lame (version 1.3.4)

nodematch: ERGM-style covariate helpers for ame() / lame()

Description

nodematch(x) returns an n x n matrix with 1 where x[i] == x[j] and 0 otherwise -- the AME analogue of ERGM's nodematch term, suitable for Xdyad as a single homophily covariate. absdiff(x) returns the absolute difference |x[i] - x[j]|, the AME analogue of ERGM's absdiff. nodefactor(x) returns the dyadic matrix of x[i] + x[j] for a numeric x (or, for a factor / character x, a named list with one matrix per level giving the dyadic co-membership count (x[i] == level) + (x[j] == level), with entries 0, 1, or 2, the ERGM nodefactor semantics).

Usage

nodematch(x, na_diag = TRUE)
absdiff(x, na_diag = TRUE)
nodefactor(x, na_diag = TRUE)

Value

An n x n numeric matrix (or, for a factor x passed to nodefactor, a named list of such matrices).

Arguments

x

a numeric, factor, or character vector of length n.

na_diag

logical: set the diagonal to NA? Defaults to TRUE.

Details

All helpers return a matrix or list of matrices that is ready to wrap into an n x n x p Xdyad array via simplify2array or array().

Examples

# \donttest{
n <- 12
grp <- sample(letters[1:3], n, replace = TRUE)
age <- rnorm(n, 40, 10)
# build a single same-group homophily covariate
Xdyad <- array(nodematch(grp), dim = c(n, n, 1),
               dimnames = list(NULL, NULL, "same_group"))
# combine homophily + age difference
Xdyad <- array(c(nodematch(grp), absdiff(age)),
               dim = c(n, n, 2),
               dimnames = list(NULL, NULL, c("same_group", "age_diff")))
# }