Free lesson

How Science Finds Things Out: From Hypothesis to Consensus

How Science Finds Things Out: From Hypothesis to Consensus — Science or Myth? How to Check Health and Science Claims Yourself. The path from hypothesis to consensus; What peer review does — and does not guarantee; Preprint, published paper, correction and retraction; A real case of self-correction: MMR and autism; Where does this study sit?. Practical…

Demo edition. The materials are undergoing editorial review.

Topics covered

  • The path from hypothesis to consensus
  • What peer review does — and does not guarantee
  • Preprint, published paper, correction and retraction
  • A real case of self-correction: MMR and autism
  • Where does this study sit?

The path from hypothesis to consensus

“A study proved” often compresses a long process into one headline. A simplified path is: researchers frame a question and a testable hypothesis; choose a method; collect and analyse data; report the result; undergo peer review; publish the work; and have other specialists criticise it, test it with new data and compare it with the rest of the evidence. Once enough studies exist, a systematic review may collect them using predefined rules. Consensus develops when the body of evidence and specialists in the field broadly converge on a conclusion.

This is not necessarily seven stations travelled in order. A paper may be corrected, its methods challenged, and new evidence may change the overall picture. A review can also be flawed. The practical question is: where does the claim sit, and what has already been checked independently? A new individual study may matter, but it is not consensus by itself.

What peer review does — and does not guarantee

In peer review, an editor and specialists in the relevant field assess a manuscript before publication. They may ask whether the method fits the question, whether the analysis is reported properly, whether the conclusions follow from the data, and whether limitations are stated. The outcome may be rejection, a request for substantial revision, or acceptance. The exact process varies between journals and disciplines.

Reviewers normally do not rerun the experiment or collect the data again. They may miss an error, selective reporting or concealed fraud. “Peer-reviewed” therefore means the manuscript passed a particular professional check, not that it received a guarantee of truth or reproducibility. The filter’s imperfections do not make every item of evidence equal: a clear method, post-publication criticism, independent repetition and a wider body of studies provide separate layers of checking.

Preprint, published paper, correction and retraction

A preprint is a scientific manuscript shared openly before formal journal peer review. It lets researchers circulate methods and results quickly and receive comments. Its version, analysis or conclusion may still change. Check the repository banner, version number and any link to a later journal article.

Publication does not make the record unchangeable. A correction addresses an error when the main result may remain valid. A retraction signals that the work should no longer be relied upon as part of the scientific record; the reason may be a serious honest error or misconduct. Retraction alone does not prove an intent to deceive. Find the journal’s notice and read its stated reason. Revising a conclusion after new evidence is an expected part of self-correction; the useful question is which evidence caused the change.

A real case of self-correction: MMR and autism

In 1998, The Lancet published a case series involving 12 children that was widely interpreted as grounds to suspect a link between the MMR vaccine—measles, mumps and rubella—and autism. That design could not establish causation by itself. Much larger studies followed, alongside a journalist’s investigation and proceedings by the UK medical regulator. In February 2010, the journal fully retracted the paper; later that year, the lead author was removed from the UK medical register.

Checking did not end with the retraction. A 2019 Danish cohort study included 657,461 children born from 1999 through 2010 and found no increased autism risk after MMR, including in the higher-risk subgroups it examined. This is one large observational study, not an absolute guarantee by itself; it also matters because its finding fits the wider evidence base. The source-checking lesson is that a prestigious journal and a dramatic first paper do not replace better designs, independent scrutiny and post-publication criticism. Personal medical decisions should be discussed with a qualified healthcare professional.

Where does this study sit?

Start with a short status map rather than an immediate true-or-false verdict: 1) Is it a preprint or a journal article? 2) Is there a correction, expression of concern or retraction? 3) Is it one study or a systematic review? 4) Have independent teams tested the result? 5) Does the public claim match what was actually measured? 6) How confident can the whole evidence base make us?

Fictional worked example. A headline says, “Common sweetener linked to memory loss.” Its link opens a preprint of one observational study. The authors report an association but cannot establish causation; there is no independent repetition or systematic review yet. A justified note is: “early, unreviewed and unconfirmed result.” It is a reason to watch for further checks, not to repeat the headline as established fact or independently change treatment or diet.

Practical assignment

Trace one scientific correction

Continue to learning

Course outline

  1. Why Myths Sound So Convincing
  2. The 10-Minute Check
  3. How Science Finds Things Out: From Hypothesis to Consensus
  4. The Evidence Ladder
  5. Correlation vs Cause: What “Linked To” Really Means
  6. Numbers That Scare: Relative and Absolute Risk
  7. From Study to Headline: Walking the Chain Back
  8. One Study Is Not a Fact
  9. Test It Yourself: A Fair Home Experiment
  10. Red Flags of Pseudoscience
  11. Experts, Consensus and “the Other Side”
  12. How to Talk About a Myth Without a Fight