Introduction to Social Network Analysis for Dissemination and - - PowerPoint PPT Presentation

introduction to social network analysis for dissemination
SMART_READER_LITE
LIVE PREVIEW

Introduction to Social Network Analysis for Dissemination and - - PowerPoint PPT Presentation

Introduction to Social Network Analysis for Dissemination and Implementation Research Miruna Petrescu-Prahova, PhD mirunapp@uw.edu Health Promotion Research Center Department of Health Services University of Washington July 1, 2018 Prime


slide-1
SLIDE 1

Introduction to Social Network Analysis for Dissemination and Implementation Research

Miruna Petrescu-Prahova, PhD mirunapp@uw.edu

Health Promotion Research Center Department of Health Services · University of Washington July 1, 2018

slide-2
SLIDE 2

Prime time for more social network analysis in HSR

◮ Social network concepts are frequently used, but rarely formally

measured

◮ Computing power and data access allow for analysis of large-scale

networks, but underlying mechanisms are not well understood

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 2 / 34

slide-3
SLIDE 3

Outline

  • 1. Terms and definitions
  • 2. Relevance of social networks in dissemination and implementation

research

  • 3. Social network measures and study design
  • 4. Example: Advice-seeking among grantees of CDC’s Colorectal Cancer

Control Program (CRCCP)

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 3 / 34

slide-4
SLIDE 4

Defining principles of social network theory

◮ Motivated by a relational, structural perspective

◮ Actors and their actions are interdependent ◮ Ties between actors are channels for exchange of resources,

information, norms

◮ Network ties enable but may also constrain actions ◮ Structure (i.e., enduring patterns of relations) influences outcomes

◮ Social mechanisms lead to certain structural patterns

◮ Homophily (“birds of a feather flock together”) ◮ Reciprocity and transitivity (“friend of a friend is also a friend”) ◮ Preferential attachment (“the rich [in ties] get richer”)

Social network analysis: study of patterns of relations in social systems

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 4 / 34

slide-5
SLIDE 5

We need to think formally about networks

“Network” is not just a metaphor: it is a precise mathematical construct

◮ Network: a collection of nodes,

together with a set of relations (ties) on those nodes

◮ Nodes: individuals,

  • rganizations, states

◮ Relations: friendship, advice,

collaboration, competition

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 5 / 34

slide-6
SLIDE 6

Dissemination and implementation research definitions

◮ Dissemination research: the scientific study of targeted distribution of

information about evidence-based interventions to a specific public health or clinical practice audience.

◮ Implementation research: the scientific study of the use of strategies

to adopt and integrate evidence-based health interventions into clinical and community settings.

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 6 / 34

slide-7
SLIDE 7

Social networks are the backbone of diffusion and dissemination

HPRC Dissemination Framework (Harris et al., 2012)

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 7 / 34

slide-8
SLIDE 8

Social networks are part of implementation context

Consolidated Framework for Implementation Research (Damschroeder et al., 2009)

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 8 / 34

slide-9
SLIDE 9

Two ways to think about social networks in health services research

  • 1. As conduits for knowledge transfer
  • 2. As part of the context in which implementers are embedded, and

which may influence their ability to implement a program successfully

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 9 / 34

slide-10
SLIDE 10

Three levels of social network analysis

  • 1. Descriptive: what are the properties of the network we observe?
  • 2. Null hypothesis tests: how does the network compare with random

networks of same size and basic structure? Is there evidence of an underlying process?

  • 3. Statistical models: what are the processes that have led the network

to its particular configuration (out of the set of all possible configurations?)

◮ Cannot simply use existing statistical methods ◮ Want to explicitly model interdependencies, relations, linkages,

structural patterns

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 10 / 34

slide-11
SLIDE 11

Steps for network study design

  • 1. Define the boundaries of the group
  • 2. Define the relationship(s) of interest and propose mechanisms
  • perating through the relationship(s)
  • 3. Collect data on the relationship through surveys, archival sources,
  • bservation, or sensors
  • 4. Create a network visualization
  • 5. Analyze the network using descriptive or statistical methods
  • 6. (Optional) Design an intervention to improve dissemination, support

implementation, or accelerate behavior change

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 11 / 34

slide-12
SLIDE 12

Network Interventions (Valente, 2012)

  • 1. Identification of opinion leaders
  • 2. Identification of clusters
  • 3. Peer-driven interventions
  • 4. Rewiring

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 12 / 34

slide-13
SLIDE 13

Advice-seeking among Colorectal Cancer Control Program Grantees, 2011-2012

◮ Colorectal cancer (CRC) is the second leading cause of cancer deaths

among men and women in the United States

◮ CRC screening reduces cancer deaths by detecting cancers at an early

stage and by detecting and removing precancerous polyps before cancer develops

◮ People without health insurance or a regular health care provider are

more likely never to have been screened

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 13 / 34

slide-14
SLIDE 14

CDC’s Colorectal Cancer Control Program (CRCCP)

◮ Goal: to increase CRC screening rates among men and women aged

50 years and older from about 64% to 80%

◮ Provided funding to 25 states and 4 tribes across the United States

for five years (2009-2014)

◮ Evidence-based strategies to be implemented by CRCCP grantees

  • 1. Client reminders
  • 2. Small media
  • 3. Reducing structural barriers
  • 4. Provider reminders
  • 5. Provider assessment and feedback

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 14 / 34

slide-15
SLIDE 15

CRCCP grantees

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 15 / 34

slide-16
SLIDE 16

CRCCP evaluation - annual grantee survey

◮ Screening promotion: use of the 5 evidence-based strategies, reasons

for using those interventions, and their ease of use

◮ Percentage of grantees implementing evidence-based CRC strategies

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 16 / 34

slide-17
SLIDE 17

CRCCP evaluation - annual grantee survey

◮ Evaluation question

◮ Do grantees seek advice from each other about evidence-based strategy

implementation?

◮ Rationale

◮ Helps identify opinion leaders, or best practice disseminators ◮ May provide a structure for the dissemination of other evidence-based

strategies, perhaps in other domains like cardiovascular disease

◮ Network survey item

◮ What other CRCCP programs do you go to for advice or support to use

the five Community Guide-recommended (evidence-based) strategies? (followed by a list of all grantees)

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 17 / 34

slide-18
SLIDE 18

2011 and 2012 CRCCP advice networks

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 18 / 34

slide-19
SLIDE 19

Descriptive analysis

  • 1. Network-level indices: density, reciprocity (mutuality), centralization,

transitivity

  • 2. Node-level indices: centrality, brokerage
  • 3. Group identification: core-periphery, clustering, blockmodels

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 19 / 34

slide-20
SLIDE 20

Descriptive measures

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 20 / 34

slide-21
SLIDE 21

Results of statistical analysis to identify mechanisms

◮ 2011: observed advice-seeking patterns can be explained by a

tendency for grantees seeking advice to both reciprocate requests and form clusters of advice-seeking, as well as homophily by region, and whether they are implementing strategies to reduce structural barriers

◮ 2012: advice-seeking was largely explained by the existence of a

previous relationship in 2011, although there is still strong homophily by region

◮ Results suggest that encouraging advice-seeking among grantees may

lead to long-lasting, mutual support relationships in this group

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 21 / 34

slide-22
SLIDE 22

Questions?

Miruna Petrescu-Prahova, PhD (HPRC) Social Network Analysis July 1, 2018 22 / 34