Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: lambda--p-value


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: lambda--p-value
Date   Mon, 8 Dec 2008 19:34:56 -0000

-lambda- on SSC (please explain where user-written stuff you refer to
comes from) was originally written, I think, because someone on the list
asked how to calculate the beast. As I recall, a P-value calculation is
problematic as well as messy, as the applicability of the theory to
small samples is a matter of faith. Hence getting a bootstrap confidence
interval and see if it straddles zero is surely the way to go. That
requires no extra programs. 

The mention of SPSS is not decisive here. Measures of association like
this proliferate in the literature like mushrooms in the dark and
conceal as much as they reveal. StataCorp's prejudice has long been that
it is better to think in terms of modelling commands. 

Nick 
n.j.cox@durham.ac.uk 

John Antonakis

Does anyone have code for calculating lambda (nominal to nomial 
association) as well as the associated approximate p-value. I know of 
the lambda ado file (by Nick Cox); however, is there anything else out 
there?

It is strange that STATA doesn't have this measure in the tab 
function--this is something that is "canned" in SPSS.


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index