Category: Social Network Analysis

New Article published at European Journal of Political Research

11/1/2018

The paper "The Power of Contact: Europe as a Network of Transnational Attachment", coauthored with Jan Delhey, Monika Verbalyte, and Auke Aplowski, has just been published online first at the European Journal of Political Research. A free pre-print version is available here.

In this article, we propose transnational attachment as a novel indicator of sense of community in Europe, arguing that this hitherto neglected dimension is substantially and structurally different from alternative ones such as cross-border trust and identification. Combining Eurobarometer 73.3 data on ties between all EU-27 countries with further dyadic data, we show empirically that the European network of transnational attachment has an asymmetric core-periphery structure centered around five extremely popular countries (the UK, France, Germany, Italy, and Spain). In line with transactionalist theory, cross-border mobility and communication are strongly related to transnational attachment. Furthermore, we demonstrate that the network of transnational attachment is much denser among those with a higher than among those with a lower level of education. Our results suggest that offering European citizens incentives to travel to peripheral countries may help counterbalance the current asymmetric structure of transnational attachment, thereby increasing Europe’s social cohesion.

New Working Paper available online

29/1/2015

A draft version of my paper The Spatial Structure of Transnational Human Activity is now available online in the arXiv.

Starting from conflictive predictions of hitherto disconnected debates in the natural and social sciences, this paper contains the – to my knowledge – first encompassing analysis of the spatial structure of human activity beyond nation-state borders worldwide (a) across eight types of mobility and communication and (b) in its development over time. It is shown that contrary to what popular terms like “death of distance” and “end of geography” suggest, the spatial structure of transnational human activity is still heavily shaped by physical space. I demonstrate that the relation between activity and distance follows a specific mathematical pattern that can also be found in the movements of many animal species as well as local-scale human motion: Lévy flights with heavy tails that obey power laws. This pattern is also shown to remain remarkably stable over time. The figure on the right shows what a Lévy flight typically looks like: the path of a walker who takes many small and few long steps and changes his direction randomly in between.

A Lévy flight

The map below shows exemplarily one type of communication that is analyzed in the paper: transnational Facebook friendships. The black lines denote the country pairs with the highest number of Facebook friendships. For the purposes of this study, they constitute the equivalent of the steps the random walker takes in the Lévy flight figure above. (NB: The Facebook data was obtained from an interactive graph, converted into a network matrix and graphically mapped using Manish Nag's fantastic free software SONOMA).

A Map of Transnational Facebook Friendships

The third figure below shows the spatial structure of these Facebook friendships. The x-axis shows the distance (in km) and the y-axis states the probability of a transnational Facebook friendship to occur. The blue circles denote binned observations and the black line is a fitted power-law curve. The inset shows the same observations on logarithmic axes, on which the power-law curve forms a straight line.

The Spatial Structure of Transnational Facebook Friendships

As both the graph itself and the goodness-of-fit measure R² show, the power-law fits the observations almost perfectly. In the logged version, the observations cluster closely around the power-law line and the R²=.968 is only slightly below the perfect fit of R²=1.
This finding (which also holds for other types of mobility and communication under study) is clearly at odds with the common notion of a "death of distance" or an "end of geography" in structuring social relations.

For more details just have a look at the paper.

Paul Valéry, the first social network analyst?

29/1/2015

While it appears to be relatively common for social scientists to reconstruct social networks of characters within literary works (see for example the network study of the characters in Dante's inferno by Amedeo Cappelli and colleagues or the social graph of the characters in the Marvel comics universe), or to analyse the real-world social networks of authors (see for example Adam Obeng's work on William Godwin's acquaintance networks), to my knowledge, little attention has been given to explicit or implicit occurences of social network analysis within the substance of world literature itself. I just stumbled upon this interesting section in the French essayist Paul Valéry's 1896 La soirée avec M. Teste that describes the narrator's thoughts upon observing a theater audience (which evidently stands for society in general):

Chacun était à sa place, libre d’un petit mouvement. Je goûtais le système de classification, la simplicité presque théorique de l’assemblée, l’ordre social. J’avais la sensation délicieuse que tout ce qui respirait dans ce cube, allait suivre ses lois, flamber de rires par grands cercles, s’émouvoir par plaques, ressentir par masses des choses intimes, - uniques, - des remuements secrets, s’élever à l’inavouable! J’errais sur ces étages d’hommes, de ligne en ligne, par orbites, avec la fantaisie de joindre idéalement entre eux, tous ceux ayant la même maladie, ou la même théorie, ou le même vice...

This paragraph made me wonder whether Paul Valéry must be considered the first social network analyst. Isn't connecting people who have something in common via ideal lines while leaving those who have nothing in common disconnected exactly what network analysts do today? After all, network studies in epidemiology connect people suffering from the same maladie, studies of co-citation networks allow for following the bonds between people with the same théorie, and network analyses of mafia clans describe connections between those who have fallen for the same vice. Thus, in a way, Valery described as a fantaisie in 1896 what network analysts would actually do in the future, decades before Jacob L. Moreno's 1930s invention of sociometry and even longer before John A. Barnes' first mentioning of the term "social network" in 1954.

On a sidenote, Paul Valéry started writing La soirée avec M. Teste in 1895 in the same house (No. 9, Rue de la Vieille-Intendance in Montpellier) in which Auguste Comte, one of the founding fathers of sociology, was born a century earlier, in 1798. What a small world.

New Article Published in Journal of Conflict Resolution

22/8/2014

A new article, "Between Collaboration and Disobedience: The Behavior of the Guantánamo Detainees and Its Consequences", is now available online ahead-of-print in the Journal of Conflict Resolution. Based on a novel dataset created from Joint Task Force Guantanamo–authored memoranda on 765 detainees which had been purloined by Bradley Manning and published by WikiLeaks in April 2011, this study examines the behavior of the Guantánamo detainees in terms of collaboration and disobedience and how it influences their chances of getting a release recommendation. The centerpiece of the analysis is a network of incriminations between detainees. The figure below, which is taken from the article, shows a graphical representation of these accusations for a sample of detainees from the four largest national groups at Guantánamo (Yemenis, Saudi Arabians, Afghans, and Pakistanis). It indicates that Yemenis and Saudi Arabians heavily overcontribute regarding incriminating statements, whereas Afghans and Pakistanis undercontribute.

Incriminations between detainees from the four largest national groups at Guantánamo (Deutschmann 2014, p. 11)

In addition, the article reports that while a few detainees incriminate many others and many detainees incriminate only few others (the distribution of incriminating statements obeys a power law), the majority (62.6 percent) of all detainees do not incriminate anyone. Disobedient behavior does not affect the likelihood of getting a release recommendation, except for hunger striking, which has a negative effect. By releasing information, detainees don’t improve their own chances of getting release recommendations but impair those of the detainees they implicate. In the discussion at the end of the article, three different groups of detainees are identified (high-level-, low-level, and non-collaborators) whose behavioral patterns seem to follow distinct logics.

Key words: Guantánamo, terrorism, social network analysis, power law, mosaic theory, learned helplessness, power-dependence, WikiLeaks

About

New Article published at European Journal of Political Research

New Working Paper available online

Paul Valéry, the first social network analyst?

New Article Published in Journal of Conflict Resolution

Categories

Archives