Writing a Research Paper › Module 3: Results and discussion
Not overclaiming
Lesson 3 of 3 · 2 min
Spin means presenting results as more positive or more important than the evidence supports. It's common, often unintended, and readers notice.
How common?
In 2010, Boutron and colleagues looked at reports of randomised trials whose main outcome was not statistically significant. In the conclusions of the abstracts, more than half still contained spin: they focused on secondary results, or described the treatment as beneficial anyway.
Spin grows as it travels. In 2014, Sumner and colleagues studied university press releases about health research. About 40% gave more direct advice than the paper did, about a third made causal claims from correlational studies, and about a third applied animal results to humans. When press releases exaggerated, news stories usually did too.
Common forms
- Causal words for correlational data: "X increases Y" when you only observed they move together.
- Generalising too far: from one dataset or population to all.
- Focusing on secondary wins when the main test failed.
- Hype words: "breakthrough", "novel", "dramatically", "for the first time" without proof.
Match words to evidence
| Evidence | Honest wording |
|---|---|
| Correlation in observational data | "is associated with" |
| One experiment on one dataset | "in our setting, X improved Y" |
| Many replications, many settings | "X reliably improves Y" |
- Correlation in observational data
- Honest wording: "is associated with"
- One experiment on one dataset
- Honest wording: "in our setting, X improved Y"
- Many replications, many settings
- Honest wording: "X reliably improves Y"
RememberMake every claim exactly as big as its evidence. If you wouldn't defend a word to a sceptical reviewer, cut it.