School choice: Digital prints and network analysis

Abstract

We apply social network analysis to examine school choice in the second-largest Russian city Saint-Petersburg. We use online data (“digital footprints”) of between-schools comparisons on a large school information resource shkola-spb.ru. This resource allows to identify clusters of city schools that have been compared to each other more often and thus reflect choice preferences of students and parents looking for a school. Network analysis is conducted in R (‘igraph’ package). For community detection, we employed fast-greedy clustering algorithm (Good et al. 2010). The resulting communities (school clusters) have been placed on a city map to identify territorial patterns formed according to choice preferences. Network analysis of the district school networks based on between-schools online comparisons reveals two main factors for community formation. The first factor is territorial proximity: users compare schools that are relatively close to each other and not separated by wide streets, parks, industrial areas, rivers, etc. The second grouping principle is the type of school: private schools always form a separate cluster which shows that they are not being compared with public schools. In one district there was also a cluster of elite or academically challenging public schools grouped together.