How Much R1b?
Today I revisited the estimate, using the data from the Giacomo paper and a table of regional populations.
Now there are couple of limitations to this analysis. First, Giacomo didn't specifically measure R1b: he reports P*(x R1a). P is the parent of haplogroups Q and R, and so P is actually measuring the frequency of "true" P, Q, R1b, and R2. Undifferentiated P is fairly rare in Europe, but it exists in small amounts along with Q (except Q3, which is largely associated with Native American populations) and R2. Thus, the reported frequency of P*(xR1a) actually overstates the frequency of R1b by a little bit.
The second problem is that Giacomo did not report all other haplogroups, though I think he got all the major ones. Still, this will also cause the R1b proxy to skew a little high.
Anyway, using his data I estimated the P*(xR1a) frequency for each region and then took a weighted average using the regional population as the weight. I got a weighted frequency of about 45%, which means that (given the caveats above) I think the frequency of R1b in Italy is maybe 40-42%, or not far off the estimate I made yesterday.