- Python 100%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| code | ||
| README.md | ||
Structural Fragility in Digital Public Health Communication
This repository is reserved for the analysis code and reproducibility materials associated with the manuscript:
Structural Fragility in Digital Public Health Communication
Study overview
This study examines structural fragility in digital public health communication using annual Twitter/X retweet networks related to U.S. e-cigarette prevention campaigns from 2014 through April 2023.
The analyses assess three dimensions of structural fragility:
- Community fragility: fragmentation, modularity, largest connected component, assortativity, and within-category retweeting
- Positional fragility: amplification, in-strength, PageRank, eigenvector centrality, and concentration
- Core fragility: deepest k-core size and actor-category composition
Accounts were classified into five mutually exclusive categories:
- Health
- Commercial
- Vape Community
- Organic
- Other
Planned repository contents
Following completion of internal review, this repository will include final scripts for:
- construction of annual retweet networks
- computation of network-level and category-level metrics
- threshold-sensitivity analyses
- attribute-permutation null models
- degree-preserving configuration-model analyses
- longitudinal trend tests
- generation of manuscript figures
- generation of supplementary tables
- documentation linking each script to its corresponding manuscript output
- software dependencies and execution instructions
Only final, reviewed scripts and approved reproducibility materials will be retained in the public repository.
Data availability
The underlying Twitter/X data cannot be shared in raw or post-level form because of privacy protections, institutional requirements, and platform terms governing the redistribution of usernames, account identifiers, and tweet content.
Where permitted, the repository will include aggregate or derived materials necessary to document the analyses without disclosing identifiable account-level information.
Questions regarding data access may be directed to the Social Data Collaboratory at NORC at the University of Chicago.
Software
The analyses were conducted primarily in Python 3.10 using packages including:
- NetworkX
- python-louvain
- python-igraph
- pandas
- NumPy
- SciPy
- statsmodels
- Matplotlib
Exact package versions and environment specifications will be included with the final release.
Archiving and versioning
The version of the code corresponding to the article will be permanently archived on Zenodo:
Reserved DOI: https://doi.org/10.5281/zenodo.21779470
The Zenodo record is currently an unpublished draft. It will be populated and published after completion of code review and institutional approval.
Subsequent development or maintenance updates may appear in this GitHub repository. The Zenodo release will preserve the fixed version associated with the published article.
Citation
Citation information will be added after the manuscript and Zenodo record are finalized.
Contact
Miao Feng
Social Data Collaboratory
NORC at the University of Chicago
feng-miao@norc.org