Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

The bild package appears to be an excellent package for serial binary responses. But it is for discrete time. I would like to specify a smooth function of time for the odds ratio connection of the current response Y with binary responses measured at earlier times, or at least a first-order Markov version of this. I believe this is called alternating logistic regression. Does anyone know of an R package that handles continuous time, i.e., measurement times can be at any follow-up time? I don't need random effects in the model.

share|improve this question
    
I haven't used it, but for alternating logistic regression, a quick search turned up this ALR package: hsph.harvard.edu/carey/vcwww_4.html –  pat Oct 22 '12 at 6:06
    
Thanks for this comment. The best I can tell from the documentation, alr is only for discrete time. –  Frank Harrell Oct 22 '12 at 13:28
add comment

1 Answer 1

The R package orth may be of some help. Here's a vignette. See also the manual.

share|improve this answer
2  
Welcome to the site, @user22299. Would you mind expanding on this w/ a few sentences to say what the package can do, & why you think it might be the answer? –  gung Mar 21 '13 at 12:58
    
The authors programmed the combinatoric calculations by generating a combinatoric dataset, making it not computationally efficient enough for large $N$. They perhaps did that to make it work in SAS IML. In addition, I can't tell if their method has the properties of a full likelihood approach, i.e., whether it is robust to non-random dropout as generalized least squares and mixed effects models are. Their method is akin to the non-full-likelihood alternating logistic regression (ALR), implying you have to use multiple imputation on top of ALR to make the missing values missing at random. –  Frank Harrell Mar 24 '13 at 19:51
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.