First of all, thanks for being out there!
Here is my question:
Ive confirmed that two methods (used to determine % nitrogen content in materials) when compared are (statistically) significantly different. The new method will save dollars but is different from the old one.
Non parametric method was used since they are clearly non normal. Transformation unsuccessful. But thats not the problem. I dont think that we can argue that result. And Im going to state it.
The total tolerance for the allowable nitrogen content for this test according to specification is 1.3% (11.5 % to 12.8 %)
Median Old Method 1 = 11.95 %
Median New Method 2 = 12.3 %
there is no overlap and as you can see the difference is 0.4% in the medians
The question is Im seeing a 0.4% difference in medians, which is about 27% of the total tolerance of 1.3 %.
Am i leaving any stone unturned to say that although the methods are statistically different, there may still be equivalent from a practical aspect?
Is there any rationale or technique on allowable difference of means wrt tolerance ……………that I can use/ exhaust at this point?
Priya
.
.