Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Creating living documents and reproducible reports with
 R markdown and Jupyter notebooks


Curious about how living documents and reproducible reports could help your research? This repo contains a workshop walkthrough about how R markdown and Jupyter notebooks can enrich your research workflow.

What are "living documents" and "reproducible reports"?

While everyone seems to have their own take on what these two terms mean and how they differ from one another, "living documents" and "reproducible reports" are ways for researchers to share code, images, and text in a single document.

For my part, I view "living documents" as work-in-progress documents. They're great for keeping notes on your data cleaning, data processing, and data analysis by allowing you to add plain text, plots, and live code in a single place. As researchers, we might spend months away from a project (while we're busy with something else or while we're waiting for reviewers to get back to us). When it comes time to start up the project again, living documents can help us jump back into the project quickly: Taking good notes about what our code does -- right next to real code -- can help us remember exactly what we were thinking and why we made the choices we did.

"Reproducible reports," on the other hand, I see as documents that are meant to publicly accompany your research outputs (e.g., talks, posters, journal articles). These are ways for other researchers (and maybe even your reviewers) to see all of the work that you did when handling your data and creating your analyses. Given ongoing concerns about transparency and reproducibility in a variety of fields (including psychology and cognitive science), using reproducible reports can provide vital information about the data cleaning, processing, and analysis that supported your conclusions.

"Reproducible manuscripts" are a specific type of reproducible report---one that has as its output a submission-ready manuscript. For a reproducible manuscript, you include code "chunks" or "blocks" that run your data analysis behind the scenes, along with the actual manuscript text (including figures, tables, and even references). Mastering the reproducible manuscript means no more tedious (and error-prone) copying and pasting of your statistics---they'll all recompile and render perfectly each time you compile your manuscript.

Why should I care?

For science

Researchers -- especially within cognitive science and psychology -- are increasingly interested in promoting transparency and reproducibility. There are badges that researchers can earn for sharing their data and materials that promote the prominence of open science, and some journals even require data and code sharing now.

Providing an explicit accounting of your data and code practices can help demonstrate the value of your work. If you share your data and your code, you're promoting immediate computational reproducibility by anyone (especially if you use open-source programming languages and packages). As an added bonus, if there's future interest in directly or conceptually replicating your work in new experiments, providing your code openly can help those future replication efforts use methods as close to your original work as possible.

By using reproducible manuscripts, you'll take the added step of helping minimize transcription errors in your manuscript's statistics. Given the surprisingly high rate of statistical errors in scientific articles generally and psychology articles specifically (see Bakker & Wicherts, 2011, Behavior Research Methods), using reproducible manuscripts can help prevent simple errors in transcription from entering the scientific field---especially ones that (as noted by Bakker & Wicherts) could lead to qualitatively different results.

For yourself

Think of this as your present self doing something nice for your future self. If your present self takes a few minutes to add some explanatory text, code comments, or a useful plot, you'll be saving your future self headaches and time. Present-you knows what you're doing because present-you is ankle-deep in things. Future-you, on the other hand, will probably have spent weeks or months away from the problem and will have to spend valuable time puzzling through the traces that then-past-you created. Do future-you a favor!

Using reproducible manuscripts takes that one step further. When reviewers come back with requests for an update to your data pipeline or a new study to add your manuscript, you don't have to worry about recalculating and painstakingly copying-and-pasting all of your descriptive statistics and results back into your manuscript. You can get right to addressing those reviewer concerns and get that revision out faster.

For yourself and science

A transformative way to think about this is to see that the effort you put in for helping your future self can be equally valuable for helping the broader research community engage with and make sense of your research.

With just a little bit of additional effort, you can tweak your living documents into reproducible reports. If you're taking good notes and adding comments to your code in your living document while you're doing the research, all you need to do is publish the document after you're done!

Workshop materials: A guide

To run the workshop materials, just click on the "launch binder" button at the top of the README file. Binder is a way of converting a GitHub repository into a cloud-based executable instance, complete with all of the files and data in the repository. Just click the button, and you'll be able to start working on, executing, and editing the code immediately---no installation required! Be patient the first time you launch it; it may take a few minutes to be ready for you to start.

Once your Binder instance is started, you can navigate the directory in the browser window much in the same way that you can navigate files on your local computer. You can click file or folder names to open them. More on opening specific files types are included below in the instructions for each of them.

You're able to come back and launch a new Binder instance of the code any time you'd like. Keep in mind that none of your changes will be saved after the Binder instance is closed; each new Binder instance will only load with the files and specifications exactly as they are in the repository.

However, if you'd like something more permanent, feel free to fork the repository or download the files. The beauty of R markdown and Jupyter notebooks lies in their flexibility -- so experiment until you find what works best for you!

R markdown (directory: rmarkdown)

To run the R markdown files, you'll need to start RStudio in the environment. If you're not familiar with RStudio, it's an integrated development environment or IDE that facilitates programming in a more user-friendly setup. (If you're familiar with MATLAB, RStudio will give you a very similar programming experience in R to the one that you're used to seeing in MATLAB.)

To start RStudio, click the "New" button in the upper righthand corner. Select "RStudio" from the drop-down menu that appears. A new tab will open within your browser that includes RStudio. You can open files by selecting the "File" menu in the top-left corner, selecting "Open", and navigating to the appropriate file. When you have this file open, it will open a new pane in the top-left section of the RStudio window. This new pane is called the source pane, and it will be where you can write and save your code. Below the source pane is the console pane, where you can run code. The top-right pane is the environment pane, which shows the variables and custom functions specified in your directory. The lower-right pane is the files/plot/help pane, which serves a variety of functions: showing files in the current directory, displaying plots, and rendering help text (depending on which tab you've selected). For more on RStudio, check out the "Further reading and more examples" section (or just play around---remember, no matter what you do, you can't permanently ruin a Binder instance, since you can just get a new one).

In the rmarkdown directory, you'll find three files:

  • rmarkdown-basic.Rmd: Start here. Playing around with this file will give you a very basic introduction to the R markdown format and its components.

  • rmarkdown-fake_experiment_data_analysis.Rmd: Once you're done with the basic introduction, try this more realistic (but still toy) R markdown. It will demonstrate how you might structure a reproducible report.

  • r_library_installation.r: This file is called by the toy reproducible report to install required libraries.

Reproducible manuscripts (directory: reproducible_manuscript)

We'll be using RStudio again for our reproducible manuscripts, so follow the instructions above to open the files through RStudio. I would strongly recommend working on the basic R markdown before starting this section, unless you are already familiar with R markdown.

In this directory, you'll find a number of files, including:

  • prep_for_reproducible_manuscript.R: Open and run this script in its entirety before beginning the rest of the reproducible manuscripts tutorial. You will not be able to proceed with this tutorial without running this script first. This script will download papaja and set up the TeX distribution with tinytex. You can open it in RStudio, highlight everything in the file, and click "Run" (or press command+enter [Mac] or ctrl+enter [PC/Linux]). Remember: You must run this each time you open a new Binder instance, since it does not permanently store the changes that you make to the instance.

  • reproducible_manuscript.Rmd: After you've run the prep_for_reproducible_manuscript.R file for the first time since opening your latest Binder instance, this R markdown will provide an example of a reproducible manuscript written in R with papaja. When you compile it, you will see a few new files with the same name (reproducible_manuscript) but with new extensions. These files will include LaTeX log files, intermediary manuscript files, and (given the specifications provided in our example) a final PDF of the manuscript and figures (already included in this repository).

  • references.bib: This is a BibTeX-formatted file of references for the reproducible manuscript. As with LaTeX, you may include more references in the .bib file than you cite in your manuscript, and only those that you cite will be rendered in the reference list. (If you're not used to using BibTeX for reference management, Google Scholar offers it as an option when you click the "Cite" button for a Google Scholar entry; just look for the "BibTex" link at the bottom of the pop-up.)

Jupyter notebooks (directory: jupyter-notebooks)

Running the Jupyter notebooks is much more straightforward than opening the R markdown files. To open, simply click on the name of the file, and it will open the Jupyter notebook in a new tab.

In this directory, you'll find three files:

  • jupyter_notebook-basic.ipynb: Start here. Playing around with this file will give you a very basic introduction to the Jupyter notebook format and its components.

  • jupyter_notebook-fake_experiment_data_generator.ipynb: Once you're done with introduction, try this more realistic (but still toy) R markdown. The demonstration will walk you through using the notebook by generating toy data (provided also by default in the data/ directory) used by various R markdown files.

  • A standalone Python script that is used in the basic Jupyter notebook.

Further reading and more examples


How to use Jupyter notebooks and R markdown to create living documents and reproducible reports.








No releases published


No packages published