nwergm
nwergm fits exponential-family random graph models to a network, after the R package ergm (statnet): maximum pseudolikelihood where the model is dyad-independent and Monte Carlo maximum likelihood otherwise.
nwergm is not part of the core nwcommands. It is a package of its own, released under the
GNU General Public License, version 3, and it needs the core installed first.
| License | GPL-3 |
| Repository | github.com/thomas-u-grund/nwergm |
| Install | nwinstall, modelpkg(nwergm) (after the core; nwinstall, everything installs the core and the four model packages) |
| Version | nwergm version; release history on GitHub |
Without nwinstall |
net install nwergm, from(https://raw.githubusercontent.com/thomas-u-grund/nwergm/main/) replace |
nwergm version shows the installed release and nwversion lists all model packages. Updating the
core never overwrites the package; to update the package, repeat the nwinstall line. Inside Stata, help nwergm gives the complete syntax, options and examples.
What it offers:
- a large set of terms: covariate, homophily and mixing terms, degree and star terms, geometrically weighted terms (fixed and curved), directed and two-mode (bipartite) terms
- missing dyads are handled as in R, constraints, offsets,
predict, and simulation from a fit - postestimation:
estat gof,estat mcmcdiag,estat ic
A first model:
. nwergm flomarriage, edges nodecov(wealth)
Section: nwergm. First tutorial: Tutorial 1: A first ERGM. nwergm follows the R package ergm (statnet) and is not a replacement for it.