For coverage factors, what can you say about the difference between the “k=2” used in Version 1 of the Toolkit, and the “95%” used in Version 2, which for infinite degrees of freedom gives “k=1.96”?
2002-05-10 Since for Version 2 of the Toolkit the objective is to calculate the coverage factor for any degrees of freedom, our choice was to be consistent with the MRA (which specifies 95% confidence), rather than with the common “k=2” approximation that we used for Version 1. For infinite degrees of freedom, “k=2” implies a confidence of 95.45%, which is not what is used in the MRA Degrees of Equivalence. Of course, there is little or no practical difference between these two numbers. Standard uncertainties are rarely known this well with any real conviction. Except for decisions about being declared an “outlier”, where the 2% broader “k=2” interval may be welcomed (note the perversity here: the higher confidence is more forgiving), the exact 95% seemed like the better choice. In several places, the Toolkit Version 2 has its Visual Basic for Applications code already set up to facilitate changing to 94.45% and “k=2” if that is what you really require. If you change the VBA code from
Related Questions
- What is the difference between the Kindergarten and the First grade version of the OLSAT? Why is the same book used to prepare children for both?
- What is the difference between the GM 160,000 km Powertrain Component Limited Warranty and my original warranty coverage?
- Whats the difference between the needle punch Im used to receiving and the ColdForm™ version?
