Using social network analysis in evaluating community-based programs: Some experiences and thoughts. Dr Gretchen Ennis Lecturer, Social Work & Community Studies School of Health
Seminar Overview What is social network analysis (SNA)? What is a network? Whole networks and Ego networks Examples from practice Thoughts.
What is Social Network Analysis? It s a lot of things - A different paradigm of research (relational) The methodical study of social networks A social network is structure made up of actors (eg. people, groups, organisations, events) and the ties between them A way of looking at the structure of things (an organisation, a group, a sector) Networks are often visualised using a social network diagram
A network can be large, small or anything in between It is made up of actors and their ties to one another It is all about relationships.
Quite a large one Arts organisation Facebook Network 727 actor 15503 ties 42 ties per actor (average degree) 14 components 2.3 hops
A small network G D C B E A F H Professional advice among work colleagues 8 employees. 11 advice relationships Actor A = influential as 3 people come to her for advice. Actor B = also influential as actor A seeks her advice. Actors F & B go to each other (the only reciprocal tie) Actor C = only seeks advice from others BDEF is an interesting little referral chain.
the diagrams are great But they are only network visualizations Aspects of SNA typically involve : Size (how big or small?) Content (what does it comprise of, who are the actors, what are the relationships?) Purpose (what function does it serve?) Density (how many ties are in the network as a ratio of how many their could be) Degree/Average Degree (number of ties each actor has & average of actors across the whole network) Cliques, clusters or subgroups Network stars (who is central to the network?) Paths, bridges and links (which actors links parts of the network together, what direction do links go in?)
Two main approaches to SNA Whole network: when you can pre- identify every actor in the network and you want to determine the ties between them. Generally a census sampling method. Ego network: This is when you start with ego (a person, a group, an organisation etc.) and explore outwards by asking them about their alters.
Whole network Data collection may involve ethnographic study of a defined group, eg. Classroom, organisation, small community, family. survey where people in the network are asked to identify who they have particular kinds of relationships with.» Eg. who do you go to for advice? (freecall) or» put a tick next to everyone you go to for advice (Roster)
Example of a whole network Kirke (1996) examined peer networks and their influence on teenagers drug and alcohol choices. She attempted to interview every single person aged 14-18 in a town of approximately 2500 in Dublin. (N = 267). Peer networks, therefore, contribute to the similarity of adolescents substance use in a profound manner by providing a pattern of peer ties in which peer influence can flourish
Ego network Data collection generally involves interviews, surveys (can use observation) Generally we ask one or more relational questions. (examples on next slide). You also ask about the relationships between the alters. Sometimes you also interview the alters and ask them the same set of questions. However doing this you may not know where it will end, so you need to draw a line based on a predetermined theoretical or other rationale.
Some questions to elicit ego networks Behavioural: Who do you drive to work with? Who do you play sport with? Who waters your plants? Cognitive (perceptual): who do you consider a friend, who do you trust with personal information?, who would you like to work with? Who do you go to for advice? Social/Collective: what groups do you belong to? What events have you attended? Where do you work? Are you a citizen of..?
Example of an ego-network Mark & Harris ( 2012) looked at Stanford Uni roommates race and the racial composition of white college students ego networks. 195 white college freshmen interviewed in 2002 about their own campus relationships over the past 6 months and also when they were at high school. (195 separate ego networks) They found the race of student s roommate affects racial composition of student s ego network. Assignment to an Asian roommate increases number of Asian friends - Parallel effects observed for black, Hispanic, multiracial, and white roommates.
My Experiences: How I ve used it so far PhD (pre and post networks of people linked to neighborhood group, and affiliations between organization's involved in projects) Evaluation a Youth Arts partnership programs (mapping links created between organizations through one partnership program). We also had a play with mapping a range of different networks to see what worked well. Evaluation of a community collective impact program SNA part of a developmental evaluation approach (mapping the changing links between staff in the main organization and staff in the partner organizations) Evaluation of a multi-site education/community development program the SNA is one element of a developmental evaluation approach, looking at how the program engages with the communities involved.
Exploring the networking value of a Partnership program Question: How has the Study organizations participation in one partnership project helped to develop further partnerships? Data Collection: mapping interview with study org and lead organization at outset and end of original project. Actors (dots) = organizations Relationship (line) = working partnership has worked with on a project
Exploring the Value of A Partnership Program Pre Study Org has a degree of 1 (1 connection to another org) 1 project Average degree (1.714) Most connected actor is lead org (6 ties) Post (1 year later) Study Org has a degree of 8 (8 connections to other orgs) 3 projects Average degree (5.4) Most connected actors: Lead Org 7, Study Org 9 Range between 2 9 connections per org.
Post (1 year later) Study Org has a degree of 8 (8 connections to other orgs) 3 projects Average degree (5.4) Most connected actors: Lead Org 7, Study Org 9 Range between 2 9 connections per org.
Exploring group cohesion using mock up* diagrams Question: How does participation in a year long arts project impact the connections between participants? Data Collection: Roster survey (all names of all other participants) and question about their level of knows and trusts for each person on roster. A whole network pre and post workshop Actors: Workshop Participants (teachers, tutors, students) Tie : knows and trusts 1 not at all 2 - a little 3 very well * We have not done this yet we went through this exercise to see if it would be a useful analysis.
Exploring group cohesion mock-up diagram Actors: Purple: Teachers Red: Tutors Blue: Students Relationship - Knows & Trusts: Green not at all (level 1) Blue a little (level 2) Brown - very well (level 3) (also line thickness)
From the mock-up we determined we could analyse the following things that would help us understand cohesion : Number (degree) of ties for each actor Average degree of the network Number of ties of different strengths (trust in network) Central actors (students, teachers, tutors Other things (eg. cliques, bridges) We could analyse any differences between these calculations pre and post.
Different layouts of same diagram (but calculations remain the same) Yifan Hu. Counter Clockwise Rotate
Some thoughts.. Software (Pajek, UCInet, Gephi). If you are brand new to it and want to work with SNA do a short course! SNA works really well when evaluation questions are about relationships, pathways, referrals and partnerships. SNA can make visible the invisible webs of relationships that have big impacts on people, groups, programs, organizations and communities. In evaluation, I think it should be used with other methods.
Some thoughts. A lot of thinking about how you will collect your data is needed (mock ups/trials are important) How hard it is to find the right relationship to map Finding the right layout How sometimes the SNA doesn t show much The more I do the more I understand how critical theory is to interpretation.
Some relevant theories Systems theories Social capital theories Actor network theory
Social network theories linking structure and culture Ann Mische (2011) outlines four different approaches : Networks as conduits for culture Networks as pipelines through which ideas, attitudes, and innovations flow. Networks as shaping culture (or vice versa) Clusters as incubators of culture, position as generating categorical identities, bridges as a source of cultural connection. How tastes, values, morals create relational affinities that shape network structure.
Social network theories linking structure and culture Networks as cultural forms Culture organized into networks of forms, concepts, categories, practices etc. Eg. A concept map. Networks as culture via interaction Networks are constituted by cultural processes of communicative interaction. Culture is created by the relationships between people and ideas.
References Carolan, B.V. (2014). Social Network Analysis and Education, Theories, Methods and Application. Thousand Oaks CA:Sage Giuffre, K (2013) Communities and networks: Using social network analysis to rethink urban and communities studies. Polity Press: Cambridge. Granovetter, M. (1973). The strength of weak ties. The American Journal of Sociology.78 (6) Granovetter, M (1983) The strength of weak ties: A network theory revisited. Sociological Theory. Vol 1. Chapter 7, pages 201-233. Kirke, D. (1996) Collecting peer data and delineating peer networks in a complete network. Social Networks 18 (4) 333-346.
References Knoke and Yang (2008). Introduction to social network analysis. Thousand Oaks CA:Sage Kadushin, C. (2012). Understanding Social Networks: Theories, Concepts and Findings. New York: Oxford University Press Mark & Harris (2012) Roommate s race and the racial composition of white college students ego networks. Social Science Research. 41 (2) 331-342. Milgram, S. (1967) The small world problem. Psychology. I. Today 61-67 Travers & Milgram (1969) An experimental study of the small world problem. Sociometry 32, 425-443. Mische, A. (2011). Relational Sociology, Culture, and Agency. Chapter 7, The Sage Handbook of Social Network Analysis, (J.Scott & P.Carrington eds) pp 80-98. London: Sage. Wasserman & Faust (1994) Social network analysis: Methods and applications. Cambridge University Press: New York White, Boorman, & Breiger (1976) Social structure from multiple networks. I. Blockmodels of roles and positions. American Journal of Sociology 81, 730-80.