[R] Simulation confidence interval

Scott Raynaud scott.raynaud at yahoo.com
Wed Feb 1 16:45:57 CET 2012


The follwing is a code snippet from a power simulation 
program that I'm using:
 
estbeta<-fixef(fitmodel)
 sdebeta<-sqrt(diag(vcov(fitmodel)))
  for(l in 1:betasize)
  {  
   cibeta<-estbeta[l]-sgnbeta[l]*z1score*sdebeta[l]
    if(beta[l]*cibeta>0)              powaprox[[l]]<-powaprox[[l]]+1
      sdepower[l,iter]<-as.numeric(sdebeta[l])
  } 
 
Estbeta recovers the fixed effects from a model fitted using lmer.
Beta is defined elsewhere and is a user specified input
that relates the data generated in the simulation to an oucome.  
So, it seems pretty clear that the third line from the bottom is
a clever test of whether the confidence interval traps 0.  My 
question is why use beta[l]*cibeta>0 rather than 
estbeta[l]*cibeta>0.  Is that because in the long run the model 
parameter etimates tend toward the betas specified by the user?  
In other words, what really matters is the standard errors, right?



More information about the R-help mailing list