Discussion: View Thread

This message has been cross posted to the following eGroups: New York City Metro Area Chapter and Statistical Consulting Section .
-------------------------------------------
Has anyone had experience in social network analysis (SNA) and mapping?  We are trying to analyze and describe how the physicians in our large health system "share" patients.  We have data on how many patients have visited buth Dr A and Dr B; likewise for A and C, B and C, etc.)  We believe this may fit into the SNA framework.

Any thoughts on software, education/training would be appreciated.

Thanks,
Marty

-------------------------------------------
Martin L Lesser, PhD, EMT-CC
Director and Investigator,
   Biostatistics Unit,
   Feinstein Institute for Medical Research
Professor, Dep't of Molecular Medicine &
     Dep't of Population Health,
   Hofstra North Shore-LIJ School of
      Medicine
Manhasset, NY 11030
516 562 0300
-------------------------------------------

Network analysis does seem appropriate. If you're an R user -- or even perhaps if you aren't -- you should have a look at the extensive resources put together by the Statnet group: http://www.statnet.org/ If you don't need the advanced statistical modeling capabilities of the Statnet approach, Ucinet and Pajek are good programs for visualizing networks and computing descriptive statistics. For general learning about these methods there's scads of literature, starting with classic studies quite analogous to your application (e.g. networks of membership in different social organizations).

-------------------------------------------
Andrew Taylor
Univ of Hawaii At Manoa
-------------------------------------------

Dear Marty,

Since you're interested in how your physicians are interacting, the nodes or vertices of your network could be individual physicians and the ties or links between them could be "shared patients". The strength of the tie between Doctors A and B could be the number of patients who have visited both A and B. In this case you would have an "undirected" graph because the tie/link between A and B has no direction. If instead you were interested in how social network analysis informed how doctors refer patients to other doctors, you would have a "directed" graph where A refers a patient to B and the direction is A to B.

I also recommend the various R packages for social network analysis.

-------------------------------------------
Eric Vance
LISA (Virginia Tech's Laboratory for Interdisciplinary Statistical Analysis)
Director and Assistant Research Professor
Blacksburg VA, United States
-------------------------------------------