Monday, June 10, 2019

Conservative Party leadership candidates and indicative vote factions

David Jeffrey has been doing very interesting work visualizing the UK Conservative Party leadership election, which started me thinking about how support for each candidate would differ depending on how an MP voted on the Brexit-related indicative votes earlier this year.

In April, I used social network analysis to identify voting factions among UK MPs during the indicative vote process. For that analysis, a faction was defined as a group of MPs who voted together on all twelve of the indicative votes. The image below shows the results by political party.

I identified eight factions within the Conservative Party with more than five members, totaling 212 MPs and 68% of the party. A further 45 MPs were a part of factions with between two and five members, and 57 MPs did not vote in lock-step with anyone. 

Below is an overview of the factions and how they voted (Euroskeptic scores come from Alexandre Afonso).

Voting factions and the leadership contest

My data on the candidates Conservative MPs comes from David Jeffrey's map of this information, downloaded on 9 June. As a result, it will be somewhat out of date, given new endorsements in the last few days. However, given that I am looking at broad groups of MPs, using slightly out-of-date should still provide useful insights and be used as a comparative baseline later.

The graphs below make clear that the factions are quite divided, and in most, a majority have not publicly endorsed (the elections are by secret ballot). But it is clear that e.g., Johnson draws much of his support from the large, Euroskeptic factions (1 and 2), while Jeremy Hunt and Michael Gove are more dependent on the 'Other' group.

Thursday, November 1, 2018

Gender representation on UACES 2018 panels

I recently read a blog post on unequal gender representation on academic panels (especially all-male panels, so-called 'manels'). In it, the authors linked to another post that estimated that all-male panels are "getting more rare each year" at the International Studies Association conference.

I realized that after my social network analysis of panels at the 2018 UACES conference and the follow up on the role of gender in the conference's network structure, I already had the data to create a detailed estimate of gender representation on UACES panels.

The caveats

This analysis comes with similar caveats to my previous post on the topic.

First, I coded gender myself, so the results are based on my own interpretation, and not self-identification.

Second, the data is based on the UACES public program, and so does not measure the people who actually participated in person. For example, my panel (Panel 205) is listed as having a 40%/60% female/male split, but was an all-male panel if measured by the people in the room.

Finally, the network is held together by panels that share formal participants, and does not address more relevant - but difficult to measure - aspects such as who went to panels as an audience member.

The findings

As a reminder, for the UACES conference as a whole, the gender split was almost exactly 50-50: 258 participants were women (49.4%) and 264 were men (50.6%).

The headline findings on panel composition mirror this 50-50 split.

Across the 127 panels analyzed, the average (mean) representation overall was 51% women, 49% men. The most common percentage was an exact 50-50 split (16 panels).

Unsurprisingly (for me) these averages masked a lot of variation.

In Figure 1, I show all UACES 2018 panels categorized by the percentage of women participating. There were two all-male panels. In a little more than half of the panels, 50% or more of the participants were women (65 panels, 51%). Around a fifth of panels had more than 80% female participants (24 panels, 19%); of these, 14 featured only women.

Figure 1: UACES 2018 panels by percentage of participants that are women

This information is visualized differently in the network map in Figure 2, where darker green panels have a higher percentage of female participants.

Figure 2: UACES panels connected by shared participants. Greener panels have a higher percentage of participants that are women.

Panels, gender, and network importance: Four angles

How is the gender balance of a panel related to its importance within the network?

In this section I will explore this question using four network statistics: betweenness centrality, closeness centrality, eigenvector centrality, and weighted degree. These three values measure the importance of nodes in different ways, which allows us to take a multifaceted approach. In Figures 3-6, I take the network presented in Figure 2 and re-size the nodes based on their ranking on these statistics (larger nodes have higher scores).

In Figure 3, the size of the nodes depends on their betweenness centrality, a measure of "how often a node appears on shortest paths between nodes in the network." Nodes will a high betweenness centrality are more important to the overall network's connectivity, and removing them is more likely to lead to network fragmentation.

Figure 3: UACES panels re-sized according to their betweenness centrality, a measure of "how often a node appears on shortest paths between nodes in the network (larger nodes have higher centrality). Greener panels have higher percentages of participants that are women.

In Figure 4, the size of the nodes depends on their closeness centrality, "the average distance from a given node to all other nodes in the network."

Figure 4: UACES panels re-sized according to their closeness centrality, which measures average distance from all other nodes (larger nodes have a higher centrality). 

In Figure 5, the size of the nodes depends on their eigenvector centrality, which measures a node's importance based on the importance of the nodes it is connected to; "a node is important if it is linked to by other important nodes."

Figure 5: UACES panels re-sized according to their eigenvector centrality, which measures which measures a node's importance based on the importance of the nodes it is connected to (larger nodes have a higher centrality). 

Finally, in Figure 6, panels are re-sized according to their weighted degree, the number of connections they have with other panels.

Figure 6: UACES panels re-sized according to their weighted degree a measure of how many connections they have with other panels (larger nodes have a higher number of connections). 

Table 1 compares the scores on these four statistics of a number of sub-groups. First, I divided panels into those that had 50% or more female participants, and those that had 49% or less. I then compared the 14 all-female panels with the 14 panels with the lowest percentages of female participants (including the two all-male panels). Finally, for each statistic, I took the top  3, 5, and 10 panels and compared their average percentage of female participants.

Panels with more women had lower betweenness centrality, but were close to even on closeness, eigenvector, and degree. The all-female panels had much lower betweenness than the bottom 14, somewhat lower on closeness, even on degree and significantly higher on eigenvector. The top scoring panels for betweenness and closeness centrality had lower than average female participation, but higher than average for eigenvector centrality and weighted degree.

In summary, different measures of network importance come to different conclusions in this case.

Table 1: Comparing sub-groups on the four network statistics and percentage of participants that are women.


I'd like to emphasize again that these findings should be interpreted with caution: as noted above, the information used to build the network (shared participants) is far from ideal and there could be a number of explanations for the findings (which are not analyzed here).

Hopefully the analysis above can help spark productive conversations on these issues.

Sunday, September 16, 2018

UACES 2018 and social network analysis: Exploring gender dimensions

Update: Using the data I compiled for this post, I have created estimates for the percentage of men and women participating in all UACES 2018 panels. That analysis can be found here.

During the 2018 UACES conference on European studies in Bath at the beginning of September, I created several social-network visualizations of panels and participants. The visualizations that focused on participants explored both the number of connections each person had to others, and the 'betweenness centrality' of each person in the network (a measure of their importance for holding the network together).

Two academics who attended the conference - Dr Toni Haastrup and Dr Katharine Wright - asked me about the gender dimensions of these topics (for much more expertise on gender in EU studies, a good place to start is the UACES Gendering EU Studies research network). 

This took additional analysis because the gender of UACES participants was not given on the conference website. So I coded the gender of each participant and added that data to the network maps (like my earlier maps, all participants were anonymized). 

This approach has important limitations: for instance, it is based on my own analysis and not conference participants' self-identified gender. In addition, the results are broad and show only the formal connections between participants, e.g. not those who attended each panel in the audience.

But the results are nevertheless interesting. Below I've focused on two aspects: network centrality and each participant's number of connections.

Network centrality

After removing duplicates, the UACES program listed 522 participants. The gender split was almost exactly 50-50: 258 people were female (49.4%) and 264 were male (50.6%).

The map below shows network participants color-coded according to gender (green is female, blue is male) with node size determined by betweenness centrality, which the network analysis program Gephi defines as "how often a node appears on shortest paths between nodes in the network".

UACES 2018 participants weighted by network centrality. Participants are connected when they participated in the same panel. Green nodes are female participants, blue nodes are male participants. Node size is related to a participant's centrality in the network (betweenness centrality). High-resolution, searchable PDF is available here.

There are obvious differences between the centrality scores of the nodes: the highest betweenness score is 21,203, while over 70% of nodes have a score of zero (because betweenness centrality is a measure of how often a node is on the 'shortest paths' between nodes).

However, there is almost no average difference based on gender, where scores are almost identical (951.1 female participants, 955.3 male).

Focusing only on the 150 participants with a betweenness centrality score above zero, men were more heavily represented in top-20 highest scores, but this evened out in the top 50 and top 150.

Percentage of female/male participants with the highest betweenness centrality scores.

Number of connections

The second way I looked at the issue was through the number of connections participants had with others (weighted degree). The average number of connections overall was 7.4. This average masks a wide range, from a low of 2 connections to a high of 32 connections.

UACES 2018 participants weighted by number of connections (weighted degree). Participants are connected when they participated in the same panel. Green nodes are female participants, blue nodes are male participants. Node size is related to a participant's number of connections (weighted degree). High-resolution, searchable PDF is available here.

On this metric, male participants had an average about 10% higher than their female colleagues (7.1 female vs. 7.8 male). Male participants also made up a high percentage of the top 20 nodes by weighted degree (although this once again was reduced to 50-50 for the top 150).

Percentage of female/male participants with the highest weighted degree scores (number of connections).


A few key findings jump out from this brief analysis.

First, the gender of participants this year was split almost 50-50, in contrast to findings in other disciplines and venues.

Second, there was almost no difference in the average network centrality by gender, and a 10% higher number of connections for male participants.

Third, this broad similarity on average masked a heavy percentage of male participants in the top scores of both centrality and number of connections.

I'd emphasize again that these results should be interpreted with caution: they are preliminary, from one year and one conference, and focus on a single aspect of the dense social networks that exist in these gatherings.

But they do raise interesting further questions:

How have these and other metrics changed over time, in UACES and beyond? What are the differences between areas of focus (e.g. gender studies, environmental policy, foreign policy)? And how do these findings fit into the rich existing academic and societal discussions on gender issues?

I hope this analysis can be an interesting, if small, part of that wider conversation.

Saturday, May 21, 2016

Mapping the Literature on the UNFCCC

In light of the ongoing climate change negotiations in Bonn this week, I decided to take a quick look at the existing academic literature on the United Nations Framework Convention on Climate Change. As of today, there were 1,375 journal articles on the Scopus database that include the terms "United Nations Framework Convention on Climate Change" or its acronym "UNFCCC". The graph below shows the number of articles per year:

A little less than 30% of those articles (nearly 400) were written in just ten journals:

I used the VOSviewer program to analyze the title and abstracts of these articles.  Using the default settings for analyzing a text corpus, it identified 25,470 unique terms, 781 of which occurred in at least ten articles. I then asked it to identify the top 60% of articles in terms of a "relevance score", which removes very common terms etc.

The map below shows the resulting 469 terms, connected when they co-occur in an article:


This second map color codes the terms according to the average year of publication for the documents in which they occur. Notice the difference between the Clean Development Mechanism and more recent REDD+ terms, as well as the very recent average for "Paris" on the left.

Finally, this heat map shows the most commonly occurring terms and when those terms cluster with each other:

**Quick Update**

I did the same analysis, but starting with terms that occurred in at least three articles (not ten as above).

Sunday, October 18, 2015

WIREs Climate Policy and Governance: Who Cites Which Journals/Books?

Following up on yesterday's post on the WIRES Climate Change Policy and Governance topic, I'm taking a look at the same data (thirty eight articles) from a new angle: which journals, books, and other sources they cite in common.

The top ten sources cited (all of which are journals) are presented in the table below.

Top Ten Journals Cited in WIREs Climate Change Policy and Governance articles.
The significance of these journals is clear in the network visualization below, with the top journals clearly visible in the center.

WIREs Policy and Governance articles (red) and the journals, books, and other sources that they cite (blue).

One interesting point to note is that some journals get a large percentage of their citations from a few papers. For example, Friman and Standberg 2014, which focuses on historical responsibility for climate change, cites the journal Climatic Change nine times, almost a third of that journal's thirty two overall citations. This seems to be due to important articles on historical responsibility that have been published in Climatic Change.

The journal Climatic Change gets around 30% of its citations from one article (Friman and Sandberg 2014).

To illustrate the pattern of some journals being cited more times per article, the chart below shows the top ten cited journals by both the number of WIREs articles citing them (horizontal axis) and the number of times they are cited overall (vertical axis, includes multiple cites by a single paper).

Number of WIREs article citing a journal (horizontal axis) versus the number of times cited overall (vertical axis), with illustrative examples.

Finally, the data for all sources that were cited by at least two WIREs Policy and Governance articles:

Saturday, October 17, 2015

Studying Climate Policy and Governance: Mapping Citations

I have recently been thinking about network analysis and visualization as a research analysis tool. One of the ideas I have been considering is mapping citations within individual journals when reviewing the academic literature.

To explore this possibility, I took a relatively small sample of thirty eight articles, which came from the Climate Change Policy and Governance topic in the journal Wiley Interdisciplinary Reviews: Climate Change (WIREs Climate Change). I limited the analysis to articles that a) are included in the Scopus database and b) have their references listed there. 

These articles had a total of 2,364 references. Of these, 106 were cited by at least two of the WIREs Climate Change articles. I used this data, Gephi, and the SciencesPo MediaLab's Table2Net tool to create a network visualization which connects two WiREs articles if they cite at least one source in common.

In the following network visualizations, article 1 and article 2 are connected if they share at least one reference in common.

The analysis results in the network below, which includes the thirty two articles that share a reference with at least one other article in the sample. The thickness of a connection shows the number of references in common (Branger et al. 2015 and Laing et al. 2014 at the bottom share nine references in common). The size of each circle shows the total number of shared references an article has (the two largest, Munck af Rosenschold et al. 2014 and Gupta 2010, share twenty one references each with other articles).  

Climate Policy and Governance Articles connected by shared references.

In the following two images, I show two ways of categorizing these articles. The first is based on WIREs Climate Change's own sub-topics, which divide the articles according to their focus (e.g., national policy or private governance). As the map shows, these sub-topics are relatively fragmented across the network.

Climate Policy and Governance Articles: Shared references, color coded according to WiRES Climate Change sub-topics.

The second image below organizes the articles into "communities" based on shared references. Most of these communities include articles from a number of WIREs sub-topics (for example, the yellow community includes articles from the multilevel/transnational, national, international, and cities sub-topics).

Climate Policy and Governance Articles: Shared references, color coded according to network "communities" identified by Gephi.

For now, I'm going to leave the analysis at the descriptive level, but I'm planning to come back to these articles in the near future.

Friday, July 31, 2015

Crossing the 1°C threshold: Science, symbolism, and climate politics

New Scientist magazine recently published an article showing that the world is on track to reach 1°C of global average warming above pre-industrial temperatures in 2015 (in this case, the average from 1850-1899). This is halfway to the 2°C warming limit which has been agreed to in international negotiations under the United Nations Framework Convention on Climate Change. It is also two thirds of the way to the stricter 1.5°C limit which some countries, notably small island nations, have argued is necessary. The 1°C 'milestone', if it is reached, will come in the same year as the crucial Paris climate negotiations, which aim to secure a global deal to reduce emissions and adapt to climate change.

Source: New Scientist

In many ways, there is not an important distinction between 1°C in 2015, 0.9°C in 2014, or 1.1°C in the future. However, as a researcher studying climate politics, I am very interested to see if and how crossing the 1°C threshold is understood in international negotiations, national policy debates, and the media.

In other words, how will 1°C of warming be framed? Will it be mentioned in media accounts alongside the 2°C limit and the rising concentrations of greenhouse gases in the atmosphere? Will national governments, NGOs, and other political actors such as the European Union refer to 1°C when pushing for more stringent emissions reductions?

And of course, if it is mentioned in policy debates, will this milestone have any effect? It is notoriously difficult to untangle the political effects of ideas and symbols. The most famous example of such an idea attracting a large amount of attention in environmental negotiations is the hole in the ozone layer which was detected during negotiations to limit emissions of ozone-depleting substances. Many participants in those negotiations mentioned the effect the ozone hole's discovery had on them. But even in this case, political scientist Edward Parson has raised doubts that the ozone hole had an important political effect. So studying the 1°C  and its political effects will be difficult.  

On a personal note, when we do 'officially' cross the 1°C threshold, I feel as if a door is symbolically closing on the world I grew up in. Again, I don't believe there is any significant scientific difference between 0.9°C and 1°C. Its importance is symbolic.

And hopefully that symbolism, of climate change being in 'the here and now', will provide a push, however minor, for ambitious climate change policy.