// learn.shawon.ch / research-methods / why-research-misleads STUDY GUIDE
← Research methods

Research methods · Foundations

Why research is easy to misread

Try this first

A headline says a study found people who drink coffee live longer. Before reading on, name three ways that could be true without coffee doing anything at all.

Picture a real result moving through five hands. A lab surveys 40,000 people and notices that the ones who ate more nuts had slightly fewer heart attacks. The university press office needs the story to land, so the release says nuts protect your heart. A journalist needs the click, so the headline becomes eating nuts cuts heart-attack risk. An influencer needs engagement, so the caption is THIS is why you're still sick. By the time it reaches you, a cautious correlation from a survey has hardened into a command. Nobody lied outright. Each hand just nudged.

The one idea

A finding gets less true and more certain at every step from the lab to your feed. Your job is to walk the chain backward to the original claim — which is almost always smaller, more hedged, and less sure than the version that reached you.

The distortion is built in, not accidental

It would be comforting if this were fraud, because fraud is rare and catchable. It isn't. Each hand in the chain is responding to an honest incentive, and the incentives all point the same way: toward a cleaner, louder, more certain claim. Five small nudges compound into a sentence the original authors would never have signed.

Two facts make the chain especially slippery. First, most real effects in messy fields — nutrition, medicine, psychology, even model benchmarks — are small and noisy, so the honest version rarely sounds impressive enough to share. Second, the words that carry the hedge are the first to get cut: associated with and linked to quietly become causes, boosts, fixes.

The same finding, five hands later
HandWhat they needWhat the claim becomes
LabTo publish"Nut intake was associated with fewer events"
Press officeMedia pickup"Nuts protect your heart"
JournalistThe click"Eating nuts cuts heart-attack risk"
InfluencerEngagement"THIS is why you're still sick"
YouAn answerA command you act on
LAB · SURVEY nut-eaters had slightly fewer events PRESS OFFICE nuts protect your heart NEWS HEADLINE eating nuts cuts heart-attack risk INFLUENCER THIS is why you're still sick YOUR FEED just buy these already
Down the chain: nuance shrinks, certainty grows.

Work one, then finish one

Worked: "A study proves multivitamins help you live longer." Walk it back. Proves is the tell — was it a controlled trial or a survey? It was a survey: people who chose to take vitamins were tracked. But vitamin-takers also exercise more, smoke less, and see doctors more often. So the pill may be a marker of a careful person, not a cause of long life. Walked back, the real claim is: "vitamin-takers, who differ in many ways, lived a bit longer." Far less sellable — and far more honest.

Your turn: A post says "new study: standing desks boost productivity 40%." Name the two questions that deflate it before you believe a word. (Was it a controlled trial or just people who already chose standing desks; and is 40% a real-world amount or a relative number with no baseline — and what was actually measured, output or self-reported busyness?)

Why this earns a place in your toolkit

This is the engine behind most of what gets sold to you. The same chain runs from a single nutrition study to a podcast segment to a supplement label to a forum thread swearing the stuff changed someone's life. A health podcaster needs a confident take every week; a brand needs a reason to buy; a community rewards the dramatic anecdote. Whether the claim is about collagen, a longevity compound, or the latest peptide, the reflex is identical: before acting — or buying — walk it back to the smallest, most hedged version the actual evidence supports. Train that one move and you've defused most of what reaches you.

Recall check · no peeking

  1. Name the five hands a finding passes through, and what each one's incentive does to the claim.
  2. Which words usually get cut first as a claim travels, and why does cutting them matter?
  3. What is the single move that protects you, stated as an instruction to yourself?

Explain it back

In one plain sentence, tell a friend why the version of a study you see online is almost always more confident than the study itself.

Learn · Shawon Chowdhury · a study guide, kept rough on purpose