social network analysis in public health

38
Social Network Analysis in Public Health Reza Yousefi Nooraie SAPHIR webinar, Nov 2012 1

Upload: blenda

Post on 15-Feb-2016

25 views

Category:

Documents


0 download

DESCRIPTION

Social Network Analysis in Public Health. Reza Yousefi Nooraie SAPHIR webinar, Nov 2012. Cool question 1. If your close friend becomes obese, your chance of becoming obese… a) will increase by 20% b) will increase by 70% c) will increase by 170% - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Social Network Analysis  in Public Health

1

Social Network Analysis in Public Health

Reza Yousefi Nooraie

SAPHIR webinar, Nov 2012

Page 2: Social Network Analysis  in Public Health

2

Cool question 1

• If your close friend becomes obese, your chance of becoming obese…

• a) will increase by 20%• b) will increase by 70%• c) will increase by 170%• d) doesn’t matter. I know what I’m eatin’!

Page 3: Social Network Analysis  in Public Health

3

• If your close friend becomes obese, your chance of becoming obese…

• a) will increase by 20%• b) will increase by 70%• c) will increase by 170%• d) doesn’t matter. I know what I’m eatin’!

Christakis N, Fowler J. The Spread of Obesity in a Large Social Network over 32 Years. N Engl J Med 2007;357:370-9

Page 4: Social Network Analysis  in Public Health

4

Networks• Networks consist of actors connected to one another

by relations • Social Network Analysis: a perspective to analyze

social relationships

• actors persons

groupsorganisationscountries

• relationsinformal

advice, trust, respect,information exchange

formalexchange of money,information exchange

multiplex

Page 5: Social Network Analysis  in Public Health

5

Social Network Analysis

• A ‘relational’ thinking in social sciences

• All social entities and concepts, e.g. power, freedom, and society, are redefined as the functions of the dynamic relationships

• Relations as the units of analysis

Page 6: Social Network Analysis  in Public Health

6

Georg Simmel (1858-1918)

• Precursor of structuralism in social sciences• Introduced dyads, triads, distance, and

network size

Page 7: Social Network Analysis  in Public Health

7

Jacob Moreno (1889-1974)

• The founder of sociogram and sociometry

Page 8: Social Network Analysis  in Public Health

8

Social Capital Coleman, Katz, Menzel (1957)

• The time to adoption of a newly developed tetracycline by physicians

• to whom they turned to for professional advice, with whom they discussed, and with whom they socialized

• the position in the network predicted early adoption more than personal characteristics.

• Physicians who were considered by more peers as advisors, discussion partners and friends were more likely to use the new drug earlier

Page 9: Social Network Analysis  in Public Health

9

Milgram (1969)Small World phenomenon

NE

MA

Six degrees of separation

Page 10: Social Network Analysis  in Public Health

(Granovetter, 1973)

• the weak ties which bridge unconnected clusters are especially important

– provide access to novel and heterogeneous resources– more likely to adopt innovations/ less bound to the

group norms

• You are more likely to hear about a job from an acquaintance than a close friend

10

The strength of weak ties

Page 11: Social Network Analysis  in Public Health

11

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

Resources flow through network

the pattern of interconnection

generates outcomes

Page 12: Social Network Analysis  in Public Health

12

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

How people benefit by connectivity

Why some people are more similar

Page 13: Social Network Analysis  in Public Health

13

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

Connectivity leads in more access to resources: social relationships and health outcomessocial capital and social support

Page 14: Social Network Analysis  in Public Health

14

Social Capital, Income Inequality,and Mortality (Kawachi et al., 1997)

Page 15: Social Network Analysis  in Public Health

15

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

transmission of traits: the patterns of disease flow (HIV, STD, obesity)diffusion of knowledge and innovation

Page 16: Social Network Analysis  in Public Health

16

The Spread of Obesity in a Large Social Network over 32 Years (Christakis & Fowler, 2007)

• social network of 12,067 people assessed repeatedly from 1971 to 2003 as part of the Framingham Heart Study.

• longitudinal GEE model

• whether weight gain in one person was associated with weight gain in his or her friends, siblings, spouse, and neighbors.

Page 17: Social Network Analysis  in Public Health

Theoretical framework

•Social influence/induction

• Social selection/homophily

• Common context

Page 18: Social Network Analysis  in Public Health

18

Christakis & Fowler, 2007

Page 19: Social Network Analysis  in Public Health

19

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

Location is power: Transactional knowledge, inter-organizational partnership

Page 20: Social Network Analysis  in Public Health

20

Partnerships among Canadian Agencies Serving Women with Substance Abuse (Niccoles, Yousefi-Nooraie, et al.)

OntarioBritish Columbia

Alberta

Prince Edward Island

Saskatchewan

Manitoba

Agency A

Agency BAgency C

Agency D

responsiveness and trustworthiness: sending referralsfriendliness: joint programming and consultation.

Page 21: Social Network Analysis  in Public Health

21

Network theories

• Steve Borgatti (2011) perspectives

models Social capitalSocial homogeneity

Network flow capitalization contagion

Network architecture coordination adaptation

Position shapes attitudes and behaviors: organizational isomorphism, etc

Page 22: Social Network Analysis  in Public Health

23

Cool question 2

• In adolescents, who is more likely to be influenced to smoke, if their friends become smoker?

• a) girls

• b) boys

Page 23: Social Network Analysis  in Public Health

24

• In adolescents, who is more likely to be influenced to smoke, if their friends become smoker?

• a) girls

• b) boys

Mercken, L., et al. (2010). Smoking based selection and influence in gender segregated friendship networks: a ‐ ‐social network analysis of adolescent smoking. Addiction, 105(7), 1280-1289.

Page 24: Social Network Analysis  in Public Health

25

Design and analysis of SNA studies

Page 25: Social Network Analysis  in Public Health

26

Types of networks

• egocentric or personal networks– relations defined from focal individuals

• compare relational structures of actors

• sociocentric or whole networks– relations linking members of a single,

bounded population

• examine internal structures and positioning of actors within one network

Page 26: Social Network Analysis  in Public Health

27

Page 27: Social Network Analysis  in Public Health

28

Data collection• Questionnaires– Name generators

• Roster / Choose from a list• Free recall

– Name interpreters• Rate the frequency, quality, … of the connection

• Interviews• Observation• Recordings– Documents– Electronic logs

Page 28: Social Network Analysis  in Public Health

29

Basic measures–Overall shape•Density: proportion of available ties to all possible•Centralization: resembling a star network

–Central actors•Degree: the number of ties•Betweenness: the mediatory role•Closeness: accessibility and distance

–Subgroups•Cliques: all connected to each other•Blocks: more connected with each other than outside

Page 29: Social Network Analysis  in Public Health

30

DDRC EMRCDensity: 41% 37%Centralization: 16% 28%

Association between co-authorship network and scientific productivity (Yousefi Nooraie, 2008)

Page 30: Social Network Analysis  in Public Health

Degree centrality• the number of connections any actor has.

• in-degree: the number of connections from other to him/her

31

A B

C

DIn-degree of actor A: 1

Page 31: Social Network Analysis  in Public Health

Information seeking for making evidence-informed decisions (Yousefi Nooraie, 2012)

32

Division 1

Division 2

Division 5

Division 2

Division 4

The Office of MOH

Nodes are sized by indegrees

Page 32: Social Network Analysis  in Public Health

Betweenness centrality

• the extent that an actor appears between the other actors’ connections in the network

33

A B

C

Dbetweenness of actor A: 2

Page 33: Social Network Analysis  in Public Health

34

Division 1

Division 5

Division 2

Division 4

The Office of MOH

11

Division 2

Information seeking for making evidence-informed decisions (Yousefi Nooraie, 2012)

Nodes are sized by betweenness

Page 34: Social Network Analysis  in Public Health

35

Sub-graphs

• Clusters based on attributes

• Cliques

• Blocks

Page 35: Social Network Analysis  in Public Health

36

Stochastic models

• The problem of the dependence of observations

• Exponential random graph modeling (ERGM)

• the effect of different structural, node-level, and dyadic factors on the formation of ties

Page 36: Social Network Analysis  in Public Health

Stochastic models

• Dynamic actor-based modeling

• How the outcome variable co-evolves with the longitudinal evolution of structural, node-level, and dyadic variables

• for any point in time, the current state of the network determines probabilistically its further evolution

Page 37: Social Network Analysis  in Public Health

39

Smoking-based selection and influence in gender-segregated friendship networks (Mercken, et al., 2010)

• Longitudinal design with four measurements.

• A total of 1163 adolescents in 9 junior high schools in Finland.

• Smoking behaviour of adolescents, parents, siblings and friendship ties.

Page 38: Social Network Analysis  in Public Health

40

Mercken, et al., 2010

• Smoking-based selection of friends was found in males and females

• Social influence only in females

• Implication: prevention campaigns targeting resisting peer pressure may be more effective in girls than boys