Statalist


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

st: RE: Nature of the covariate


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Nature of the covariate
Date   Tue, 8 Sep 2009 13:38:53 +0100

I think the word "covariate" has morphed in meaning. 

The original strict sense I take to be a continuous predictor, included in an analysis of variance, making it an analysis of covariance. 

An increasingly common use is that covariate is yet another word for any independent or predictor variable. That easily could be categorical, as you say. 

The latter terminology is not universal, but I'd regard it as common e.g. in any discussion influenced by generalised linear models. For some reason I associate John Nelder and similar British statisticians with this kind of usage. 

The existence of these two senses, narrow and wide, in overlapping literatures may lie behind your confusion. 

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

Muhammad Riaz


I have a bit concerns about some statistical books which says that the covariate should be a continuous variable. But in practice and also in the assumptions of regression analyses I have never seen that a covariate must be continuous. I have seen many model keeping sex as a covariate, now sex is categorical not a continuous. Then I don’t understand why some books define covariate as a continuous variable. Although I believe that covariate is a variable which does have significant relationships with the dependent and independent variables.

I will really appreciate to discuss this further and explain something which I am missing here.

*
*   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