#Statistics in the editorial process (short thread) 🧵

🔍What does @sjn_16 look for when peer reviewing for clinical journals? 🔍

‼️ May not reflect the views and peer review approach of other statisticians ‼️
#Statistics in the editorial process: WHEN?

Recognising statistical problems can be difficult for a non-statistician (and even for statisticians!)🤔🤯
Tips to spot problems are available:
https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-015-0510-5
But if in doubt, consult a statistical reviewer! 🤓
#Statistics in the editorial process: WHEN?

✅Reporting guidelines such as #CONSORT, #PRISMA and #STROBE may help indicate where important information is missing:
https://www.equator-network.org/ 

❌But these are not conduct guidelines and do NOT guarantee good statistical analysis
#Statistics in the editorial process: WHAT?

Myth: Statisticians only look at methods - WRONG❌

I have to read the whole paper to follow the 'story' to determine if methods, results and conclusions are appropriate and match the objective 🧐
#Statistics in the editorial process: WHAT?

BEWARE: If you have a published protocol / analysis plan or study registration, I will read this and compare what you planned what you actually did 😈
(It's OK - just add a 'changes from protocol' section 😇)
#Statistics in the editorial process: METHODS

❌Don't skip the analysis section to save word count
❌Don't cite a protocol or previous work for ALL methods
✅Methods section MUST have sufficient detail for a standalone paper
✅Use an online statistical appendix if needed 👍
#Statistics in the editorial process: METHODS

Using a complex statistical method?
✅Explain what the method does related to your clinical objective in plain language.
‼️ Can't explain it? = Should you use it?
❌AVOID jargon & complex statistical terms without explanation
#Statistics in the editorial process: RESULTS
❌ Lengthy results text is boring 🥱
✅ Tables & Figures help to visualise results 📈
‼️ Use of colour, legends & footnotes can help with interpretation of results
‼️ Check that numbers in Tables & Figures match results text 🧐
#Statistics in the editorial process: RESULTS

More specific examples of common statistical errors and misinterpretations coming this week 🎉
#Statistics in the editorial process: HOW?

I write a statistical review which highlights:
‼️ MAJOR comments which may require re-analysis or rewriting of results with explanation of the issues & suggestions for improvements (if I can!)
❗️ MINOR comments - such as wording
#Statistics in the editorial process: HOW?

Disagree with my comments? 🙅
Have I completely misunderstood your work? 🤔
NO PROBLEM 👍explain to me in your response.

But if I don't understand your #Statistics, clinical readers have no chance 🤷
So do add some clarifications!
#Statistics in the editorial process: WHY?

Statistical peer review is SO IMPORTANT to provide highest quality evidence for clinical decision making

📢Calling statistical reviewers! Be constructive, be nice and most importantly, don't be #reviewer2 😱
You can follow @ColorectalDis.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: