CRAN release 1.3.4
This is a substantial update for modelling of clustered time-to-event data.
New features:
- Added -log link (for additive hazards models) and Aranda-Ordaz link
- Added Z design matrix for random effects (e.g. random slopes). The current implementation only includes univariate random effects.
- Added both adaptive and non-adaptive quadrature for normal random effects
- Added robust standard errors for clustered data
- Experimental support for clustered data with left-truncation.
Bug fixes include:
- Can now include parametric smoothers in the formula
C++:
- Refactored stpm2 now includes the link as a data member, rather than using a template parameter and inheritance. This decreased the binary size by about a half.