Seminars: 2015 Winter / Spring


2015 Winter / Spring


András  Vörös (RECENS, University of Oxford): Dimensions of peer perceptions in Hungarian high schools

Psychologists have long been interested in how the behavior of children is related to their social success (or failure) in peer groups. They realized early that peer perceptions play a crucial role in this phenomenon: how children interpret and evaluate the behavior of others in their group will determine how they interact with each other, who becomes popular, who will be rejected, bullied, and so on. Several measures of peer perceptions of behavior have been designed over the past six decades, which collect data in a large number (30-50) of behavioral dimensions. The analyses of these tests led to the identification of typical “behavioral profiles” that are associated with popularity, rejection, and other forms of sociometric status in school. It can be accepted that dimension reduction based on the number of nominations received by students in each behavior item suits the aim of the psychologists interested in individual outcomes (sociometric status). However, such an approach simply ignores most of the structure in the rich multiplex network data at hand, obscures the impact of community structure (e.g. the presence of cliques) on the results, and thereby potentially leads to unreliable or false interpretations.
Drawing on existing statistical methods and measures, we propose a procedure to reduce the dimensions of multiplex network data measured in multiple groups. We achieve this by clustering the network dimensions using their pairwise similarities and constructing composite network measures as combinations of the network dimensions in each resulting cluster. In case an appropriate similarity measure is chosen, our approach is able to produce a small number of networks which represent the original pool of relations directly on the level of ties. The procedure is demonstrated on a longitudinal dataset of 21 perception network dimensions in 18 Hungarian high-school classrooms. The results indicate that the network items organize into three well-interpretable clusters: positive, negative, and social role attributions. While actual peer-perceptions are dynamically evolving in the observed period, the cluster solutions at the first and second time points are essentially identical. Finally, we show that the composite networks defined on the three relationship groups overlap but do not fully coincide with more standard network measures in adolescent research, such as friendship and dislike. This suggests that the fine-grained measurement of social relations in school, combined with dimension reduction techniques appropriate for multiplex data, can help to advance our understanding of group processes in adolescent communities.



László Lőrincz (MTA KRTK KTI, RECENS): Inter-ethnic dating preferences in secondary school

Romantic relationships in adolescence are sources of social influence considering educational achievement and delinquent behavior. Integrated schooling is known to induce inter-ethnic friendship relations, however, it also creates the opportunity of inter-ethnic dating. Based on contact theory, inter-ethnic personal relationships or long-term exposure decreases ethnic prejudice, thus it is proposed that willingness to date between ethnic groups may increase. The question arises, whether in the school context exposure is enough for this mechanism to work, or personal contact is necessary. It must be also taken into account, that romantic relationships are embedded in status relations within schools. Previous studies on intermarriage and homogamy found a “social exchange” mechanism, that lower status members of majority groups are more likely to choose minority partners. Translated to the adolescent society, it is assumed, that the less popular members of the majority groups are those, who are more willing to form inter-ethnic relations. To address the above questions empirically, the first wave of the Hungarian network panel "Wired into Each Other” was used, containing data of 1214 9th grade students in 43 classes of seven secondary schools. Inter-ethnic dating preferences of Roma and non-Roma students were measured by dyadic attribution of physical attractiveness (liking), and nominations of willingness to date between class members. Statistical analysis was carried out using multilevel p2 models.



Melinda Ratkai (University of Málaga): The Facebook model of opinion dynamics and emotional contagion

Opinion dynamics in our age takes place at social media sites and is interrelated with emotional contagion (Kramer et al., 2014). Models of opinion dynamics, even if they took into account interpersonal attraction and emotions, have considered pairwise interactions that take place randomly or in a pre-specified network. We present a new opinion dynamics model that fits much closer to actual features of opinion dynamics on social media, considers opinion walls instead of pairwise interactions and takes account of filtering that is a new feature of social media sites, such as Facebook. We find an answer to the problem of classical opinion dynamics models about how polarization of opinions can be stable. We demonstrate that polarization of opinions is a result of emotional contagion and filtering on these sites. Our model is also capable of describing the conditions that assist the survival of extreme opinions and views.



Michał Bojanowski (University of Warsaw, RECENS): Composition and structure of networks in Polish school classes: a multilevel perspective

We study the structure of networks among Polish 6-year-olds in a sample of 176 school classes. We look into explaining the structure of networks within classes using, among other things, mechanisms of homophily (e.g. according to gender) and hierarchy (e.g. with respect to social status of parents) as well as selected structural effects of ERGM models. Results show substantial variability between classes in terms of network structure. To explain these differences between school classes, a multilevel model is used to asses the importance of compositional and contextual (e.g. school neighborhood) factors.



Sándor Soós (MTA KIK TTO): Scientometrics as network science: the latent face of a misperceived research field

A common (mis)perception within the academia pictures scientometrics as a “quantitative ritual” of research evaluation, or, the measurement of scientific performance, at best. In contrast with this attitude, scientometrics as a research field is defined as the quantitative study of scholarly communication, or, in the broad sense, the empirical study of the cognitive and institutional organization of science. Modelling this phenomenon, among the dominant methodological frameworks of the state-of-the-art is the theory and analysis of complex networks. The main aims of this presentation are (1) to provide an overview of the various models of social and information networks applied in scientometrics, with a special emphasis on the less salient types called “science maps”, and (2) to show how evaluative scientometrics – that is, seeking for proper measures of research performance – is fundamentally linked to this hidden subfield, driven by network science, and sometimes called “structural scientometrics”.



Károly Takács (MTA TK "Lendület" RECENS): No Sword Bites So Fiercly as an Evil Tongue? Gossip Wrecks Reputation, but Enhances Cooperation

Social norms in general, and norms of cooperation in particular, are the cement of all human societies. For the difficult problems of the maintenance and enforcement of social norms and of cooperation, humans have developed surprisingly complex solutions. Reputation mechanisms and gossip are certainly among the compound informal solutions.
According to common wisdom, gossip channels mainly negative and often fictitious information. If it is the case: how can dishonest gossip and the resulting biased reputations legitimize social order and promote cooperation?
This is the main puzzle we tackle in the this project exploiting a wide set of instruments. We use analytical modeling and agent-based simulation to derive hypotheses. We test simple hypotheses in small group experiments. We develop new methodological tools to appropriately analyze the triadic nature of gossip embedded in network flows of information. We utilize dynamic network datasets from primary and secondary school classes, and we gather qualitative and quantitative information from organizations to test conditional hypotheses about the role that gossip plays in reputation and cooperation in different developmental and social contexts of life. In addition, we apply new communication technologies currently under development to explore the hidden world of gossip and the dynamics of reputation in dormitories and organizations.
With the insights gained, we can overcome common stereotypes about gossip and highlight how gossip is related to credible reputational signals, cooperation, and social order. Expected results will help us to outline the conditions that can promote cooperation in work groups, and they will help to construct successful prevention strategies for social exclusion and for other potentially harmful consequences of the evil tongue.
The presentation is followed by a reception.