# TGraphAsymmErrors:BayesDivide - adding separate measurements

Dear all!

I have a naive question concerning TGraphAsymmErrors:BayesDivide. I have only one bin and let’s assume I expect N and get x in that bin. I can easily put N and x in two histograms and let TGraphAsymmErrors:BayesDivide calculate the efficiency and the errors. How do I deal with the situation that I repeat the same experiment, i.e. I assume the same efficiency, and the result is that I expect N’ (because of different conditions) and get x’. How do I combine the two measurements? Is it safe to add N and N’ and x and x’? Should I call BayesDivide for both cases and combine the two TGraphErrors somehow?

Sebastian

Hi,

I have not fully understood what are N and x in your case. For calculating the efficiency (let’s assume only 1 bin) you need two values, K and N and eff = K/N.
For combining the measurements you would need two new values, K2, N2 and you combine summing them separately. You cannot then combine the two TGraphAsymmErrors, you need to combine first the two histogram by merging them using TH1::Add.
I hope I have understood your question,

Best Regards

Lorenzo

Dear Lorenzo!

Thank you very much. K=my N, N=my x.

So you are sure that it is statistically correct to sum two measurements? K=K1+K2, N=N1+N2? For the mean efficiency for sure, but also for the errors?

Best regards,

Sebastian

Hi,

the error will also be calculated correctly if you sum both the k (# in the numerator) and N ( # in the denominator).

Lorenzo