Month to month the two data sets vary by around 25% at the power plant level. But the variations are negative some months and positive in others. So on an annual basis the variation between the two data sets is around 2% or 3%. Various statistical test indicate that the variations are largely due to bias (in the statistical sense), and are mostly not random.
So on an annual basis would you decide that your ANNUAL data is precise within 2% or 3%? Or since the monthly data is varies widely between two methods would you put your precision more at 25%? (I'm assuming this kind of statistical test can only verify precision and not accuracy, but maybe it has implications for accuracy as well.) Can an aggregate measure be more precise than the sum of the individual measurements upon which it is based? If it can, when can it and when can't it?
Thanks
Gar
-- Facebook: Gar Lipow Twitter: GarLipow Solving the Climate Crisis web page: SolvingTheClimateCrisis.com Grist Blog: http://grist.org/author/gar-lipow/ Online technical reference: http://www.nohairshirts.com