This repository provides the material for a workshop on latent variable modeling (LVM) applied to human-coded data of reports on sexual violence in armed conflict (SVAC). This empirical work forms part of ongoing research by Ragnhild Nordås and Jule Krüger, with advice from Christopher Fariss.
The LVM approach we follow here builds on Schnakenberg and Fariss (2014), whose model is an extension of Treier and Jackman (2008).
We will be using the free and open-source R programming language in this workshop. To actively follow the workshop, edit R scripts, and compute the models, you need to have installed R and the RStan R package on your machine. Check out this RStan Getting Started Guide, as well as the RStan documentation. These two vignettes are also very useful for understanding RStan: (1) RStan: the R interface to Stan and (2) Accessing the contents of a stanfit object . The free and open-source RStudio Desktop interface is recommended for editing and testing R scripts before they are run from the command line.
The organization of this repository follows HRDAG's principled data processing guidelines. This includes the organization of tasks into separate directories (e.g., import/, estimate/, visualize/, present/, etc.), the division of task directories into input/, src/, and output/ directories, and the use of GNU Make to achieve a self-documenting workflow. To run a Makefile from your command line, navigate to the relevant task repository (e.g., "$: cd ~/git/SVAC-LVM-tutorial/import/") and execute "$: make -f Makefile". To run Makefiles on Mac OS, install the XCode Command Line Tools.
You won't be able to access the data when you download this repo to your machine via a browser. Follow this link to download the data (csv file) directly.