Doing network analysis in Stata seems to become more and more popular. In October, we saw the release of centpow by Zachary Neal, which computes various centrality measures. Last week I learned of a new Stata package called networks, developed by Hirotaka Miura at the Federal Reserve Bank of San Francisco. Networks computes betweenness centrality, distances, and clustering, and lets you create adjacency matrices and node lists from edgelists that are stored as Stata variables. I played around with it and it seems to work really well and reasonably fast, even on the 1000-node network that I threw at it. Actually using the results takes some work, because all results are returned as Stata and Mata matrices, and/or csv-files. Check out the Stata Graph Library website for downloads and details.
On a more general note, even though I am exited to see SNA in Stata gaining momentum, a remaining problem is that there is not yet a clear standard for the treatment of network data. networks, stata2pajek, and neplot assume edge- or arc lists stored as dta files, but centpow uses a comma-delimited text file, while pajek2stata returns as a list of nodes in a Stata variable and the network in a square Mata matrix. Who comes up with the one format to rule them all?