Presented by Dr. Matt Hamilton, The Ohio State University, USA and Dr. Caleb Gallemore, Lafayette College, USA, on 10 November 2020 at "International workshop: Enhancing wetland management and sustainable development"
3. Network analysis:
A set of approaches for understanding interactions
???
What is the structure
of those interactions?
4. Network analysis:
A set of approaches for understanding interactions
???
What explains the
structure of those
interactions?
???
What is the structure
of those interactions?
5. Network analysis:
A set of approaches for understanding interactions
??? ???
What are the
implications of
those interactions?
What explains the
structure of those
interactions?
???
What is the structure
of those interactions?
6. Components of a network: nodes and links
Representation
LinksNodes
people, organizations,
institutions, concepts/factors,
places, habitat patchesโฆ
collaboration, exchange of
information, co-occurrence,
biophysical connectivity...
capacity, economic sector,
preferences, location,
size, typeโฆ
direction (or lack thereof),
frequency, magnitude, sign,
typeโฆ
Measurement
7. Different levels of network analysis /
Examples of research questions
Node-level: Do local conservation groups have more influence (e.g., more
incoming links) than local authorities?
Substructure-level: Which types of organizations broker the exchange of
information about PES between communes and provincial-level
organizations?
Network-level: How well connected are forest governance networks?
8. โข People connect with visualizations of networks โ opportunities for
engagement with stakeholders and decision-makers.
โข Highly appropriate methodology for studying collaborative resource
management, coupled human-natural systems, and other environmental
social science fields that emphasize relationships.
โข Increasingly accessible and powerful tools for data collection, analysis,
and visualization.
โข Network perspectives are increasingly common (and expected) in many
environmental social science fields.
Strengths
& Opportunities
This is my
reality!
9. โข Respondent fatigue
โข Network analysis is more
sensitive to missing data
(because observations
are interdependent).
Weaknesses
& Challenges
โข Can be difficult to identify the boundary of a network.
โข Conceptualizing a complex system as a network requires
simplification and abstraction.
10. How networks help us understand forest
governance
Collaboration between two
organizations
Organization Organization
Organizations collaborate for
many different reasons.
What predicts collaboration
among organizations involved in
PFES/REDD+?
11. How networks help us understand forest
governance
Collaboration between two
organizations after participating in
the same PFES or REDD+ workshop
Organization Organization
If organizations participate in the
same workshops, are they more
likely to collaborate in the future?
If so, workshops may be playing
an important role in sparking
cooperationWorkshop
12. How networks help us understand forest
governance
Collaboration between two
organizations that work in the same
place
Organization Organization
If organizations work in the same
place, are they more likely to
collaborate?
If so, those partnerships can help
organizations avoid inefficiencies
Place
13. Exploring these questions by bringing
together multiple network datasets
Organization Organization
Longitudinal collaboration
network from REDD+ Policy
Network Study
Longitudinal network of
organizations participating in
REDD+ and PFES workshops
Network of where organizations
work
WorkshopOrganization
Organization Place
14. Preliminary results from network model
(stochastic actor oriented model)
Model 1
Rate parameter period 1 7.39 (0.49)***
Rate parameter period 2 5.58 (0.45)***
Outdegree (density) -1.61 (0.18)***
Reciprocity -0.00 (0.20)
Organizations co-attended workshops in prior time period 0.53 (0.22)*
Organizations work in the same place -0.31 (0.15)*
Governmental organization 0.51 (0.33)
Collaboration between governmental organizations 0.56 (0.20)**
Iterations 2669
***p < 0.001, **p < 0.01, *p < 0.05
15. Yes, organizations that attended the same workshops
are more likely to collaborate in the future!
Model 1
Rate parameter period 1 7.39 (0.49)***
Rate parameter period 2 5.58 (0.45)***
Outdegree (density) -1.61 (0.18)***
Reciprocity -0.00 (0.20)
Organizations co-attended workshops in prior time period 0.53 (0.22)*
Organizations work in the same place -0.31 (0.15)*
Governmental organization 0.51 (0.33)
Collaboration between governmental organizations 0.56 (0.20)**
Iterations 2669
***p < 0.001, **p < 0.01, *p < 0.05
16. But organizations that work in the same places are
more less likely to collaborate!
Model 1
Rate parameter period 1 7.39 (0.49)***
Rate parameter period 2 5.58 (0.45)***
Outdegree (density) -1.61 (0.18)***
Reciprocity -0.00 (0.20)
Organizations co-attended workshops in prior time period 0.53 (0.22)*
Organizations work in the same place -0.31 (0.15)*
Governmental organization 0.51 (0.33)
Collaboration between governmental organizations 0.56 (0.20)**
Iterations 2669
***p < 0.001, **p < 0.01, *p < 0.05
17. Taking stock of (preliminary!) results
In addition to disseminating information and training participants,
workshops may also catalyze collaboration, which can contribute to
more cohesive forest governance
Organizations seem to avoid one another when working in the same
areas. This may present challenges for coordination of forest
management activities
29. So letโs think about some of these
networks together.
Corridors
Watersheds
Governance
Supply
Chains
Information
30. New Partnerships for Sustainability (NEPSUS)
Project, Tanzania
โข Coordinated by Stefano Ponte at Copenhagen Business School and
Christine Noe at University of Dar es Salaam
โข Studies three different natural resource sectors:
39. Assessing PFES in Vietnam
โข Random nation-wide sample of 250,000 pixels that were forested as of
2000 from Global Forest Watch data, combined with random sample of
250,000 pixels that were forested in 2000 but deforested by 2018
โข Data on provincial activities carried out by organizations recorded as
active in REDD+ and/or PFES by CIFORโs GCS-REDD studies carried out
since 2010
โข Control variables - elevation, terrain ruggedness, surrounding cropland,
distance from road
โข Random effects by region (will move to province with a larger sample)
43. Summary
โข Conservation efforts take place in a complex network of
processes
โข Emerging evidence both institutional (policy) and relational
(network) variables affect success in stemming forest loss
โข Network building for conservation isnโt just about building
bigger networks, but also about building networks with the
most effective structures