Guilbeault, D. & D. Centola. 2021. Topological Measures for Identifying and Predicting the Spread of Complex Contagions. Nature Communications.

This repository contains:

• Raw edgelists of Addhealth topologies (82 adolescent friend networks).
• The largest connected component extracted from each Addhealth topology.
• An R script that replicates all of the figures from the manuscript and supplementary material using the raw simulation data generated for this study (data provided below).
• An R script that (i) calculates complex path length and complex centrality for all nodes in a given graph g (including methods for parallelization), and that (ii) uses these functions to replicate our main results on simulated and empirical graphs (from the Addhealth dataset).

Features in development:

• A Python code base for reproducing all of the measures defined in Guilbeault & Centola (2021). In the long term, these measures may be released as a package in Python and/or R.

Data:

• Here is a link to all of the raw data used to generate the main and supplementary figures for this study: https://drive.google.com/file/d/1M9sO51Viv9h6e-p83Q6FThOiYZZpdUhJ/view?usp=sharing
• The topologies and household-level statistics for the Banerjee et al. (2013) analysis were downloaded from the following data repository (https://osf.io/qu57w/); this accessed repository made use of the Banerjee et al. (2013) data for “Study 2” in the following article: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0245476

Any questions regarding this code or data can be sent to douglas.guilbeault[at]haas.berkeley.edu.