Reviewing the 2014 Political Studies Conference with Twitter Data

By Justin Murphy, Lecturer, University of Southampton  @jmrphy

What were the most talked-about issues during the 2014 Political Studies Conference in Manchester last week? How are those issues connected? How many attendees used Twitter? How was that communication network structured? How did sentiment change over time throughout the conference?

To get some leverage on questions such as these, below I analyze all the tweets containing the hashtag #PSA14 between Monday April 14 (roughly 9 a.m.) until Wednesday April 16th (roughly 8 p.m.). The Twitter API only provides 1500 tweets per call but, conveniently, that allows us to analyze every tweet during this period. As with all social media data, this data is almost certainly not representative of the whole population of attendees. Needless to say, these analyses do not describe the conference as such, but only that portion of the conference which took place on Twitter!

The dynamics of conference tweeting

No big surprises here. Conference tweeting typically peeked in the middle of each day, and the middle day of the conference saw the most Twitter activity. Each bar represents a 30-minute interval.


Who tweeted the most?


Who received the most retweets?


Who received the most retweets, as a ratio of total tweets?

Retweets are an indicator of influence, but people who tweet frequently will be retweeted more than infrequent tweeters, so a better measure of influence is retweeted tweets as a ratio of total tweets.


What was the network structure of retweets?

Who retweeted who, visualized as a network.


Centrality and Brokerage

Number of retweets is a good first look at influence, but not all retweets are equal in network terms. A tweet which is retweeted from someone who is also often retweeted is more influential than receiving retweet from someone who is rarely retweeted. This is what is captured by the network concept of “eigenvector centrality.” On the other hand, there might be nodes in the network which are not necessarily retweeted very frequently or by influential nodes, but are important nodes because they connect many other nodes who otherwise would not be connected. “Betweenness centrality” captures the degree to which a node sits on the shortest path between all other actors. Typically, eignevector centrality and between centrality will be positively correlated, and this is a straightforward indication of the power of that node in the network. But nodes off the diagonal represent other kinds of power. Nodes with high eigenvector centrality but low betweenness centrality are relatively powerful nodes but relatively outside the community. Nodes with high betweenness centrality but low eigenvector centrality are “brokers” who are not high-visibility but are powerful because a relatively high number of nodes go through that node to stay connected.

Thus, this graph helps us identify interesting nodes in the conference network, which we would not notice by just looking at who tweeted the most and who was retweeted the most. The businessman and journalist @MarkFox__ did not tweet much on #PSA14, and he’s a relative outsider to the network, but when he tweeted he was disproportionately retweeted by the most influential nodes within the network. @chrisgold was not retweeted very frequently by very influential nodes, but a relatively high number of nodes are connected to the rest of the network through him.


Note: That some of the usernames here have an “@” in front but some of them don’t suggests that there might have been a small error in some of the text-processing of usernames. I would fix this, if it were anything more than a blog post! It doesn’t seem to be much of a problem, as everything else is very consistent with what we would expect.

How positive/negative was the conference tweeting?

I used the Hu and Lui sentiment lexicons to do a basic “sentiment analysis” of the conference tweets. A sentiment lexicon is basically one large list of negative words and one large list of postive words. It sounds naive but it works, as evidenced by how much this is used in marketing. The sentiment of a tweet is simply the number of postive words minus the number of negative words.


How did the overall sentiment change over the course of the day, on average?


How did the overall sentiment change throughout the conference, by the hour?


What were the most frequent terms tweeted, and how were they associated?

Below are the 50 terms which appear most frequently throughout the 1500 tweets (I use stems of words rather than whole words to avoid redundancy, as is conventional in text-mining). The first several are terms related to the conference, but then several substantive political issues emerge. Some notable themes include: youth politics, UKIP, engagement and participation, and media.

Freq. Term

119 new
95 great
94 now
84 group
83 suit
77 research 76 paper
73 day
72 journal
71 today
66 present
65 teach
64 peopl
61 polici
60 prize
58 local
57 manchest 57 time
55 chair
55 ukip
55 will
53 award
53 young
52 confer
50 dinner
50 discuss
48 interest 48 vote
47 specialist 46 thank
45 need
45 year
45 youth
44 issu
44 studi
43 congrat
43 make
43 talk
42 best
42 come
42 media
42 onlin
42 work
41 just
41 roundtabl 41 session
40 coproduct 40 see
39 chang
38 evid

How were these main themes connected?

In other words, how did the most-tweeted terms cluster in the entire network of terms discussed? To understand this better, we can visualize the entire vocabulary of the 1500 tweets as a network. Every word which appears in a tweet with another word, represents a non-directional connection. Below is a graph of this network with the most central nodes highlighted and labeled.

download10 This post revealed the who, what, and when of how the 2014 Political Studies Conference took to Twitter this year. Feel free to explore or download the raw dataset available here, or use the other materials in this Github repository to check/extend these analyses yourself.

Follow Justin Murphy on Twitter @jmrphy.

The question of revolution is not romantic, it is unavoidable

By Justin Murphy, Politics & International Relations

Cross-posted here. Follow Justin on Twitter: @jmrphy

If I insist on the revolutionary position, it is not to insist on the dichotomy between revolution and reform. Most of us today will agree with Gorz that there exists a class of revolutionary reforms, at which point the relevant distinction becomes the distinction between revolutionary reforms and reformist reforms. Today, Nancy Fraser suggests the critical distinction is between “system-conforming” changes and “system-transforming” changes, but it seems to me that the long-standing theoretical and practical difficulty remains the same: which types of projects (individual or collective) effectively oppose capitalism and push society toward justice, and which types of projects (whether through mystification, co-optation, or defeat) merely improve capitalism for some at the price of renouncing the system-level opposition which would be the maximally true, coherent, and just position.

To my mind, this is the essence of the revolutionary position: To believe that the organisation of the world’s institutions are unjust, to see empirically that a key feature of these institutions is precisely that they offer particular groups small gains in return for their renunciation of system-level opposition, to therefore locate this precise mechanism as the essential and perhaps only mechanism which is able to maintain such massive worldwide system-level injustice, and finally to assume the theoretical and practical position to never renounce system-level opposition in exchange for any particular gain less than the absolute system-level transformations which are required for justice, no matter how relatively transformative such gains might be.1

Because of the almost primordial or, in any event, perennial quality of this tension and its unavoidable need for resolution in any theoretically defensible political project, I see no way that any political theory today can innocently elide the question of revolution. I do not say that any political theory today must be explicitly revolutionary in any specific sense. I say only that distinctions between “system-conforming” and “system-transforming” beg the crucial question which will always arise for those who agree to pursue system-transforming collective action: when the state, the market, and/or the thousands of institutions such as the university (defined by irrevocable cognitive and material allegiances to the state and market) offer us a particular “transformation” on condition that we demobilise just enough to not threaten the equilibrium of the institutional arrangement as such, should we accept that transformation or not?

That is, after individuals and groups choose “system-transforming” rather than “system-conforming” agendas, what are the conditions under which it is justified for them to demobilise their system-transformative demands in exchange for some political victory which improves the world but is less than what they see as a fully adequate transformation of the system? For this is the perennial dilemma with which all system-transforming political projects constantly struggle, and indeed it is this dilemma to which the very notion of an anti-capitalist (system-level) perspective is supposed to counter.

Finally, while I zealously affirm that any defensible revolutionary position will not only be anti-capitalist but also feminist, anti-racist, ecologically sustainable, and inclusive of many other human differences which “revolutionaries” have a long history of betraying, in no way do these inclusions obviate the question of revolution. Indeed, it is precisely because gender, race, and ecological as well as class struggles so urgently require truly system-level institutional transformations, that it is all the more important for us to maintain what is specific about the question of revolution and the meaning of the revolutionary position.

No matter how naively romantic the revolutionary position rings to our contemporary ears, it’s naiveté and romanticism is only a function of the merely contingent strength of the current status quo, and moreover our aversion to this seeming romanticism of the revolutionary position is itself merely the cognitive inheritance of a politics now more than ever defined by collaboration with injustice rather than resistance to it. Perhaps the meaning of the revolutionary position is indeed nothing more than an integral naiveté, but on the wager that real integrity to a truth is exactly the most emancipatory political force in the world. And perhaps the most dangerous romanticism existing today is the notion that humans have suddenly been absolved of having to decide whether they will negotiate with oppressive institutions or overthrow them.

  1. This phrasing is purposely agnostic about what exactly constitutes justice or what any ultimate institutional configuration should look like (or how this would be determined). This is because, for the moment, I am trying to sketch what is essential and specific about the revolutionary position as inclusively as possible with respect to any particular vision of political justice. Thus, the only essential premises with which one has to agree here are: 1) that there currently exist system-level injustices in the arrangement of institutions, and 2) that we can at least in principle admit the possibility of a globally just arrangement of institutions. One does not even have to agree that capitalism is the name of the currently unjust institutions, to see how a commitment to system-level injustice necessarily implicates one at least in the question of revolution.