Should we go without
such uncertainty intervals and argue purely from the likelihood of certain point value choices for candidate values for the unobservable parameter?
Not exact matches
Point estimates of population parameters (e.g., mean, correlation coefficient, slope) or comparative measures (e.g., mean difference, odds ratio, hazard ratio) should be accompanied by a measure of
uncertainty such as a standard error or a confidence
interval.
By contrast, a (frequentist) x % one - sided confidence
interval with a limit of y can, if accurate, be thought of as one calculated to result in values of y
such that, upon indefinitely repeated random sampling from the
uncertainty distributions involved, the true parameter value will lie below y in x % of cases.
But it is very hard to understand how
such large
interval can be derived from the evidence IPCC itself gives: Palaeosens gives a likely 2.2 - 4.8, CMIP5 models is in the range 2.1 - 4.7 and the
uncertainties in inter-annual feedback observation analysis is mainly from cloud feedbacks which is said to be «likely» positive.
Such inflated
intervals would be a valid description of the
uncertainty in the maximum or minimum of the reconstructed temperature series.