On Wed, Aug 1, 2012 at 4:34 PM, Xu Jun wrote:
> Thanks Michael. Now I switched my approach after doing some google.
> Following are my new codes:
>
> ###
> library(foreign)
> readin <- read.dta("ordfile.dta", convert.factors=FALSE)
> myvars <- c("
I am not that proficient in R. I found some codes on web using those
indicator variables to sum up log likelihood. I block out bols in the
codes, but I also tried using them as start value for my estimation of
ologit. Didn't work. Thanks for your suggestion.
Jun
On Wed, Aug 1, 2012 at 5:44 PM, Be
Disclaimer: I have not followed this thread at all, but only wish to note:
1) Indicator variables are (almost?) never needed in R -- that you are
fooling with them suggests that there is probably a better approach.
2) Your bols is just least regression, no? -- If so, there are far
better ways to
Thanks Michael. Now I switched my approach after doing some google.
Following are my new codes:
###
library(foreign)
readin <- read.dta("ordfile.dta", convert.factors=FALSE)
myvars <- c("depvar", "x1", "x2", "x3")
mydta <- readin[myvars]
# remov
On Tue, Jul 31, 2012 at 7:57 PM, Xu Jun wrote:
> Dear R listers,
>
> I am learning the MLE utility optim() in R to program ordered logit
> models just as an exercise. See below I have three independent
> variables, x1, x2, and x3. Y is coded as ordinal from 1 to 4. Y is not
> yet a factor variable
Dear R listers,
I am learning the MLE utility optim() in R to program ordered logit
models just as an exercise. See below I have three independent
variables, x1, x2, and x3. Y is coded as ordinal from 1 to 4. Y is not
yet a factor variable here. The ordered logit model satisfies the
parallel regre
6 matches
Mail list logo
