Prepare_and_run_replication

Prepare and run the replication

• Carefully read the README provided with the replication materials.
• Follow the instructions for running the code.

Config file

It will be useful to include the config file provided with the LDI replication template files (see the example in the template):

• Depending on the software being used, create a configuration file with the following names:

  • Stata: config.do
  • R: config.R
  • Matlab: config.m

⚠️ You must change the name of config file from template-config.do to config.do and delete unnecessary lines!

• Add the basepath for the replication materials

For Stata, remember to use forward slashes ("/"), even on Windows. The R function file.path() will do this automatically.

  Stata: global pwd "/path/to/base/directory";
  R:     pwd <- file.path("path","to","base","directory")
  Matlab: (NEED EXAMPLE)

• Check if the author's programs create a logfile (i.e. are there .log files in the deposit?). If not, add sample code to create such a logfile to the "config" file (the template config file already contains code for this):
• Stata:

        local pwd : pwd
        local c_date = c(current_date)\
        local cdate = subinstr("`c_date'", " ", "_", .)\
        log using "`pwd`/logfile_`cdate'.log", replace text

The replication itself

• Continue to follow instructions in the README or the main program.
• Whenever the authors use a filepath, use the basepath macro you created in config.do
  • For this to work, you need to call your "config.do" file within the author's script.
• Stata example: the line

      use "C:\Documents\Author\data\thisfile.dta"

becomes

      do "config.do"
      use "${basepath}/data/thisfile.dta"

• R example:

      source("config.R",echo=TRUE)

• Matlab :

      NEED EXAMPLE

• Commit all the changes you made to make the code work: Example of commands::

    git status
    git add *log
    git add programs/table1.do
    git add programs/figure10.do
    git commit -m "All the changes I made to make the code work on the system"
    git push origin master

• Also: Important to keep note in REPLICATION.md about whether you change the directory, or any issue/progress you make on the replication.

Capturing output

Programs may generate output in a variety of formats. Typing

git status

will show you any new files that were not previously there, or that were modified, after each program run.

Datasets

Some programs generate datasets. You should treat those as any other datasets as well - i.e., if your instructions are to NOT save them, then don't.

Tables

Tables may be output as csv files, Excel files, Latex files, etc. You should git add these, as they are evidence of your replication, and your assessment in the REPLICATION.md.

Figures

Figures may be output in many ways, including on-screen only.

• See if PDF, PNG, EPS, or JPEG files are output by the program
• If not, see if the on-screen display in the program (Stata, Matlab, etc.) has an option to save, and save in PNG or PDF format
• If all else fails, do a screenshot, and save the screenshot (possibly cropped), with the name of the figure it is supposed to represent (e.g., Figure1.png).
• All figures should be placed in the Bitbucket repo through a git add, git commit, git push sequence.

You are not quite done yet

▶️ Write your replication report.