[R] Calling procedures

[Steven Yen]

Dear All

Below are calls to functions to calculate bivariate and univariate 
logistic probabilities.It works for the following sample program (with 
results p1=p2 and p3=p4), but similar calls in a more elaborated program 
produced unpredicted results.

My question is whether I am doing something bad (which I should avoid) 
in my calls to mycdf2 and mycdf to obtain p2 and p3, respectively. Thank 
you.

Steven Yen

pbivlogis <- function(x,y,rho){
# *********************************************
# Bivariate logistic CDF
# *********************************************
   p<-(1+exp(-x)+exp(-y)+(1-rho)*exp(-x-y))^(-1)
return(p)
}

mycdf <- function(q,logistic=FALSE){
# *********************************************
# Univariate CDF: normal or logistic
# *********************************************
   if(!logistic){
     p<-pnorm(q)
   } else {
     p<-plogis(q)
   }
return(p)
}

mycdf2 <- function(x,y,rho,logistic=FALSE){
# *********************************************
# Calling bivariate CDF: normal or logistic
# *********************************************
   if(!logistic){
     p<-pbivnorm(x,y,rho,recycle=T)
   } else {
     p<-pbivlogis(x,y,rho)
   }
return(p)
}

set.seed(123)
x<-runif(n=5,min=-3,max=3)
y<-runif(n=5,min=-2,max=4)
rho<-0.5

p1<-pbivlogis(x,y,rho); p1
p2<-mycdf2(x,y,rho,logistic=TRUE); p2

p3<-mycdf(x,logistic=T); p3
p4<-plogis(x); p4

Results

 > set.seed(123)
 > x<-runif(n=5,min=-3,max=3)
 > y<-runif(n=5,min=-2,max=4)
 > rho<-0.5
 > p1<-pbivlogis(x,y,rho); p1
[1] 0.04937376 0.65977865 0.35821101 0.72243120 0.63881214
 > p2<-mycdf2(x,y,rho,logistic=TRUE); p2
[1] 0.04937376 0.65977865 0.35821101 0.72243120 0.63881214
 > p3<-mycdf(x,logistic=T); p3
[1] 0.2184819 0.8493908 0.3667608 0.9087199 0.9335661
 > p4<-plogis(x); p4
[1] 0.2184819 0.8493908 0.3667608 0.9087199 0.9335661
 >