1 / 69

Social Network Analysis & Behavior Change Dynamics

Explore the impact of social networks on behavior change and behavior diffusion in various areas such as smoking, substance abuse, family planning, physician practices, and sexually transmitted infections.

strader
Download Presentation

Social Network Analysis & Behavior Change Dynamics

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Social Network Analysis & Behavior Change Dynamics Robert Wood Johnson Foundation Program Meeting November 16, 2006 Thomas W. Valente, PhD Associate Professor, Director MPH Program Preventive Medicine, Keck School of Medicine University of Southern California tvalente@usc.edu

  2. UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  3. Social Networks Applied • Improve use of organizational resources • Better deployment of human resources • Improve structural arrangements • Measure and develop partnerships UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  4. Social Networks Influence Behavior • Smoking (Alexander; Ennett & Baumann; Unger; etc.) • Substance abuse (Valente, Latkin, Freidman, Neaigus, etc.) • Family Planning & Fertility Regulation (Valente; Casterline; Montgomery; Watkins; Behrman; Entwisle; etc.) • Physician Practices (Lomas; Soumerai; ) • Sexually Transmitted Infections (Klovdahl; Rothenberg; Aral; Ellen; etc.) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  5. Organization of Talk • Individual level behavioral influences • Network level influences • Individual-network level linkages • Diffusion effects • Coalitions/Collaboratives • Interventions UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  6. UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  7. Knowledge Attitude & Practices UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  8. 1. Individual Level Effects • Awareness and information passes thru network contacts • Detailed knowledge and know-how gets transmitted via networks • Perceptions of norms, peer pressure flow thru networks

  9. Network Exposure = Non User = User PN Threshold=33% PN Threshold=66% PN Threshold=100% UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  10. Physicians are Influenced by Peers • Studied MD adoption of Ablation Therapy for treatment of Barrett’s Esophagus • Baseline and 1-year followup using a professional mailing list of Gastroenterologists • Measured practice and case scenario • Thought of all the possible things that could influence these behaviors. UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  11. Table. Physician Beliefs, Characteristics, and Practice Style Associated With Use of AT. UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  12. Table: Adjusted Odds Ratios for Contraceptive Use. Women in Voluntary Organizations, Yaoundé Cameroon. . UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  13. Personal Network Exposure Weighted by Indirect Ties = Non User = User PN Exposure=54% UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  14. Influences Stronger for Closer/Stronger Ties

  15. Baltimore NEP • Time Period: August 12, 1994 - February 12, 1997 • Repeated interviews with 1,184 respondents at baseline, 2-week, 6-month, 1-year, 18-month • Included ego-centric questions on survey • “Provide the initials or nicknames of up to 5 your closest friends”

  16. Graph of reported syringe sharing by friendship rank and survey wave

  17. Adjusted Odds Ratios for Syringe Sharing (N=3,907).

  18. Three Studies with Data on Time-of-adoption & Social Networks

  19. Table. Regression on Time to Adoption by Network Exposure & External Contacts

  20. Table. Maximum Likelihood Logistic Regression on Adoption by Time, Ties Sent/Received & Network Exposure.

  21. Exposure Adoption? • Represents a challenge to the diffusion and other behavior change models • Could be a function of location on the diffusion curve – more likely after critical mass • Very disappointing from a replication perspective • What model can explain this?

  22. Network Threshold = Non User = User PN Threshold=33% PN Threshold=66% PN Threshold=100%

  23. Graph of Time of Adoption by Network Threshold for One Korean Family Planning Community 100% Threshold 0% Time 1973 1963

  24. Table: Adjusted Odds Ratios for the Likelihood of Low and High-threshold Adoption.

  25. 2. Network Level Effects • Density associated with more rapid diffusion (Valente 1995) • Centralization associated with more rapid diffusion (1995) • Clustering speeds/slows diffusion (Watts 2002) • Bridges accelerate diffusion (Granovetter) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  26. Dense Networks (D=36.4%) Decentralized (9.1%) Centralized (50.9%) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  27. Sparse Networks (D=18.2%) Decentralized (0.0%) Centralized (87.3%) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  28. Graph of diffusion by net density & centralization (clustering) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  29. Diffusion of Tetracycline for Marginal versus Integrated Doctors UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  30. UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  31. STEP Project Communities USC  Control  Prevention Training  Prevention Training + Technical Assistance UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  32. Community-level Networks – Coalitions & Inter-organizational Relations • 24 Communities being trained in substance abuse prevention (STEP) • Assigned to 3 conditions: • Control • Satellite TV training • Satellite TV training + TA • In each coalition, social networks of community leaders in coalitions were measured • Outcomes: Adoption of evidence-based programs UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  33. Structure & Adoption • Density is associated with more rapid diffusion (Valente 1995) • Centralization is associated with more rapid diffusion (Bavelas, 1955; Valente, 1995) • Lower clustering faster diffusion (Watts, 1999) Of course, in these settings this might not hold: • Density can overwhelm functioning • Centralized decision-making would be less egalitarian • Clustered networks provide reinforcement for adoption decisions UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  34. Coalition with High Density UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  35. Coalition with Low Density UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  36. Results:Intervention Effects Mediated by Density UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  37. Centraliz. & Clustering • No main or mediation effects of Centralization or Clustering • No association between network diameters and outcomes • No association between network clustering and outcomes UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  38. Increasing Density Inhibits Adoption Outcome Change Change in Density UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  39. SEM Testing Mediation UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  40. Coalition Type (Form) Affects Coalition Structure 1 2 3 UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  41. Considerable Variation in Coalition Network Indicators (N=24) UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  42. 3. Individual-Network • Individual network effects are captured by the immediate personal network • When the complete network is mapped, individual positions within the network can be determined: • Leaders • Isolates • Bridges • Group members UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  43. Network Positions Bridges Group Members Central Members Isolate UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  44. Popular Students Began Smoking

  45. Peer Networks and Adolescent Cigarette Smoking:Adolescent Health Survey (Bearman, Udry, et al.) • Randomly selected schools in which all students were surveyed and asked to name 10 best friends (5 male and 5 female). • 13 schools (2,590 students) collected sociometric data once. • Some outcomes measured in the household data only.

  46. Table: Adjusted Odds Ratios for Smoking Last 30-days. Data are from Adolescent Health Study (N=2,525).

  47. How To Speed diffusion? • Target interventions to Low Thresholders • Reduce Thresholds • Change networks • Increase exposure - have those who influence others be the first adopters • Sociometric Segmentation rather than geographic, socio-demographic, pyschographic UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  48. 4. Diffusion Effect • Susceptibility: Connections to others who already have adopted • Infectiousness: Connections to others who adopt after you UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  49. Calculating Infection = Non User 4 = User Adopted T=4 3 Infectivity=2 Infection %=50 4 5 7 UNIVERSITY OF SOUTHERN CALIFORNIA  INSTITUTE FOR PREVENTION RESEARCH

  50. Table. Maximum Likelihood Logistic Regression on Adoption by Time, Ties Sent/Received & Network Exposure.

More Related