The 4-Hour Body: Self-Experimentation and Spotting Bad Science

Your dermatologist prescribes you a drug. You take it for months. Turns out it does nothing. Happens more often than you’d think.

That’s exactly what happened to Dr. Seth Roberts, a psychology professor at UC Berkeley. He was taking tetracycline for acne. Just for practice, he decided to actually measure whether the pills worked. He counted pimples every morning while varying the dosage. Six pills a day, four pills, zero pills. The result? Zero pills produced the same number of pimples as six pills. The drug was useless. Years later, research papers about antibiotic-resistant acne started appearing. But Roberts already knew.

These last three chapters of The 4-Hour Body are about two things: why you should test things yourself, and how to tell when “science” is lying to you.

Chapter 42: The Value of Self-Experimentation

This chapter was written by Seth Roberts himself. His story is a good argument for why non-experts can sometimes learn things that experts miss.

After the acne discovery, Roberts tackled his sleep problem. For years he’d wake up at 4 AM, unable to fall back asleep. He tried exercise. Didn’t help. Tried eating cheese in the evening. Nothing. Ran out of ideas after several years.

Then something accidental happened. He switched his breakfast from oatmeal to fruit. His early waking got worse - from half the time to every morning. Because he was tracking everything on paper, he noticed the correlation. He tested it back and forth. Oatmeal, fruit, oatmeal, fruit. The pattern held.

Eventually he figured out: any breakfast made early waking more likely. The best breakfast was no breakfast.

This actually matched animal research. Rats fed at the same time every day become active about three hours before feeding time. Roberts was eating at 7 AM and waking at 4 AM. Same pattern. Humans are animals. We just don’t like admitting it.

He kept going. Found that standing for 8+ hours a day helped sleep (not practical). Standing on one leg to exhaustion helped too (weird but doable). Animal fat improved sleep. 250 grams of pork belly had a clear effect.

Why Self-Experimenters Beat Professionals

Roberts identified three advantages:

Speed and cost. Self-experiments are fast and free. A conventional sleep study takes a year to fund and costs thousands. Roberts could test an idea in weeks. Fast iteration means more mistakes, and more mistakes mean more learning. He also found unexpected side effects - like flaxseed oil improving his balance - that no planned study would have caught.

Stone Age solutions are easy to test. Many health problems come from the gap between modern life and how humans evolved. No breakfast, more standing, more fat - these are returns to pre-industrial life. Drug companies won’t fund studies on “skip breakfast.” There’s no money in it. But you can test it yourself tomorrow.

Better motivation. Roberts studied his own sleep for 10 years before a breakthrough. No professional researcher would do that. They need publications, grants, tenure. You just want to sleep better. That’s a stronger motivation than any career pressure. Mendel was a monk. Darwin was rich. Wegener was a meteorologist doing geology as a hobby. Freedom from institutional pressure lets you think clearly.

Chapter 43: Spotting Bad Science 101

Headlines are confusing on purpose. “Panel Urges Hour of Exercise a Day” in 2002. “Why Exercise Won’t Make You Thin” in 2009. Same newspapers, opposite conclusions.

Ferriss gives you five questions to ask whenever you see health advice. He calls them the Big Five.

1. Are They Using Percentages to Impress You?

“Coffee drinkers lose 20% more fat!” Sounds amazing. But 20% of what? If the control group lost 0.25 pounds and the coffee group lost 0.30 pounds over eight weeks, that’s your 20%. In real terms, it’s meaningless. Always ask for absolute numbers. Percentages alone are a red flag.

2. Is This an Observational Study Claiming Cause and Effect?

This is the big one. Observational studies watch groups of people and note patterns. They can show correlation - A and B exist together. They cannot show causation - A causes B.

The Pastafarian example makes this clear: as the number of pirates decreased, global warming increased. Therefore, global warming is caused by a lack of pirates. Somalia has the most pirates AND the lowest carbon emissions. Coincidence?

A real-world example: observational studies showed women on hormone replacement therapy (HRT) had less heart disease. Media ran with it. Then proper randomized controlled trials showed HRT offered no protection and might even increase risk. The observational studies hadn’t accounted for the fact that doctors were prescribing HRT to healthier women in the first place.

3. Does the Study Rely on Self-Reporting?

In an Antarctic research station, men weighed and recorded all their food. When asked to recall what they ate the day before, they underestimated by 20-30%. These were people who had literally weighed the food that same day.

The Women’s Health Initiative - a $415 million study with 49,000 women - relied on questionnaires asking things like “In the last three months, how many half-cup servings of broccoli did you eat?” or “When you ate chicken, did you eat the skin?” Try answering that honestly. You can’t. Nobody can.

4. Is This Diet Study Claiming a Control Group?

In nutrition, true control groups are nearly impossible. If you reduce fat, you must increase carbs or protein to keep calories equal. Now you’re testing three variables at once. Any claim that a single nutrient was responsible is suspicious.

5. Who Funded the Study?

Fred Stare at Harvard got over a million dollars from General Foods in 1960 - the makers of sugar-loaded cereals, Kool-Aid, and Tang. He then spent the next decade defending sugar and food additives. Always check the “conflicts of interest” section. Look for “consulting fees.” Follow the money.

Chapter 44: Spotting Bad Science 102

This chapter was written by Dr. Ben Goldacre, a British doctor who writes the “Bad Science” column in The Guardian. He frames it as a tutorial: imagine you work for a drug company and need your mediocre pill to look good. What tricks can you use?

Study it in winners. Test your drug on young, healthy people who are likely to improve anyway. Don’t test on the complicated cases that doctors actually treat.

Compare against placebo, not against existing treatments. Your drug just needs to beat a sugar pill. That’s a low bar and everyone knows it, but the graphs look great.

Sabotage the competitor. If forced to compare against another drug, give the competitor at too high a dose (more side effects) or too low a dose (less effective). This actually happens.

Don’t ask about side effects. SSRI antidepressants cause sexual dysfunction in 2% to 73% of patients - depending entirely on how you ask the question. One review of 3,000 patients didn’t even list sexual side effects in its 23-item table.

Use surrogate outcomes. If your cholesterol drug is supposed to prevent heart deaths, just measure cholesterol levels instead. Easier to show improvement, cheaper trial, more positive result.

Hide bad data. Don’t publish negative results. Or bury them in text nobody reads. Drug companies did exactly this with SSRI data - hiding evidence the drugs performed no better than placebo.

Torture the statistics. Run every possible statistical test until something comes up positive. Slice into subgroups. “Your drug works very well in Chinese women aged 52 to 61.” As they say: torture the data long enough and it will confess to anything.

The Takeaway

Ferriss isn’t saying his book is bulletproof science. He admits his self-experiments lack randomization and control groups. The point is different: don’t wait for perfect research to improve your own life. But don’t blindly trust headlines either.

Track your own data. Test one thing at a time. And when someone shows you a study, ask those five questions before you change your behavior.

You don’t need a lab. You need a notebook and some skepticism.

Previous: Can You Actually Live Forever?

Next: Final Thoughts on The 4-Hour Body

This is part of my 4-Hour Body retelling series. New posts every Saturday.