Reproducible Research

Dr. Ajay Kumar Koli \(\cdot\) 2nd SARA Bootcamp

Our Common
Research Frustrations

Image source: R. Azad at Unsplash

Repetitive Data Cleaning


  • Spending hours manually cleaning survey data in Excel

  • Making the same corrections across multiple data sets

  • Risk of human error in copy-paste operations

Version Control Nightmares


  • Files named: “Analysis_final.xlsx”, “Analysis_final2.xlsx”, “Analysis_FINAL_FINAL.xlsx”

  • Forgetting what changes you made between versions

  • Collaborators overwriting each other’s work

Lack of Reproducibility


  • How did I generate this graph six months ago?

  • Can’t replicate your own analysis for revisions

  • Reviewers asking for different analysis - starting from sratch

Limited Visualization Options


  • Stuck with basic Excel charts

  • Spending hours formating graphs manually

  • Can’t create interactive or dynamic visualizations

Time-Consuming Report Writing


  • Copy-pasting results into Word/PPT documents

  • Updating 50 tables when data changes

  • Inconsistent formatting across documents

The Magic of Quarto

Artwork source: Allison Horst

Automated,
Reproducible Research with Quarto

Artwork source: Allison Horst

Reproducible Research


“Reproducible research is when someone else (or your future self) can take your data and code, run it, and arrive at the same results you reported.”

AI generated information

Why Should Researchers Aim for It?

  1. 🔍 It builds trust: If others can verify your work, your findings carry more weight. Science runs on trust — reproducibility is how you earn it.

  2. 🐛 It catches mistakes: When you document every step, errors are easier to spot — by you or by reviewers. Many high-profile retractions could have been avoided with reproducible practices.

  3. 🕐 It saves your future self: Come back to a project after 6 months and you’ll thank yourself for writing everything down. Undocumented analysis is a nightmare to revisit.

  4. 🤝 It enables collaboration: Teammates and collaborators can pick up exactly where you left off — no guessing, no “it works on my machine” problems.

  5. 🌍 It advances science faster: When research is reproducible, others can build on it confidently instead of wasting time redoing or questioning your work.

  6. 📋 It’s increasingly required: Journals, funding bodies, and institutions are now requiring reproducibility as a standard. Getting ahead of it is a career advantage.

AI generated information

The Problem:

  • We run analysis in SPSS, create charts in Excel, copy-paste everything to Word

  • Your boss asks for changes - you redo everything manually

Quarto Solutions

  • Your research narrative (written text)

  • Live Code that runs analysis

  • Automatically generated tables and figures

  • Citations and references

Artwork source: Allison Horst

Quarto Produces:

Articles & Reports ~ Presentations ~ Websites ~ Books & more

Artwork source: Allison Horst

The Learning Curve

Artwork source: Allison Horst

Common Fears, Overcoming Barriers


🫠 “I’m not good at math/technical things”

  • Coding is more about logical thinking than math

  • You already do logical thinking in research design

  • Many successful coders struggled with math

Common Fears, Overcoming Barriers


🏃🏽‍♂️ “I don’t have time to learn”

  • Initial investment pays dividends

  • Even 30 minutes daily for a month makes a difference

  • The time you save later is worth it

Common Fears, Overcoming Barriers


😬 “What if I get stuck?”

  • Massive online community ready to help

  • Error messages are normal—even experts Google them

  • AI assistants (like ChatGPT) can help with code

Common Fears, Overcoming Barriers


🥱 “My field doesn’t really need this”

  • Every field is generating more data

  • Funders increasingly expect computational methods1

  • Sets you apart in job market