Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | lan zhang <cat1984@126.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: gologit2 |
Date | Wed, 28 Aug 2013 10:58:29 -0400 |
On Aug 28, 2013, at 10:41 AM, Maarten Buis <maartenlbuis@gmail.com> wrote: > On Wed, Aug 28, 2013 at 3:51 PM, lan zhang wrote: >> i want to conduct a gologit2 model, however, the correlations between my independent variables are very high, almost 0.9. Is it still possible for me to use the gologit2 model? > > If the correlation is that high you first need to ask why that might be. > > For example, it could be these variables are just different ways of > measuring the same thing. In that case you should not add them all to > the model. Say, we think that occupation and education are different > ways of estimating socioeconomic status. If we add both to our model, > we would get effects of one approximation of status while adjusting > for another approximation of that same status. That would not make any > sense. However, this is only true if we think that these are both > measuring the same concept. If we think that there are different > dimensions of socioeconomic status, say an economic dimension captured > by occupation and a cultural dimension captured by education, adding > both to your model would again be perfectly fine and actually > desirable. So the real question remains: why is that correlation so > high? > > -- Maarten > > --------------------------------- > Maarten L. Buis > WZB > Reichpietschufer 50 > 10785 Berlin > Germany > > http://www.maartenbuis.nl > --------------------------------- > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/faqs/resources/statalist-faq/ > * http://www.ats.ucla.edu/stat/stata/ * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/