Hi all, from time to time, the problem of zero values in the pattern is discussed. I have done an indepth Bayesian approach: Have assumed identical probability of all values prior to measurement and have calculated the Bayesian distribution of probabilities after having measured n counts, its estimated value and esd. The results for time preselection: -Having measured n counts, the estimated value is n+1 -Having measured n counts, the esd is also sqrt(n+1)! Some detail: The correction of the estimated value is invisible, if one uses constant time preselection in the pattern: The values simply are shiftet, and the background changes. Things change with variable counting times. We have done measurement with variable (sometimes very low) time preselection, and the above correction for the estimated values gave significant better R values.
Greetings Joerg Bergmann, Dresden
