Simulate binary correlated data

Dear R experts,

I am trying to simulate correlated binary data and have stumbled upon the
following problem:

With the help of "binarySimCLF" or "mvpBinaryEp" I have been able to
simulate correlating binary vectors given certain mean values and a desired
correlation. My problem is that these procedures do not allow you to specify
the exact vector for which you want to generate a correlated vector. Is
there anyway to do this?

Maybe I can clarify my question by explaining what my goal is:
I want to generate one Binary Vector (A), generate a correlated binary
Vector (B), then generate a third binary Vector (C) that is correlated to B
so that I can then see the occuring correlations between A and C.

IIRC, knowing the correlation between A & B and the correlation
between B & C is not enough to uniquely specify the correlation
between A & C (or perhaps even to bound it). Therefore i think your
question is ill-defined. Though I might be wrong in the specific case
of binary variates...

Cheers,
MW

Thank you in advance,
Marbles



--
View this message in context: http://r.789695.n4.nabble.com/Simulate-binary-correlated-data-tp4660366.html
Sent from the R help mailing list archive at Nabble.com.

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Thank you for your response!

I know that it is not enough to uniquely specify the correlation. That is
why i would like to simulate it so that i can see how the resulting
correlations between A and C are distributed in dependence of the
correlations between A&B and B&C.



--
View this message in context: http://r.789695.n4.nabble.com/Simulate-binary-correlated-data-tp4660366p4660380.html
Sent from the R help mailing list archive at Nabble.com.

Btw: I have done this for non-binary variables already. If anyone is
interested, it looks like this.
I am pretty new to R, so excuse the potentially unelegant code.

# Implement library 
library(ecodist)

# Prepare ACOutcome vector
ACOutcomes = c()

# Set desired variablelength & Number of simulations
VarLength = 1000
NSim = 10000

# Set desired range for correlations
rangeAB = c(0.0, 1.0)
rangeBC = c(0.0, 1.0)

# Start n simulation runs
n = 0
while(n < NSim) {n=n+1;

# Set desired correlations between Variables
DesCorAB = runif(1, rangeAB[1], rangeAB[2])
DesCorBC = runif(1, rangeBC[1], rangeBC[2]) 

# Simulate A and B with desired correlation between them
DatasetAB = corgen(len=VarLength, r=DesCorAB, epsilon=0.00)
A = DatasetAB$x
B = DatasetAB$y

# Option of saving correlation between A & B
cor(A, B, method = "pearson")

# Simulate C with desired correlation to B
C = corgen(x=B, r=DesCorBC, epsilon = 0.00)$y

# Calculate correlation between A & C
corAC = cor(A, C, method = "pearson")

# Save correlation AC into vector
ACOutcomes = append(ACOutcomes, corAC)

;}

# Show results: Mean, Minimum, Maximum
hist(ACOutcomes, breaks = 100)
mean(ACOutcomes)
min(ACOutcomes)
max(ACOutcomes)



--
View this message in context: http://r.789695.n4.nabble.com/Simulate-binary-correlated-data-tp4660366p4660382.html
Sent from the R help mailing list archive at Nabble.com.