Get set up

Everything here is free, and nothing needs an account.

You do not need all of it. Git and one way to analyse data are the working minimum; which editor is whichever one you already like.

Write the analysis

RStudio Desktop

The familiar one. Open a run folder as the working directory and everything in the starter scripts resolves. → posit.co/download/rstudio-desktop · free, AGPL-3

Positron

Posit’s newer editor — one place for both R and Python, VS Code underneath. Worth it if you move between the two languages during the jam. → positron.posit.co · free

You need R itself as well if you are running scripts outside the container: → cran.r-project.org

Or do not write it

Point-and-click statistics, for people who would rather not start at a blank script. All three are free and open source, and all three are real analysis tools rather than teaching toys.

jamovi

A statistical spreadsheet with R underneath. The useful part for the jam: it shows you the R syntax for every analysis it runs, so a point-and-click session can become a script you can record. → jamovi.org · desktop or in the browser

JASP

The same idea from the University of Amsterdam, and the one to take if you want Bayesian alongside frequentist — it puts both in reach of the same drag-and-drop. R packages do the work. → jasp-stats.org · University of Amsterdam

Orange

What jamovi and JASP are to R, Orange is to Python: you place widgets on a canvas and connect them instead of writing code, and it is Python all the way down for anyone who wants to extend it. Strong on visualisation, clustering and text mining. → orangedatamining.com · University of Ljubljana

One thing to know before you rely on them. The record the jam keeps is of a command run against files. An interactive session is not that: nothing re-runs a sequence of clicks, so a result produced only in a GUI cannot be handed to the other team to reproduce. Use these to explore and to decide what is worth doing — then take the syntax out (jamovi prints it; Orange can export a script) and record that. Exploring is free; the thing you defend has to be re-runnable.

Share the work

Git

Everything the jam produces is committed and pushed, so the other team can read exactly what you did and run it themselves. If you have never used it, that is fine — the three commands in the repository’s README are all you need. → git-scm.com/downloads

Docker Desktop

Optional, and worth it if you want the other team’s re-run to mean something. A rocker image is a pinned R — the same version and the same packages for everybody, now and in a year:

docker run --rm -v "$PWD:/work" -w /work rocker/r-ver:4.3.3 Rscript work/yourname-murder/analysis.R

Without it your analysis still runs; it just runs against whatever R you happen to have, and so does theirs. → docker.com/products/docker-desktop

Then

git clone https://github.com/gitmick/asif-jam.git
cd asif-jam

Everything is in there: every dataset in data/, already fetched, and three worked examples in examples/. Copy one into a folder of your own, with your name on it:

cp -r examples/murder work/yourname-murder

Each example folder is self-contained — its data is already in inputs/, and there is an .Rproj file, so in RStudio or Positron you open work/yourname-murder/murder.Rproj and everything resolves. Work down the chunks in analysis.Rmd, or run analysis.R for all of it at once; the figures and tables land in out/.

In jamovi, JASP or Orange, open the CSV from data/ directly instead.

Commit as you go and push, so the rest of the team can see what you did and the other side can re-run it:

git add work/yourname-murder && git commit -m "what I tried" && git push