Am attending a talk at UCI's "Social Network Brown Bag" seminar.
Mapping Networks. George Tita, Katherine Faust.
(George gave a talk last year, if I recall correctly, on maps of social networks in Los Angeles barrios--specifically looking at how gang territory and alliances were negotiated over space. In general, he was noticing that natural barriers--like freeways and major roads--were strong demarcations of networks. Groups across these barriers were less likely to either fight or to make alliances. It occurs to me, now, that it might be interesting to look back at his data in terms of tempo--that one might look at the rhythm, or approximate frequency, of interactions in terms of distance. And look at Rick Grannis: "“The Importance of Trivial Streets: Residential Streets and Residential Segregation.” American Journal of Sociology 103: 6, 1530-1564" for a similar perspective.)
Spatial work
* Interested in variations in networks over space.
* Space matters (Tobler's first law?) in the development of social ties
* Festinger, Schachter, and Back; Butts; Huckfeldt and Baybeck.
George's field site is in Hollenbeck: heavily Latino, a little bit poor. Lots of gangs, a lot of gang-driven crime. And most families have been there for a long time.
In particular, they are interested in egocentric networks. Which makes me happy already.
Their data collection is ego-centric networks and the characteristics of the alters.. Thus, when they collected the set, they asked each person what the interactions of the alters were. ("Know each other? Don't know each other?") They also collected their addresses, and the addresses of the alters. (Take THAT, Orkut mappers.)
Fairly straight forward name generator from standard social surveys--plus the question "who are you cool with because you want to stay on their good side."
An interesting challenge: their inteview asks (among other things): Living With but Not Married; Spouse; Boyfriend/Girlfriend. In contrast to "children" and "parent." And, well, they've found that their interviewees seem to come up with--on the average--2 or 3 people: a girlfriend and a spouse, for example. They suspect this is probably a translation problem.
68% of alters live in Hollenbeck. You get up to 85% within a several mile radius of hollenbeck. (And you don't get any hits in Santa Monica or Beverly Hills.)
---
Data! They've got (initially) cool data! Social networks (that kinda look like mine) superposed over geography. Basically, using Arcview to draw networks.
* "Loca", who had moved (by about 10 blocks) about a year ago. Still refers to her old address as her neighborhood (and thinks of her gang as has community.) But interestingly, all the social support questions came up with her new neighborhood.
* There are virtually no ties between this (Mexican) community and the slightly-west bordering (Salvadoran) community.
* "Who do you avoid" is only local people (unsurprisingly)--there's no reason to avoid people who live far away.
--
And so that leaves them with LOTS of future work. Euclidean distance isn't necessraily as good a measure as travel time distance, for example.
Carter Butts just suggested the idea of mapping networks against "sociological gradients." It doesn't make sense to compare geography of New York city blocks against 150 feet in Ohio: instead, one can draw a gradient of population, or of "social impendance" (the cultural distinction between the Mexican and Salvadoran communities, for example). Apparently, he did some work on a similar topic for his dissertation. (This page is one of his working papers; ignore the inccorect author.) He suggests that for many networks, this sort of gradient can account for most of the internal network structure.
Social distance and space are heavily theorized, but are often not backed up with real data. Lots of the audience is very excited about seeing distance correlated--for example--"who do you fight?" After all, you don't fight people who are VERY far away--and you don't fight people who are immediately close.
Ok, this just led us to getting the idea of tying