I didn't suggest that mac's were immune to threats, that would be stupid. Given their BSD base, though, mac's are inherently more secure than windows boxes.
I think the difference is great enough that you can't use virus penetration as a measure of system popularity.
In that case, I didn't suggest that virus penetration be the metric, but attempts at penetration.
Although it is likely that there is a correlation between attempts and the perceived vulnerability of the system, you would also expect a correlation between attempts and expected total success (which goes up with market share).
I think the difference is great enough that you can't use virus penetration as a measure of system popularity.