Keto Forums and Confirmation Bias

Quick answer: A keto group is full of people keto worked for, because the people it didn't work for quietly left. That's not a conspiracy, it's a filter, and it means the advice skews toward "more keto" for every problem. Use the community for recipes, encouragement and company. Take symptoms, lab results and medication questions to your doctor.

On this page

What confirmation bias looks like in a keto group
The people you don't hear from
Before-and-after posts aren't a fair comparison
What forums are good for
When to ask a doctor instead
What the research says
KetoDick's take
Key takeaways
Related

What confirmation bias looks like in a keto group

Confirmation bias is the habit of noticing what agrees with what you already believe and skipping what doesn't. Everyone does it. A group built around one diet does it as a crowd.

Post "my blood sugar is weird" in a keto group and you'll get keto answers: cut more carbs, add more fat, test your ketones. Post "I've had palpitations for a week" and someone will say electrolytes. They might be right. But a group where every answer is keto can't tell you when the answer isn't.

The people you don't hear from

Think about who stays in a keto group for a year. Mostly the people it's working for. Someone who felt awful, stalled for months or got a bad cholesterol result tends to drift off without posting a goodbye.

That's survivorship bias, and it makes the remaining group look more successful than the diet is on average. Trials have the opposite problem built in: they count the dropouts. In one year-long trial of low-carb versus low-fat, retention was low in both arms (Foster et al., 2003). The people who left are part of the result. In a forum, they vanish.

So what you do is read success stories as "this is possible", not "this is typical".

Before-and-after posts aren't a fair comparison

Feeds show people's best day. Starting weights get rounded up, timelines get rounded down, and nobody posts the morning they gained two pounds.

Starting points also differ. Someone coming off a diet heavy in fast food and salt will drop a lot of water in the first weeks, and their early photos will look dramatic. If you started from a cleaner baseline, your first month will look slower, and that says nothing about whether it's working.

And some posted rates of loss aren't healthy. A comment section rarely questions a fast loss, because it isn't polite. If a post makes you want to eat even less, close the app.

What forums are good for

Plenty, honestly:

Recipes and swaps. Real people cooking real food on a budget is the best thing keto groups do.
Company. Changing how you eat is lonelier than people admit. A group that gets it helps.
Practical tips. What's in the store, what's hiding carbs, what to order at a restaurant.
Accountability. Posting your week to people who'll ask how it went.

None of that needs a group to be right about medicine.

When to ask a doctor instead

Some questions don't belong in a comment thread:

You take diabetes or blood pressure medication. Cutting carbs can change what you need, sometimes fast.
You have kidney disease, or you're thinking about potassium or magnesium supplements.
You're pregnant or breastfeeding.
You have a history of an eating disorder. Bring in your therapist as well as your doctor.
Symptoms that hang on: cramps, sleep trouble, night sweats, palpitations, anything that doesn't settle after the first few weeks.
Lab results, especially cholesterol after a few months.

What the research says

There isn't much research on keto forums as such, but a few findings are useful:

Online posts tell a real but partial story. A content analysis of online posts found reported "keto flu" symptoms clustered in the first weeks and were mostly described as passing (Bostock et al., 2020). That's useful. It's also a sample of people who chose to post.
Averages hide big individual differences. In DIETFITS, 609 adults on healthy low-fat or healthy low-carb lost similar amounts over a year, and genotype and insulin secretion didn't predict who'd do better on which (Gardner et al., 2018). Forums rarely hear from the people for whom the other diet would have worked better.
Lipids vary. The National Lipid Association says low- and very-low-carb diets lower triglycerides and raise HDL, but have a variable effect on LDL (Kirkpatrick et al., 2019). "My numbers got better" in a forum doesn't predict your numbers. A blood test does.

KetoDick's take

This one is close to his heart, and he's aimed it at his own side. In his 2017 video answering "keto is dangerous", he said the real danger he worried about was confirmation bias: a keto forum gives you only one answer, and keto may not suit everyone's gut or genetics (video 1m1hPgaRIg4). "Beware of confirmation bias," from the video. He also suggested that some of the weight people credit to keto may come from simply paying attention to food, tracking and meal prep.

His advice in the same video was to get blood work, talk to your doctor and follow up on symptoms, and not just ask internet forums. That's notable coming from someone who was answering keto-forum questions live on camera at the time (video cTGYIFOTPl8).

On before-and-after posts, he's said social media can be worse than the scale: feeds are cherry-picked, some numbers are exaggerated, and some posted rates may hide crash dieting or disordered eating (video dNUvzMl-OEU). And he's said he doesn't believe keto works for everybody (video cTGYIFOTPl8).

The literature and his position line up here. The one thing to add is that he's also been critical of strictness inside keto groups, the people policing every gram, because it makes others feel they're failing (videos DWI8qQl5BbY, ffwBPoKxlDI). Confirmation bias and purity tests tend to travel together.

Key takeaways

Keto groups over-represent people keto worked for; the rest quietly leave.
Read success stories as "possible", not "typical".
Starting points differ, so early results aren't comparable.
Use groups for recipes, company and accountability.
Take medication changes, kidney questions, pregnancy, eating-disorder history and lingering symptoms to a doctor.
A blood test beats a comment thread for cholesterol questions.