# Wise or mad crowds? The cognitive mechanisms underlying information cascades
This reprository contains data and code for "Wise or mad crowds? The cognitive mechanisms underlying information cascades" by Alan Novaes Tump, Tim Pleskac and Ralf H.J.M. Kurvers; Wise or mad crowds? The cognitive mechanisms underlying information cascades. Sci. Adv. 6, eabb0266 (2020). DOI: 10.1126/sciadv.abb0266
This reprository contains the following folders:
- `\Data` containing the file `sDDM_data.csv` and `Variable_description.txt` with the data and the description of the data, repectively.
- `\Package` containing a package with the social DDM code in C++ to run the model faster.
- `\Code` containing the code to reproduce all results
----------
The Code folder contains following subparts:
- The personal phase can be analysed with the 2DSD model with `2DSDanalysis.R`.
- The social-DDM analysis is conducted with `run_MCMC.R` which runs a hierarchical Differential-Evolution-MCMC algorithm. Afterwards the chains can be diagnosed with the `run markdowns.R`file.
- Linear model statistics are conducted with `Social-DDM_final_stats.R`, which also saves the results in LaTex tables.
- `all_plots_posteriour.R` produces all important plots (exept recovery analysis)
- Recovery analysis is performed with `Parameter_Recovery.R`.
- `\Simulations` runs agent-based simulations using the posterior estimates and plots the results
Subfolders with results:
- `\brms_models` containing results of the linear models
- `\final_plots` containing final figures for talks and paper
- `\final_tables` containing LaTex tables with the results of the social DDM and linear models
- `\mcmc_analyis` containing html files with results and chain diagnosis from the social DDM analysis
Importantly, to conduct the social DDM analysis please install the 'Powerfunction' package contained in the `\Package` folder with the file `Powerfunction_0.1.0.tar.gz`.
Other used packeges are: 'dplyr', 'tidyr', 'tidyverse', 'parallel', 'doParallel', 'compiler', 'Rcpp','reshape', 'ggplot2', 'gplots', 'cowplot','gridExtra','RWiener', 'rstan', 'brms' and 'knitr'.
If you have any questions or issues running the code, please do not hesitate to contact us via: tump@mpib-berlin.mpg.de
