Who is Studying What? An Anatomy of Climate Change Think Tanks in Europe

Note: This post's underlying data and an acronym key are available here. Network data files (nodes, edges, GDF, Gephi) are available by request by sending an email to bpmoore@gmail.com. This is a work in progress, feedback is greatly appreciated.

The International Center for Climate Governance has built a fascinating database and map of climate change think tanks on every continent. Each think tank is classified according to nine research fields (e.g., adaptation, policy and institutions, forestry and land use, etc.).

I was curious to see which research fields were most popular among climate think tanks in Europe, and how the organizations themselves related to one another. There are 141 European think tanks listed on the website, and each organization is linked to between two and six research fields. The most common research field is Policy & Institutions (a focus of 85% of the think tanks), and the least common is the Carbon Finance field (27%):

Among the think tanks, which ones work on the same research issues? To explore this question, I first converted the ICCG data into an affinity network which contains two types of nodes: think tanks and research fields. In this type of network, a think tank can only be connected to one or more research fields, not to other think tanks. The result is shown below, illustrating the central position that research fields such as Policy and Institutions play in the climate think tank idea ecosystem, as well as the fields such as sustainable cities which attract less attention:

My one issue with this affiliation graph is that it does not directly show when two think tanks are working on similar issues. I therefore created a second, projected graph, which removes the research field nodes and connects two think tanks directly if they share at least one research field in common:

Using Gephi, I identified three highly-connected communities within the overall think tank network (color coded green, blue, and yellow). However, as should be obvious, there is not a clear-cut separation between these communities. This is because almost all of the think tanks are connected to each other. In fact, this network has a very high density, .975 out of 1, meaning that in this case almost all of the possible connections between think tanks actually exist. In addition, the network has a very high average clustering coefficient (.98), which measures how likely it is that any three nodes are connected to each other.

Finally, for those interested in the networks of specific European regions, the image below highlights think tanks from six regions (the British Isles, German-speaking Countries, Mediterranean Countries, Nordic Countries, the Low Countries, and Eastern European Countries):    

Below are each of the six regional networks, removed from the overall graph to highlight their unique structures and connections.

United Kingdom and Ireland

Germany, Switzerland, & Austria

Italy, Spain, Portugal, & Greece

Norway, Sweden, Denmark, & Finland

Holland and Belgium

Hungary, Lithuania, Poland, & the Czech Republic

The IPCC as a Social Network: The Working Group I Summary for Policymakers

Just a quick post to get this blog going again. Below is a network map of the authors of three chapters of the Intergovernmental Panel on Climate Change from the 3rd, 4th, and 5th Assessment Reports (released in 2001, 2007, and 2013, respectively).

The chapter in question is the Working Group I Summary for Policymakers, the working group dealing with the physical science of climate change. There are 151 authors pictured in the image, with 26 of those authors writing for more than one Assessment Report. Four authors (in the center) were involved in all three years. A red connection means that that author was a Lead Author in that year, a blue connection denotes Contributing Authors.

By my count, 24 of the 26 (92%) "multi-report" authors are from Annex I (industrialized) countries. This includes all four authors in the center of the image.

Visualizing the UK National Ecosystem Assessment

For my recent master's dissertation, I studied the UK National Ecosystem Assessment (UK NEA), a government-initiated scientific assessment of the UK's ecosystems, their value to society, and policy options for their protection. One of the biggest challenges when researching an assessment of this type is keeping track of the large number of contributors involved. In the case of the UK NEA, the contributors numbered upwards of 500 people. In a process where some authors contributed to multiple chapters, keeping track of everyone and their relationships to each other is difficult.

To handle this complex situation, I used Gephi to create a network map which presents the UK NEA's authors as a social network held together by co-authorship. Using author data from the Assessment itself, I was able to create a map which included all 403 people who were listed as authors for at least one of the UK NEA's chapters.

UK NEA Authorship Network Map

The image above is based on a very simple foundation. Each circle represents one of the UK NEA's authors. A connection between two authors means that they co-authored at least one chapter together. In the center of the network, purple connections identify authors who co-authored the high-level Synthesis for Policy Makers. Magenta connections represent co-authorship of the Assessment's introductory chapters, while green connections represent habitat-based chapters that focused on specific types of ecosystems such as woodlands or freshwaters. Light green connections (found at the bottom of the image) are related to ecosystem services chapters, and gold connections signify country-focused chapters (in this case the four constituent countries of the United Kingdom: England, Scotland, Wales, and Northern Ireland). Finally, blue connections are related to ecosystem service valuation and scenario building (the valuation aspect of the UK NEA was the main focus of my dissertation research).

So what insights, if any, does this network map provide? First, the map confirms - and visualizes - a number of features of the UK NEA authorship network that I noticed during my research. For example, many of the authors were involved in only one chapter of the assessment, and these groups form dense, easy-to-distinguish clusters within the network. Connecting these clusters are a relatively small number of multi-chapter authors, as can be seen here in this close-up:

Author Clusters with Two "Connector" Authors (Center)

Second, the map helps illustrate the relationship between chapters written mainly by ecologists (green connections) and other chapters largely written by economists and social scientists (blue connections). Some ecology-focused chapters were relatively closely connected to the valuation chapters because they included ecologists or economists who also worked on ecosystem service valuation (Clusters A and B below). Other ecology chapters were not connected directly to the valuation chapters (Cluster C). Why the difference? This topic was not my focus, but it could be an interesting one to explore.

Ecology Chapters and Their Connection to Valuation Chapters

However, I believe that these network maps may be most useful for identifying effective research strategies. For example, if a researcher was studying interdisciplinarity in scientific assessments, they could use the network map above to identify the UK NEA authors who connected different chapters. They could then request an interview with these "connector" authors in order to learn more about the process of exchange both between chapters and between academic disciplines.

So, is creating a network map worth the time investment? (around 10-15 hours for the UK NEA map above) The answer depends on the research questions being pursued. But given the (relatively) easy learning curve for network software like Gephi, network visualization could be a valuable addition to researchers' intellectual toolkit. 

A (Network) Map of Europe

As part of my last job, I got a chance to work with network visualization and analysis as it related to biodiversity research in the EU. In the past month or so, I have also been working on a few fun side projects like the one below: a network map of Europe. 

In the map below, each point in the network (or "node") represents a European country that is a member of the United Nations. Two countries are linked in the network if they share a border; for example, at the top right Turkey and Greece share a border and so are linked together. Some non-European countries - such as China - are included because they share a border with a European country. 

A Network Map of Europe
(Full-size, Creative Commons-licensed image available here)

The size of a node is determined by how many other countries it shares a border with. On one extreme, Russia borders 14 other countries and is the largest node in the network. On the other end of the spectrum, the island nations of Malta (top-left corner) and Iceland (top-right corner) do not border any countries, and are visualized as tiny, free-floating nodes.

The network also contains five "communities" of nodes that are closely connected to each other, color-coded below as green, dark blue, light blue, dark orange, and light orange. These communities were defined by running a statistical test (for "modularity") on the free network visualization software Gephi.

This map was just for fun, but what could techniques like this offer to researchers or policy-makers working on European issues?

Sources

Border information is drawn from Wikipedia. Unlike the Wikipedia article, I did not include Kosovo because it is not a member of the United Nations and its sovereignty is disputed.