By
Jacqueline Olvera
When
the Center for Health Innovation (CHI) announced its Summer Scholar Program in
2014, I jumped at the opportunity to apply. CHI made it possible
for AU faculty to enroll in a variety of summer skill building courses and
seminars offered at the Mailman School of Public Health at Columbia
University. Since I had been working on
a project on Mexican migration, I was particularly interested in the Social
Network Analysis (SNA) course. I had conducted in-depth interviews with
migrants from Tlaxcala, Mexico with funding from the Russell Sage Foundation
and was interested in finding out if network analysis would enrich my study. Based
on preliminary analysis of these interviews, the data revealed that
relationships between migrants were meaningful structures. That is, I was
finding that social ties are how migrants find housing, jobs, and information
about community resources. And equally as important, the formalized structures
of these relationships seemed to be the basis for inclusion and exclusion when
forming community. Given my interests in social relationships, I was delighted
when I found out that I would indeed be able to enroll in the SNA course.
The course I participated in provided
an excellent introduction to the conceptual and computational principles of
SNA. On our first day, we covered what Social
Network Analysis is and is not, what counts as network data, and how to collect
it. We spent time using R, a language
and platform for statistical computing and graphing in order to manipulate
network matrices and visualize network data.
Thereafter, we quickly moved on to a discussion of ego-networks and the
meaning of distance, density, and balance within an individual’s networks. By day three, we covered higher-order network
structures: the group and entire networks. All the while, we focused on
important network structural features such as equivalence, clustering,
centrality, and cohesion. For example,
in a migrant network computing centrality indices might tell us which
individual in a network is the most central or popular.
The fourth day of the course was by
the best part – we put our knowledge to work by focusing on applications of
SNA. We explored examples from
epidemiological research such as the transmission of AIDS, the structure of
adolescent romantic and sexual networks, and the dynamics of smoking in large
networks. Each of these empirically
motivated problems gave us a glimpse into how network analysis is applied. More specifically, we focused on three network
processes: Search, Diffusion, and Influence.
And, we replicated analyses of classic studies that examine how network ties
facilitate the exchange of employment information as well as the efficiency of
job search strategies in networks.
On
the final day of the course, we covered statistical models and worked through
tutorials in R. The instructor
introduced network autocorrelation and stochastic actor oriented (SIENA)
models. Since I had experience with spatial analytic techniques, the
network autocorrelation model was familiar. In a recently published paper, I
used neighborhood-level data to estimate spatial lag models that follow similar
autocorrelation properties.
After completing the course, I have
spent much time thinking more about how I might use the insights of network
analysis to ask, what network practices in the small migrant community I have
been studying might tell us something about the raising and blurring ethnic boundaries?
My hope is that by integrating social network analysis into my research I will
make new inroads into the ways in which migrant communities come to make sense
of the places they live and work in.
