Heart Podcast Top 10 statistical errors in submitted papers...and how to avoid them
7 snips
Oct 7, 2025 Dan Green, a Senior Teaching Fellow at Aston University and a statistical reviewer, shares invaluable insights into common statistical errors found in submitted manuscripts. He discusses the importance of avoiding incorrect causal language and emphasizes the need for well-structured abstracts. Green highlights tips on reporting statistical analyses accurately, utilizing clear participant flow diagrams, and avoiding pitfalls in regression modeling. He also advises against relying solely on univariable p-values for model building, promoting transparency in reporting missing data.
AI Snips
Chapters
Transcript
Episode notes
Lessons From 200+ Statistical Reviews
- Dan Green has done over 200 statistical reviews for journals including Heart and Addiction over five years.
- That experience formed the checklist behind his Top 10 statistical pitfalls paper.
Use Causal Language Only When Justified
- Avoid causal words like "proves", "shows" or "leads to" unless you have a well-conducted RCT or strong causal-inference design.
- Check your conclusions and remove causal language if your study only supports associations.
Format Abstracts To Journal Requirements
- Follow the journal's abstract structure and include the five W's: what, who, where, when and why.
- Compare your abstract to published articles in the same journal to ensure completeness.