```*****  To join INSNA, visit http://www.insna.org  *****

Hi Aaron,

You've rediscovered Ron Breiger's classic duality of people and groups!

Breiger, R.L. 1974. The Duality of Persons and Groups. Social Forces. 53 (2):181-190. (https://urldefense.proofpoint.com/v2/url?u=http-3A__www.jstor.org_stable_2576011&d=DwIGaQ&c=pZJPUDQ3SB9JplYbifm4nt2lEVG5pWx2KikqINpWlZM&r=uXI5O6HThk1ULkPyaT6h2Ws3RKNKSY__GQ4DuS9UHhs&m=m-OrXt6SiN-s6PQsTgj2rAE9o9Za3CY_eZlHQXDp0kM&s=0T2MuMtjuTvien-dpqrAF-AEqNnEwtB1b8rLVrJ8f8I&e=)

Stick with your first coding because it allows you to do more. From this, you can induce the second coding as well as the induced socimatrix among people with a little bit of linear algebra. What you have is an n x k incidence matrix of n people by k events.

Dunno if you're an R user, but here's a little snippet of R code that shows how:

require(igraph)
### This is an 18 x 14 matrix of women x events from Davis, Gardner and Gardner (1941)
### to get the 18 x 18 matrix of women co-attendance, do some matrix multiplication
### matrix multiplication by %*% (inner product)
### inner dimensions must match; resulting matrix has outer dimensions
### e.g., 18 x 14 %*% 14 x 18 => 18 x 18 (i.e., person-by-person)
### t(davismat) is the transpose of the davis matrix

## f2f == "face-to-face", i.e., the sociomatrix
f2f <- davismat %*% t(davismat)

### this gives you the number of women at each event (diagonal) or mutually at 2 events
e2e <- t(davismat) %*% davismat

## plot the event network -- I think this is what you want
ge2e <- simplify(ge2e)
plot(ge2e, vertex.color="lightblue")

Best of luck.

Cheers,
Jamie

--
James Holland Jones
Associate Professor of Earth System Science &
Senior Fellow, Woods Institute for the Environment

473 Via Ortega, Room 140
Stanford, CA 94305-4216

phone:	650-721-2484
fax:	650-498-5099
url:	https://urldefense.proofpoint.com/v2/url?u=https-3A__profiles.stanford.edu_james-2Djones&d=DwIGaQ&c=pZJPUDQ3SB9JplYbifm4nt2lEVG5pWx2KikqINpWlZM&r=uXI5O6HThk1ULkPyaT6h2Ws3RKNKSY__GQ4DuS9UHhs&m=m-OrXt6SiN-s6PQsTgj2rAE9o9Za3CY_eZlHQXDp0kM&s=mY31cRYCeceC8d7-AJnB0hRBfbtiiwCs1uT-iz3OKP8&e=

>
> ***** To join INSNA, visit https://urldefense.proofpoint.com/v2/url?u=http-3A__www.insna.org&d=DwIGaQ&c=pZJPUDQ3SB9JplYbifm4nt2lEVG5pWx2KikqINpWlZM&r=uXI5O6HThk1ULkPyaT6h2Ws3RKNKSY__GQ4DuS9UHhs&m=m-OrXt6SiN-s6PQsTgj2rAE9o9Za3CY_eZlHQXDp0kM&s=slqHR9u4zJ5GnLD-Ps0LuBbMf0zQP3H8Gf3BfOGhUAE&e= *****
> Hello Everyone,
>
> This may be too simple and I may be overthinking it but I was hoping to get some feedback on the best way to manage some data. I have individual records where everyone is a member of two to three groups. There is a total of about 30 groups people can be a member of, but each is a member of at least two.
>
> Now, we are not necessarily concerned about the individuals, but we need them to show the connection between the groups. We are really interested in the number of connections between the 30 groups and identifying areas of overlap and possible intervention to encourage collaboration among the silos.
>
> Currently, the data looks a little something like this:
> (Example 1)
>
> Group A
> Group B
> Group C
> Group D (…etc)
> Individual 1
> 0
> 1
> 0
> 1
> Individual 2
> 1
> 1
> 0
> 1
> Individual 3
> 0
> 0
> 1
> 0
>
> One idea someone suggested was to alter the data as (example 2)
>
> Group A
> Group B
> Group C
> Group D (…etc)
> Group A
> 1
> 1
> 1
> 1
> Group B
> 1
> 0
> 0
> 1
> Group C
> 1
> 1
> 1
> 1
>
>
>
>
>
>
> The challenge with this is that Group A could have 100-200 individuals who are a member of group A and group B.
>
> Is the data in example 1 the best way to handle it? Or is there an obvious solution I am blanking on right now?
>
> Thanks for any help-
>
> Aaron Guest, MPH, MSW, CPH
> PhD Candidate in Gerontology
> Graduate Center for Gerontology | University of Kentucky
> Multi-Disciplinary Science Building | 725 Rose Street, Room 448