p2ML: MCMC estimates for the (multilevel) p2 model

In dyads: Dyadic Network Analysis

MCMC estimates for the (multilevel) p2 model

Description

Estimates the (multilevel) p2 model parameters,as described in Zijlstra, Van Duijn and Snijders (2006) <doi: 10.1027/1614-2241.2.1.42>.

Usage

p2ML(nets, sender = NULL, receiver = NULL, density =~ 1, reciprocity =~ 1, 
adapt = NULL, burnin = NULL, center = NULL, separateSigma= NULL, seed = NULL,
densVar = NULL, recVar = NULL) 

Arguments

nets	List with n dichotomous dependent directed networks.
sender	Optional matrix with a stacked actor-level sender covariate, corresponding to the n networks. Multiple sender covariates can be added as a formula object, see example below
receiver	Optional matrix with a stacked actor-level receiver covariate, corresponding to the n networks. Multiple receiver covariates can be added as a formula object
density	Optional stacked matrix with a density covariate, with dimensions similar to the n dependent networks. Multiple density covariates can be added as a formula object, see example below
reciprocity	Optional stacked matrix with a symmetric reciprocity covariate, with dimensions similar to the n dependent networks. Multiple reciprocity covariates can be added as a formula object
adapt	Optional minumum effective sample size to be reached for the model parameters before ending MCMC simulations (default is 300 for the total sample, including burn-in).
burnin	Optional specification of number of burn-in iterations (default is 2500).
center	Optional argument for centering predictors (default is TRUE).
separateSigma	Optional argument for estimating separate variance-covariance matrices for the random sender and receiver effects in the dependent networks (default is FALSE).
seed	Optional specification of random seed (delfault is 1).
densVar	Optional argument for estimating density variance at the network level instead of separate density effects (default is TRUE).
recVar	Optional argument for estimating reciprocity variance at the network level instead of separate reciprocity effects (default is TRUE).

Value

Returns a fitted model of class 2ML, to be opened with the function summary().

Author(s)

Bonne J.H. Zijlstra [e-mail]

References

Zijlstra, B. J., Van Duijn, M. A., & Snijders, T. A. (2006). The Multilevel p2 Model A random effects model for the analysis of multiple social networks. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 2(1), 42.

Examples

# create two very small networks with covariates for illustrative purposes
Y1 <- matrix(c(0,1,0,1,0,1,0,1,0,0,
               0,0,1,1,0,1,0,1,0,1,
               1,1,0,0,1,0,0,0,0,0,
               1,1,1,0,1,0,0,0,0,1,
               1,0,1,0,0,1,1,0,1,1,
               0,0,0,0,0,0,1,1,1,1,
               0,0,0,0,0,1,0,1,0,1,
               1,0,0,0,0,1,1,0,1,1,
               0,1,0,1,0,1,0,1,0,0,
               0,0,1,1,1,0,0,0,0,0), ncol=10)
Y2 <- matrix(c(0,0,1,0,1,0,0,1,0,0,
               0,0,0,0,0,0,0,1,1,0,
               0,0,0,1,0,1,0,1,0,1,
               0,0,1,0,0,0,1,1,0,0,
               1,0,0,1,0,0,1,0,0,1,
               0,0,1,0,0,0,1,1,0,0,
               0,1,0,0,1,0,0,0,0,0,
               1,0,1,0,1,1,1,0,0,1,
               0,1,0,1,0,0,0,0,0,0,
               0,1,0,1,0,0,0,1,0,0), ncol=10) 
Y <- list(Y1, Y2)                
Sa1 <- c(1,0,1,0,1,1,0,1,0,1)
Sa2 <- c(1,0,0,1,0,0,1,1,0,1)
Sa <- list(Sa1, Sa2)
Sat <- Rat <-  do.call(plyr::rbind.fill.matrix, Sa)
Sb1 <- c(0,1,1,0,1,0,1,0,1,0)
Sb2 <- c(1,0,1,0,0,1,0,1,0,1)
Sb <- list(Sb1, Sb2)
Sbt <- do.call(plyr::rbind.fill.matrix, Sb)
Da1 <- abs(matrix(rep(Sa1,10), byrow = FALSE, ncol= 10) -
              matrix(rep(Sa1,10), byrow = TRUE, ncol= 10))
Da2 <- abs(matrix(rep(Sa2,10), byrow = FALSE, ncol= 10) -
              matrix(rep(Sa2,10), byrow = TRUE, ncol= 10))
Da <- list(Da1, Da2)
Dat <- do.call(plyr::rbind.fill.matrix, Da)

# estimate p2 model for two networks
M1 <- p2ML(Y, sender= ~ Sat + Sbt,  receiver= ~ Rat, density = ~ Dat, adapt = 10, burnin = 100)
summary(M1)
# Notice: burn-in, and number of adaptive sequenses are 
# much smaller than recommended to keep computation time low.
# recommended code: 
## Not run: 
M1 <- p2ML(Y,sender= ~ Sat + Sbt,  receiver= ~ Rat, density = ~ Dat)
summary(M1)

## End(Not run)

# estimate p2 model for a single network
M2 <- p2ML(list(Y[[1]]),sender= ~ Sat[1:10,] + Sbt[1:10,],  receiver= ~ Rat[1:10,],
density = ~ Dat[1:10,], adapt = 10, burnin = 100)
summary(M2)
# Notice: burn-in, and number of adaptive sequenses are 
# much smaller than recommended to keep computation time low.
# recommended code: 
## Not run: 
M2 <- p2ML(list(Y[[1]]),sender= ~ Sat[1:10,] + Sbt[1:10,],  receiver= ~ Rat[1:10,], 
density = ~ Dat[1:10,])
summary(M2)

## End(Not run)

dyads documentation built on Sept. 18, 2026, 5:08 p.m.