### Replication Materials for Quantifying the Impact of Misinformation and Vaccine-Skeptical Content on Facebook
#### Experiment Materials (`experiment_information/`)
##### Survey Materials (`surveys/`)
- `vax_experiment_survey_flow_1.docx` : Qualtrics survey flow and questions for Study 1
- `vax_experiment_survey_flow_2.docx` : Qualtrics survey flow and questions for Study 2
##### Pre-Registrations (`pre_reg/`)
- `study1_prereg.pdf` : Pre-Registration for Study 1
- `study2_prereg.pdf` : Pre-Registration for Study 2
##### Stimuli (`stimuli/`)
- `study1/`: Folder of png files containing screenshots of Study 1 content
- `study2/`: Folder of png files containing screenshots of Study 2 content
#### Data (`data/`)
- `experiment1_data_cleaned.tsv`: Study 1 Cleaned Individual-Level Data
- `experiment2_data_cleaned.tsv`: Study 2 Cleaned Individual-Level Data
- `estimates_labels_filt.csv` : Combined treatment effects for Study 1 and 2 and Content / Crowdsourcing Labels
- `total_impact_by_cluster.csv` : Total Views, Total Impact, and Misinformation Rating for URL Clusters
- `total_impact_by_domain.csv`: Total Views, Total Impact, and Impact Per URL for all domains
- `url_impact.csv`: URL IDs, domain, predicted treatment effect, view count, fact-check-rating, total-impact, and predicted scores for each URL (NOTE: not currently available, please contact jnallen@mit.edu for instructions on how to generate data from `url_rids.csv` once access to Meta Social Science One dataset is granted)
- `url_rids.csv` URL IDs from Facebook Social Science One Dataset
##### Intermediate Data (`intermediate_data/`)
- Intermediate data files from analysis script
#### Domain Lists (`domain_lists/`)
- `experiment_low_quality_domains.txt` : Domains used to identify low-quality vaccine URLs from CrowdTangle for Study 2. Taken from iffy.news (circa May 2022)
- `experiment_mainstream_domains.txt` : Domains used to identify mainstream vaccine URLs from CrowdTangle for Study 2. Taken from Pennycook and Rand (2020).
- `lasser_domain_list_clean.csv` : Domains used to label URls as high vs. low credibility in our Facebook dataset. URLs labeled as "unreliable" were labeled as "low credibility" in the paper; all other domains were labeled "high credibility". Taken from Lasser et al (2022).
#### Code (`code/`)
- `./run_analysis_generate_figs.sh`: Code to generate all Figures and In-Text calculations (will not run without access to Meta Social Science One data, email jnallen@mit.edu for access instructions).
- `figX.R` : Code to generate figure X (figures 2 and 4 are unable to run without access to Meta data).
- `sectionX_intext_calcs.R` : Code to generate all intext calculations in Section X (sections 2 and 4 are unable to run without access to Meta data).
