Fragmentation of a longitudinal population-scale social network: Decreasing network closure in the Netherlands

TL;DR

This study analyzes population-scale longitudinal network data for the Netherlands (2010–2021) across five edge types to quantify social cohesion via excess-closure in ego networks. Despite a near-stable average degree, average closure declines by approximately $15.5\%$, indicating growing fragmentation driven by ego-network rewiring rather than demographic shifts. Decomposition and ego-timeline clustering reveal that individual-level changes in multiplexity and geographic dispersion are the primary drivers, with a notable relocation paradox where moving temporarily increases local closure but long-run dispersion reduces it. The findings imply policy and planning considerations around housing and urban structure, and they highlight the potential tension between locally dense networks and broader social cohesion in increasingly mobile and digitally mediated societies.

Abstract

Population-level dynamics of social cohesion and its underlying mechanisms remain difficult to study. In this paper, we propose a network approach to measure the evolution of social cohesion at the population scale and identify mechanisms driving the change. We use twelve annual snapshots (2010-2021) of a population-scale social network from the Netherlands linking all residents through family, household, work, school, and neighbor relations. Results show that over this period, social cohesion, quantified as average closure in the network, declines by more than 15%. We demonstrate that the decline is not due to changes in demographic composition, but to rewiring in individual ego networks. Statistical models confirm a decreasing overlap of social contexts and greater geographical mobility as drivers. Residential relocation, however, temporarily increases closure, suggesting that local cohesion-seeking behavior can yield global network fragmentation, with implications for policies related to housing, urban planning, and social integration.

Figures (8)

• Figure 1: Temporal evolution of the nationwide Dutch multilayer population-scale network, 2010--2021. (A) Number of nodes. (B) Number of edges. (C) Average degree. (D) Average local clustering coefficient. (E) Average excess closure. (F) Share of multiplex ties. (G) Spatial dispersion in terms of the average alter distance (in km) and the share of alters in same municipality.
• Figure 2: Relative contribution to decreasing closure. Signed relative contributions to the average change in excess closure from $t-1$ to $t$ of the group, individual, and interaction terms (A) by age group, (B) migrant status.
• Figure 3: Four largest clusters of ego network degree and closure trajectory shapes (indicated by C1--C4 and colors). (A) Normalized degree ($k_{u,t}^{norm}$) timelines. (B) Share of sample belonging to the normalized degree timeline clusters. (C) Normalized closure ($C_{u,t}^{exc,norm}$) timelines. (D) Share of sample belonging to the normalized closure timeline clusters. (E) Cross-membership between degree (rows) and closure (columns) clusters.
• Figure SI1: Pearson correlation matrix of network, spatial, and demographic variables used in regression models. The heatmap displays pairwise correlations for 15 variables measured across 50,000 individuals over 12 years in the panel database (2010--2021).
• Figure SI2: Distribution of years in sample. Histogram showing the count of individuals by the number of years they appear in the longitudinal sample (2010--2021). The majority of individuals ($N\approx 26,918$) are present for all 12 years.