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

st: RE: RE: how to adjust for covariates?

From   "Nick Cox" <[email protected]>
To   <[email protected]>
Subject   st: RE: RE: how to adjust for covariates?
Date   Mon, 11 Jan 2010 17:57:38 -0000

To strengthen Paul's point: -logistic- is going to treat -ethnics-
literally, meaning numerically. That's more than treating it as ordinal.

[email protected] 

Visintainer PhD, Paul

In your first syntax:

xi: logistic <DV> i.snp sex ethnics

The model assumes that "ethnics" is an ordinal value.  The coefficient
is giving you the change in the dependent variable, DV, for each unit
change in ethnics.  This doesn't make any sense because "ethnics" is a
4-level categorical variable (nominal), not an ordinal variable.  Treat
your ethnics variable the same way you treated your SNP variable.  

Your second syntax is the correct approach:

xi: logistic <DV> i.snp sex i.ethnics
[email protected]

I'm doing a SNP analysis using the xi:logistic function.
To exclude possible confounder, I'm adjusting my calculations for
different variables, e.g. sex and ethnics.
My study population consists of 4 different ethnicities, coded as 1 2 3
4. sex is coded 1 2. SNPs are coded 1 2 3 (e.g. CC/CT/TT).
I don't understand the following:

When I calculate:

xi:logistic i.snp sex ethnics

i get different results to

xi:logistic i.snp sex i.ethnics

Why is that, and what is the correct way to adjust for ethnics, using
i.ethnics or just ethnics?

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index