Asking for collaboration on nutrition and healthy lifestyle, biases and reproducibility issues


Anyone interested in collaborating in a study (literature review, metaanalysis…) on nutrition and healthy lifestyle, biases, reproducibility and recommendations?

I could share my current hypothesis, findings and scope with anyone interested. I would be open (and happy) to adapt scope and assumptions while it still relates to the subject.

Moreover I will be glad not being first (or 2nd, or 3rd or…) author if it means this research gets traction. Actually my current motivation is curiosity to go deeper on a hard problem that relates to several of my personal interests (statistics and causal inference, observational studies, reproducibility, open science, health, life-style, nutrition…) and to publicly communicate results.

“Publicly communicate results” would imply any code required to run the published analysis or charts (in github?). And any data (at least the macro-data, but also micro-data when possible, according to legal issues). Also publishing lack of results, that may be considered also as another valuable type of results.

Let me know :slight_smile:



Hello Enrique,

This sounds very interesting, although the area of the study sounds very broad! Could you be a bit more specific on what you plan to do? Is the scope of your study already planned or are you looking more to discuss with somebody on a potential specific hypothesis? Are you planning to do an original study or a review?

Could you also specify a bit on what type of collaborator you are looking for? (e.g. a data analyst or someone with a specific background on the topic of your study?) What is your own background and what do you plan to bring to the project?

I think this will help the discussion!

1 Like

Hello Jelle,

In principle, my purpose was to complete a review, with specific hypothesis and scope: “Meat & colorectal cancer, fruits & vegetables: exploring confounding and selection bias”. Let’s say:

  • Vegetables and fruits are good to avoid colorectal cancer
  • Usually, low consumption of vegetables and fruits is associated with another unhealthy life-style habits
  • WHO has classified processed meat as carcinogenic Group 1, and red meat as Group 2A. These conclusions were based on IARC monograph. But it’s possible they were highly influenced by a strong association between high meat intake and low consumption of vegetables, that was found in the larger cohort study included in IARC metaanalysis.
  • The field of causal inference has developed methods to help to identify and control biases in observational studies: Matching, Inverse Probability Treatment Weighting, etc. These methods are more common today that 10 years ago, but they are still not universally understood and applied. There is room for improvement also on how to differenciate good controls from bad or unnecessary controls.

Anyway, I am open to adapt the scope or pivot almost completely the idea, regarding circumstances, expertise or interests of any possible collaborator. Furthermore, despite whatever results on this research, I would not like to sound like a plead in favor of meat, but a critical review and collection of best practices to diagnose and adjust for biases in observational data. So perhaps instead of meat consumption, the first review could be focused on another less controversial dietary habit that may be recommended or discouraged as a collateral effect of any other typical bias.

My background is in MSc, with postdegree in Signal Processing and a few old publications related to Biometrics. And I’m currently working in the private sector as “Data Scientist” (though I dislike this umbrella or fuzzy term). Anyway, I’m an alien to domain specifics on health or nutrition. But I have practical expertise within statistics, control groups, biases, confidence intervals, credible intervals… and I’ve suffered directly how hard is it to measure properly and even harder (and risky) to infer causality when you can’t perform (for any reason) Randomized Control Trials.

Regarding what I’m looking for, any collaborator with expertise or interests in the statistical approach is welcome. And any collaborator with domain knowledge on nutrition is specially welcome. Particularly if she or he is already studying at individual physiological level the benefit or harm of any particular nutritional aspect or dietary action.