Is the QDE Toolkit based on the Bayesian or frequencist schools of statistics?
2002-06-03 We think that the QDE Toolkit is compatible with the best elements of both schools, being based on the ISO Guide to the Expression of Uncertainty in Measurement, which uses both schools. For a thoughtful attempt to address this issue from a statistician’s perspective, see L.J. Gleser’s paper Assessing Uncertainty in Measurement, Statistical Science, 13, 277-290 (1998). To use both schools, one needs to be adept at transforming back and forth between a frequencist’s distribution (of measurements) and a Bayesian distribution (of measurand). In our applications, at worst this requires care to “flip” distributions that are not symmetric about zero. Many statisticians will also expect scrupulous distinction to be maintained between the two pictures, even when there is no observable asymmetry. In our view of metrology, the first contact between experiment and statistics is in the frequencist picture: we measure something repeatedly and we simply record what happens. Any unexpected
Related Questions
- Are high schools required to enter attendance once-Division time, every period based on the school’s restructured day (block – 4 per day), or for every period (breaking clock into 7-8 periods)?
- Statistics from Piwik and my other web analytics tool (log analyzer, webalizer or javascript based) are different, why?
- May a district establish transportation zones within a district based on the geographic location of schools?
