Quoting Stats and R - What statistical test should I do?

What Statistical Test Should I Do?

Reference: What statistical test should I do? from Stats and R

A Question You Must Ask Yourself in Your Research Career

I am definitely that student in Antoine Soetewey’s blog post, “What statistical test should I do?” Even after working many years in the industry, I still had not asked myself this question—until I worked on an LLM-as-a-Judge project last year.

That’s when I realized I could no longer bear such an annoying problem. It was time to ask myself this question and answer it every single time. I know it seems ridiculous for a senior researcher to not know how to choose the right statistical test, but for me, I take it as paying a knowledge debt to my past self.

“The best time to plant a tree was 20 years ago. The second best time is now.”

Why I Strongly Recommend This Blog

When I was searching Google for the answer to this question, I discovered that Antoine Soetewey had already created a structured flowchart (as you can see in the above-mentioned post). I’m grateful for his work in summarizing such dense content that I had hoped to create myself someday.

The blog offers:

• Clear decision-making framework: A prepared instruction set you can follow
• Comprehensive coverage: Almost every statistical test with detailed explanations
• LLM-friendly format: In the era of large language models, the flowchart serves as an excellent reference document that can be used with ChatGPT for your specific tasks