Always ~ teaser. Details on my previous statements after having checked.
I confirm that my 'reasonable' values are based upon the histogram (non parametric distribution) and not the lognormal. They rule out the tails of the parametric distribution, that is the extreme behaviours.
reasonable LB= 70 mgN-kg-1d-1 = 0.44 g kg-1 d-1
reasonable UB= 160 mgN-kg-1d-1 = 1.0 g kg-1 d-1
My previous estimate was that 90% of the data are included within those bounds, now, how many exactly?
n=235, The number of values outside the range, counting the number of values each bin = 25 (pls check my calcs if possible).
25/235= 0.11 = 11% , hence 100%-11%= 89%, practically equal to my previous quick-look estimate of 90% .
Again, table one uses a parametric model, based on the lognormal distribution fitted upon the non histogram. So the range here is more restricted than the histogram range.
If we use the lognormal distribution as reference, my statement , answerign to mushymindy, should be rewritten like this:
Quote:
The complete, honest answer according to the WHO, 2002 study is that, with about 100% certainty (absolute certainty), your minimum requirement will probably be something within the range of 25 to 56 grams of proteins.
|
And I absolutely agree that using the lognormal would be a benefit sicne we have a narrower range, hence a more precise determination.
continues next.