They're usually quite surprised when they see how little power a good SS system actually uses (see below). (This is my first major objection to the W3OTC "analysis" to which you refer -- he assumed that just because 100W SS is allowed, it will always be used.)
You can assume the local propagation loss varies according to an