Estimation of twostage model with cluster truncation in bivariate situation

Estimation of twostage model with cluster truncation in bivariate situation

twin.clustertrunc(
  survformula,
  data = parent.frame(),
  theta.des = NULL,
  clusters = NULL,
  var.link = 1,
  Nit = 10,
  final.fitting = FALSE,
  ...
)

Arguments

survformula

Formula with survival model aalen or cox.aalen, some limitiation on model specification due to call of fast.reshape (so for example interactions and * and : do not work here, expand prior to call)

data

Data frame

theta.des

design for dependence parameters in two-stage model

clusters

clustering variable for twins

var.link

exp link for theta

Nit

number of iteration

final.fitting

TRUE to do final estimation with SE and ... arguments for marginal models

...

Additional arguments to lower level functions

Author

Thomas Scheike

Examples

library("timereg")
data(diabetes)
v <- diabetes$time*runif(nrow(diabetes))*rbinom(nrow(diabetes),1,0.5)
diabetes$v <- v

aout <- twin.clustertrunc(Surv(v,time,status)~1+treat+adult,
     data=diabetes,clusters="id")
aout$two        ## twostage output
#> 
#> Dependence parameter for Clayton-Oakes model
#> Variance of Gamma distributed random effects 
#> With log-link 
#> $estimates
#>              log-Coef.        SE         z    P-val Kendall tau         SE
#> dependence1 -0.4125038 0.3719616 -1.108996 0.267432   0.2486826 0.06949713
#> 
#> $vargam
#>             Estimate Std.Err   2.5% 97.5%  P-value
#> dependence1    0.662  0.2462 0.1794 1.145 0.007178
#> 
#> $type
#> [1] "clayton.oakes"
#> 
#> attr(,"class")
#> [1] "summary.mets.twostage"
par(mfrow=c(2,2))
plot(aout$marg) ## marginal model output

out <- twin.clustertrunc(Surv(v,time,status)~1+prop(treat)+prop(adult),
     data=diabetes,clusters="id")
out$two        ## twostage output
#> 
#> Dependence parameter for Clayton-Oakes model
#> Variance of Gamma distributed random effects 
#> With log-link 
#> $estimates
#>             log-Coef.        SE       z     P-val Kendall tau         SE
#> dependence1 -0.391893 0.3637734 -1.0773 0.2813464   0.2525534 0.06866958
#> 
#> $vargam
#>             Estimate Std.Err  2.5% 97.5%  P-value
#> dependence1   0.6758  0.2458 0.194 1.158 0.005978
#> 
#> $type
#> [1] "clayton.oakes"
#> 
#> attr(,"class")
#> [1] "summary.mets.twostage"
plot(out$marg) ## marginal model output