Dear R users
Recently I received advice from this fine group on gee() and sample weights
One suggestion was to use geeglm()
I hope someone can help me to solve a problem that arises when converting a
code from gee to geeglm.
*Here is a code that I wrote with the original data, not weighted: *
> m1 <- gee( Bin ~ educ+agemean+ residencysize + yearx , id = rad09 ,
data = Males, subset = marp1 == 1 , + family = binomial, corstr
="unstructured" )
(Intercept) educ agemean residencysize yearx
-0.23875 -0.17931 -0.01470 -0.07418
-0.15200> se <- summary(m1)$coefficients["yearx", "Robust
S.E."]
> efrinedri <- coef(m1)["yearx"] + c(-1, 1) * se * qnorm(0.975)
> printco( y1 = summary(m1)$coefficients["yearx",
"Estimate"] , uppdown efrinedri )
[1] 0.85 ( 0.59 , 1.24 )
*
*
*Trying to convert it to geeglm() with sample weiht: *
m1 <- geeglm( Bin ~ educ+agemean+ residencysize + yearx , id = rad09 ,
data = Males, subset = marp1 == 1 , family = binomial, weights Vigtpan ,
corstr ="unstructured" )
*I get the following error message and not sure how to work on that. Any
suggestions appreciated*
Error in geese.fit(xx, yy, id, offset, soffset, w, waves = waves, zsca, :
nrow(zsca) and length(y) not match
In addition: Warning messages:
1: In eval(expr, envir, enclos) :
non-integer #successes in a binomial glm!
2: glm.fit: algorithm did not converge
3: glm.fit: fitted probabilities numerically 0 or 1 occurred
*Regards*
*Stefan Jonsson*
*
*
[[alternative HTML version deleted]]
