Hello all, I am wondering if there is a way to specify sampling weights for an ols model using sample weights. For instance, right now, my code is: fit.ex<-lm(y~x1+x2+x3+...xk,data=dataset,weights=weightvariable.) summary(fit.ex) But, there is almost no difference in the coefficients nor standard errors. I am skeptical that I am using the option correctly since many posts state that the "weights" option is for weighted least squares. Or is this the same thing? Any help is appreciated! Thanks Carlos -- View this message in context: http://r.789695.n4.nabble.com/sample-weights-in-ols-tp3510865p3510865.html Sent from the R help mailing list archive at Nabble.com.
On Tue, May 10, 2011 at 2:50 PM, jour4life <jour4life at gmail.com> wrote:> Hello all, > > I am wondering if there is a way to specify sampling weights for an ols > model using sample weights. > > For instance, right now, my code is: > > fit.ex<-lm(y~x1+x2+x3+...xk,data=dataset,weights=weightvariable.) > summary(fit.ex) > > But, there is almost no difference in the coefficients nor standard errors. > I am skeptical that I am using the option correctly since many posts state > that the "weights" option is for weighted least squares. Or is this the same > thing?You want svyglm() in the survey package. -thomas -- Thomas Lumley Professor of Biostatistics University of Auckland
I have a follow up question. When using svyglm, it does not matter that I am not using survey design and only weights? In other words, fit<-svyglm(y~x1+x2+...xk,data=dataset,weights=weightvariable) Or am I going to have to construct a survey design variable, using only the weight variable? Thanks, Carlos -- View this message in context: http://r.789695.n4.nabble.com/sample-weights-in-ols-tp3510865p3514920.html Sent from the R help mailing list archive at Nabble.com.