4 reasons to stop believing the science 👩‍🔬

“People who eat bacon are twice as likely to get colon cancer”

We’ve all seen headlines like this, here’s a quick explainer on why they are mostly BS, and why food science is a pseudoscience.

Enjoy 15% off Force of Nature Meats’ incredible pastured Bison, beef and ancestral organ blends when you shop with Seed Oil Scout 🥩 🫡

1. Healthy User Bias:

This is a common issue in observational studies. For example, if someone is vegan, they likely think about their health more than your average person, and are therefore also more likely to exercise.

Statistically it looks like the veganism is their secret, but in reality they are a cohort of health conscious users being compared to couch potatoes.

There’s no real way to separate this in observational studies, and it is a big reason for bacon effects like above (bacon eaters might not have gym memberships on average etc)

2. Hawthorne Effect:

This effect refers to the change in behavior of study participants simply because they are being observed. For example, if people know they're being studied for their eating habits, they might eat healthier than usual, skewing the results.

This is particularly bad when control and treatment are handled differently

The famous PREDIMED study often cited showing tremendous health benefits of EVOO, may have actually just been testing the health benefits of dropping off the EVOO every so often to remind people to eat healthy, since the control group was never visited during the study.

See “Olive oil debunked” by Fire in a Bottle

3. Recall Bias:

Many nutrition studies are survey based, relying on questions like “How many times did you eat bacon in the last 6 months”. These surveys are wildly inaccurate to the point that they are basically useless. Do your best to ignore any survey based studies.

4. Absolute vs Relative Reporting:

This is a big one. Let's say the risk of colon cancer is .1% in non-bacon eaters and .2% in bacon eaters. While it's true to say bacon eaters have a "100% increased risk," it sounds much scarier than saying their risk increased by just .1 percentage point. Always check if the reported risk is relative or absolute.

Types of Studies:

Different studies have different levels of evidence and potential biases:

• Observational: These studies observe people in their natural environments without intervening. While they can find associations, they can't prove causation. They're also more prone to biases like the "healthy user bias."

• Randomized Control: Participants are randomly assigned to one of two (or more) groups: a treatment group and a control group. These are the “gold standard”, but are nearly impossible to do with diet unless you use soldiers or prisoners. These are still prone to the Hawthorne effect, and bad randomization (randomizing whole villages instead of individuals etc)

• Meta-analysis: This is a study of studies. Researchers compile data from multiple studies to get a more comprehensive view. While they can increase statistical power, they can also amplify biases present in the original studies. Garbage in garbage out.

Always take dietary headlines with a grain of salt. One single food item rarely means much.

When seed oils have been put to the test with RCT’s on prisoners, they tend to cause heart disease:

Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968-73)

🫡