[R list package] Replication problem
Juan Felipe Riaño Rodriguez
jf.riano48 at uniandes.edu.co
Fri Apr 3 02:33:08 CEST 2015
Thank you very much! I will be wating for your email and for the update
On 2/04/2015, at 3:35 p.m., Graeme Blair <graeme.blair at columbia.edu<mailto:graeme.blair at columbia.edu>> wrote:
Thanks for your patience. This error was caused by a change in another software package list depends on, and we have now resolved the problem. I will send you a test version of the package off-list, and we will release an update to CRAN including the fix soon.
On Fri, Feb 27, 2015 at 1:52 PM Juan Felipe Riaño Rodríguez <jf.riano48 at uniandes.edu.co<mailto:jf.riano48 at uniandes.edu.co>> wrote:
Working with my data to verify ceiling and floor liars, I used your replication code as a guide and got error messages about the Log likelihood not being monotonically increasing. But then I realized that I also get this error with your own code on your data, that is simply running the replication of your paper.
In particular when trying to replicate Table 7, columns 3-6, as below, I get the error messages shown below (if I change the fit.start parameter, the error that pops out may change from object not found to non non monotonically increasing Log likelihood, but I always get an error anyways).
In short, before concluding there is problem with my data specifically, given this finding I would like to point it out and make sure there is not a more general problem with the code (perhaps something got messed up in an software update or that of package dependencies?).
Thank you very much,
> ceiling.results <- ictreg(y ~ age + college + male + south, treat = "treat", J = 3, data = affirm, method = "ml", fit.start = "ml", ceiling = TRUE, ceiling.fit = "bayesglm", ceiling.formula = ~ age + college + male + south)
Error in logistic(x.treatment.curr %*% fit.control.coef) :
object 'fit.control.coef' not found
> summary(ceiling.results, boundary.proportions = T)
Error in summary(ceiling.results, boundary.proportions = T) :
object 'ceiling.results' not found
> # Fit standard design ML model with floor effects alone
> # Replicates Columns 5-6
> floor.results <- ictreg(y ~ age + college + male + south, treat = "treat",
+ J = 3, data = affirm, method = "ml", fit.start = "glm",
+ floor = TRUE, floor.fit = "bayesglm",
+ floor.formula = ~ age + college + male + south)
Error in ictreg(y ~ age + college + male + south, treat = "treat", J = 3, :
log-likelihood is not monotonically increasing.
> summary(floor.results, boundary.proportions = T)
Error in summary(floor.results, boundary.proportions = T) :
object 'floor.results' not found
listpackage-discuss mailing list
listpackage-discuss at lists.r-forge.r-project.org<mailto:listpackage-discuss at lists.r-forge.r-project.org>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the listpackage-discuss