Wednesday, July 7
8:45 am - 9:00 am


  • Daniel Alexandrov (Speaker) Professor and Deputy Director, National Research University - Higher School of Economics - St. Petersburg


Authors: Daniel Alexandrov ; Valeria Ivaniushina; Vera Titkova

There is a long-term interest in the relationship between network structure, positions, behavior, success and health of students [Vaquero, Cebrian, 2013; Grund, 2014; Howell et al., 2014; Mojzisch et al., 2020]. Academic achievement depends both on network structure and students' centrality in networks [Rijsewijk et al., 2018; Saxena et al., 2019]. High degree centrality has both positive and negative effects at the same time: students are happier and more prosocial but in stress and relatively more aggressive [Howell et al., 2014; Bos et al., 2018]. There is an interaction between attitudes and centrality: students with a central position but no identification with their group are stressed [Howell et al., 2014]. The same place in the same peer network could have an opposite influence on different sides of students' well-being.

Our study focuses on network cohesion/modularity and individual network position as predictors of students' achievement and their sense of school belonging. We use various centrality measures: indegree, outdegree, betweenness, closeness and such measures as eigenvector centrality and hub/authority [Vignery, Laurier, 2020]; and main network characteristics: density, reciprocity, transitivity, number of cliques and modularity (based on greedy optimization of modularity), diameter – on class-level, and density, reciprocity, transitivity on clique-level.

This study's data has been collected in Saint-Petersburg and Moscow urban and rural areas. The sample is 15796 students nested in 2860 cliques in 827 classes from 250 schools. For the analysis we used (1) only most complete networks (>75% response rate); (2) only cliques with three or more students. For information on friendship the students wrote down the students' names from the class with whom they socialize most of all. Our data contains information on academic achievements, sense of belonging, and socio-demographic characteristics. Network structures and three-level HLM were assessed using R's software environment (igraph, sna, lme4).

The well-being of students in school depends on complex relational effects. Students have positive belonging and achieve academic success if their networks are well integrated on class level (closeness), and if they have friends with high eigenvector centrality. Positive effect of high network diameter and a negative effect of density on the GPA we interpret as the rich-club effect [Vaquero, Cebrian, 2013]. The adverse effect of such network structure and position is low level of sense of belonging. Similar adverse effect of lowering sense of school belonging is produced by strong division into groups (n of cliques, modularity). The analysis of hub scores produces counterintuitive results. Contrary to our expectations, the ties with authority peers are negatively related to GPA, but positively related to the sense of belonging.

Add to my calendar

Add to Google Add to Outlook (.ics)

Create your personal schedule through the official app, Whova!

Get Started