On Jun 12, 7:25 am, djh <halitsk...@att.net> wrote: > 1. Tool for genersating correlations > > Thanks for re-doing the correlations reported in Tables I and II of my > last post. Once we are through this initial phase of verifying > meaningfulnes of existing "driver" correlations (and/or defining > better ones), I will either switch to my wife's copy of Minitab or > purchase a very good but cheap stat package (about $175) now > available. > > 2. Significance testing on Tables I and II (relative to questions > 1-3) > > I've never pestered you before and I deeply apologize for doing so > now, but if there's any way you could have some even preliminary > significance testing results re questions (1-3) by 130pm US EST > tomorrow (6/13), that would be really great. I'm meeeting with Bob > Lewis at Fordham and the two students he's directing at that time, and > would really like to be able to report that: a) ln(c/u) on ln(c/L) is > "meaningful" when run "relative to e"; or b) ln(c/u) on ln(c/L) is NOT > doing anything trustworthy enough to be valuable as a "driver > correlation" for the structural alignability analyses we will > eventually be doing. > > Table I.
The overall regression of ln[c/u] on ln[c/L] is different in the two sets (random vs actual), but none of the differences relative to e are reliable.