Pondering my defaults for Quarto websites and reproducible research projects

reproducible research
Quarto websites are a great way to package and distribute assets from a reproducible research project. I’m just about to start another project, and I’m thinking about issues with how I get started.

Matt Crump


June 27, 2023

This post is mainly a long ramble where I attempt to start working this morning.

I’ve been using Quarto almost exclusively for a while now, and it’s great. I ❤️ Quarto.

This morning I’m about to start another project, and TBH I’m procrastinating a little bit. I don’t have writer’s block. I know that as soon as I get the project started I’ll rattling off for the rest of the day, week, or however long. I have mild getting-started-with-a-new-project-block.

One of the things that drew me into Rstats, R Markdown, and now Quarto in the first place was the ability to collect nearly all of the research assets that I typically assemble in one place and to make most of them computationally reproducible, at least to a reasonable standard for my field. For example, a typical project might have some data or simulated data from a model, scripts to analyse the data, maybe some custom functions for the project, a research paper, maybe a slide deck or poster. All of these things can be housed in a Quarto project, and Quarto can handle most of the different outputs as well. It’s also pretty easy to put all of these things into a Quarto website and share everything on GitHub.

Before Quarto, my colleague Matti Vuorre and I wrote a paper for psychologists to consider using R Markdown and other R-centric tools to create and share reproducible research projects (Vuorre and Crump 2021). We called that approach Vertical. Now, I’m mostly using Quarto to accomplish the same goals.

Some quick examples of Quarto projects:

Overall, these are quick to set up using templates provided by RStudio. So, what’s my issue?

Like I said I’m procrastinating. What I should do is open RStudio, create a new Quarto website project, and then get on with life. The website won’t look quite right, and I’ll have to modify the _Quarto.yml file. I won’t be able to remember what I need to modify so I’ll look at previous examples and scrape something together. It will take probably all of 15 minutes for me to do this.

It would be nice if my previous self had set up templates that make it even easier and faster for me to get started. Part of the reason for this post was to consider whether I should make those templates right now.



Maybe I need to go through the process like I normally do, and keep notes on things that I might put in a template. Or, at least think about it.

Making of a Quarto project

The flavor of this project is computational modeling. Ideally, I would share the scripts and analysis, and some kind of global write-up, perhaps some slides.

This is the vanilla Quarto website project.

This is the default yml that controls the website build:

  type: website

  title: "GPTAttentionPerformance"
      - href: index.qmd
        text: Home
      - about.qmd

    theme: cosmo
    css: styles.css
    toc: true

Things I usually change later:

One issue is how to translate the documents in the project to pages on the website. This can be rearranged at any time using the _quarto.yml file, which defines website navigation and connects individual .qmd files to the navbar at the top or on the side. For example, the two pages in the default project are given links “home” and “about” at the top of the website.

In some previous projects I use a flexible starting position in terms of website structure. Quarto will automatically build sidebar links for .qmd documents. I think I’ll add that here as a starting point.

Something like this. Need to consider ways to template this for later. For now, I can copy and paste from here.

  type: website
  output-dir: docs
    port: 4200
    browser: true
    navigate: true

  title: "GPTAttentionPerformance"
  description: "A computational modeling project examining whether or not a GPT model can simulate human performance in classic attention and performance tasks."
  repo-url: https://github.com/CrumpLab/GPTAttentionPerformance
  repo-actions: [source, edit, issue]
  site-url: "https://crumplab.com/GPTAttentionPerformance"
  search: true
      - href: index.qmd
        text: Home
    - icon: github
        href: https://github.com/CrumpLab/GPTAttentionPerformance
        aria-label: GitHub
    - title: "Home"
      style: "floating"
      collapse-level: 3
      align: left
      contents: auto

    theme: minty
    css: styles.css
    toc: true

The landing page, defined in index.qmd, might start out like this:

title: "Simulating Attention and Performance tasks with an LLM"
  - name: Matthew J. C. Crump
    orcid: 0000-0002-5612-0090
    email: mcrump@brooklyn.cuny.edu
      - name: Brooklyn College of CUNY
description: "A computational modeling project examining whether or not a GPT model can simulate human performance in classic attention and performance tasks."
date: 6-27-2023
title-block-style: default

## Project Information

Brief notes on the project

## Repository Information

Brief notes about the repo

The next step is to make new folders and add .qmds to them. I made a modeling folder, and added a .qmd that will act as a scratchpad/to-do list for me. Now any page in the modeling folder will automatically show up in the left-side navigation.

This is good enough for now. I’m going to pass on making templates. I think I’m still in the stage of figuring out what I want by making projects from scratch. Hopefully I will really like one of these project structures, and then I can make a template out of it later.

Time to get going on this actual project!


Vuorre, Matti, and Matthew J. C. Crump. 2021. “Sharing and Organizing Research Products as R Packages.” Behavior Research Methods 53: 792–802. https://doi.org/https://doi.org/10.3758/s13428-020-01436-x.