/Achievement_Gap_PISA

Primary LanguageTeX

An analysis of the achievement gap between top-bottom SES students over time.

In this repository I keep the files for the second chapter of my PhD thesis.

Below is an explanation of all the files in the repo.

  • The main files are .paper/Paper.Rnw and transform_data.R. The first document contains all the text and analysis of the paper and the second one downloads, manipulates and saves the data.

  • If you open the *.Rproj file you should open the previous two scripts (although I simply load the data produced by transform_data.R and don't run it in the .Rnw file. Check transform_data.R for details on how I download the data)

  • The data used in the analysis is in the folder ./paper/data (not in the Github repo, but it contains all data created from transform_data.R) and it is called pisa_listcol.Rdata because it is actually a tibble with a list column containing the six PISA waves that will be used. This file is nearly 9GB.

  • All PISA waves I use are taken from Przemyslaw Biecek and his PISA*lite packages, except for PISA 2012 and PISA 2015, which I download in the transform_data.R script (as temporary files) and store everything in pisa_listcol.Rdata.

  • All other .Rdata in .paper/data belong to estimations that take too much time to compute, so I saved provisional versions in this folder. For more details, see the modeling chunk in the Paper.Rnw.

  • Inside ./paper/ there is a packrat document containing all the packages used in this analysis together with it's versions.

  • Don't worry about overloading Github because I excluded both the cache and data folders in .gitignore

  • The playing_around folder has a script I was playing around. See comments inside script for details.

  • Finally, the paper should run seamlessly by compiling Paper.Rnw with Sweave and a pdf file should be saved in the ./paper folder. Without messing up the cache chunks, it currently takes around 4 minutes to compile.