While I was driving down to Portland to take care of my nephew and niece for the evening, it occurred to me there's a better way to calculate climate sensitivity from the data, if you assume atmospheric CO2 levels are increasing exponentially with time t:
C(t) = C0 exp(at)
where "a" is some constant to be determined by the data. Then the climate sensitivity S is
S = [mT/a]*ln(2)
where mT = ΔT/Δt is the linear slope of temperature change with time. This avoids having to pick an N-year moving average and just uses the long-term trends in the data. It's what you'd expect for an exponentially increasing CO2 level, since then the time for CO2 to double (D) is just
D = ln(2)/a
S = D mT
which, of course, is just what you expect.
The constant "a" can be calculated by linear regression of ln(C/C0) versus time, and using the NOAA CO2 Mauna Loa data I find D = 166.0 years. The exponential assumption is a very good one (as Keeling discovered, of course,), since the Pearson correlation coefficient is R2 = 0.9914.
From the data I find:
mT(Hadley) = 0.126°C/decade
mT(UAH LT) = 0.135°C/decade
The difference is that the calculation for Hadley starts in 1958, when the CO2 data starts, and surface temperatures really didn't head upward strongly until the mid-70s when, shortly after, UAH started to record data. So
S(Hadley) = 2.1°C
S(UAH LT) = 2.2°C
You can calculate the statistical uncertainty too, since you know both of these from the data, which I'll do sometime soon.