[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: deming regression

From   "Nick Cox" <>
To   <>
Subject   st: RE: deming regression
Date   Thu, 28 Feb 2008 13:40:08 -0000

-deming- is an unofficial command written by Yulia Marchenko (who just
happens to work for StataCorp). 

It expects to be fed the coefficients of variation, or strictly 100
times those. 
Recall that coefficient of variation is standard deviation/mean. 

The method you refer to requires quite different input and on the face
of it is different. 
However I think it would need someone to read that paper and also look
inside Yulia's program 
to see whether they are or are not equivalent. 

My experience in this territory is that there are many different
procedures all aimed at 
situations in which both variables are measured with error. What you do
has to be based 
on what you can assume about the underlying situation. The literature is
also scattered across
disciplines and full of papers claiming that they know the best
procedure and that others are 
wrong or not nearly so good. The same procedure also often appears under
quite different names in 
various places. Recently I came across a reference to "neutral
regression" for example, which sounds a 
good term, except that I still seek a precise definition.  

See also -concord- (-findit concord-) which includes a reduced major
axis (SD line) option and
includes various little tools for method comparison. 


Subramanian Swaminathan

I have a problem in giving inputs to the 'deming'
command in stata. From a 
paper by "Cornbleet, PJ and Gochman, N (1979).
Incorrect least squares 
regression coefficients in method-comparison analysis,
Clinical Chemistry 
25, 432-438", I understand that they used a parameter
called 'lambda', which 
is the ratio of the within-individual variance of the
indepdent variable and 
the within-individual variance of the dependent
variable. This parameter can 
be estimated from repeated measurements of depdent and
indepdent variables. 
My question is, do the the values of cv1 and cv2 in
stata are based on the 
values of var1 and var2? if so, how is it related to
the 'lambda' parameter 
of the above stated paper? If such repeat measurements
not available, what 
value should be used for inputting to cv1 and cv2?

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index