On Nov 12, 12:50 am, djh <halitsk...@att.net> wrote: > Thanks for your comments on applicability of the B&M paper, > and where/how the scope of the present inquiry exceeds theirs. > > I. > > When you have a moment, and if you have the patience, I'd still like > to understand if/why a referee would reject the argument in my post of > 11/10@1:09. i.e. the argument "from p's" based on behavior of dicodon > set 2 vs dicodon sets 1,3 with respect to the three simplified > regressions ln(c) on ln(e), ln(c) on ln(u), and ln(c) on (lnc,lnu).
I can't follow the logic. First, the simplified model is not a legitimate replacement or substitute for the original model, but you seem to be treating them as interchangeable. Second, I thought that one of your requirements was that all the 3-way interactions had to differ significantly from zero in both the statistical and practical senses of the term. Third, are you now saying that you want a 4-way interaction, that the 3-ways should differ across dicodon sets? Or are you saying that it's OK to ignore dicodon sets that don't show a 3-way interaction?
> > II. > > Am I even partially correct in thinking that this page of David > Howell's: > > http://www.uvm.edu/~dhowell/StatPages/More_Stuff/Ogle.html > > might suggest a way of fruitfully studying the pairs (c,e), (c,u), > and (e,u) > > What I'm thinking here is to get an idea of what bivariate normals > would look like in our case by doing the following: > > 1) express each of our variables c,e,u within the interval N(0,1) > as c', e', u'; > > 2) choose one pair of variables, say (c',e') > > 3) apply Ogle's algorithm but starting with two randomly selected > values of the variables (c',e') at a given L (i.e. two values that > are not necessarily associated with the SAME observation) > > 4) repeat 2-3 for (c',u') and (e',u') > > 5) repeat everything for each value of L
Don't bother. All bivariate normals look alike, differing only in their ellipticity, which reflects the correlation. Means and SDs (location and scale) affect only the labeling on the axes.