reporting with numbers

Health and Medicine

Constructing headlines

1) When covering medical research or studies, identify the source of any numbers or statistics included in your headline


"Sleeping 5 Hours or Fewer Every Night Could Put You at Risk of Multiple Chronic Diseases: Study"


"Sleeping 5 Hours or Less Raises Risk of Multiple Chronic Diseases"


To help audiences think critically about health data, it’s important to let them know where this comes from. Even including a simple word like “study” (or a phrase like “study finds”) helps here, as it points toward the source of the information and provides insights into how it was produced.

2) “Data” isn’t a person -- be explicit about who is interpreting it


"Biden Said the COVID-19 Pandemic is Over, Here’s What Experts Think"


"Joe Biden says the COVID-19 pandemic is over. This is what the data tells us”


It is fairly common in everyday conversation to hear things like “the medical research says [x]” or “the data tells us that [y].” But running articles with headlines like these is unhelpful, because they treat data as an unquestionable truth. Moreover, in saying “this is what the data tells us,” we’re assuming that “the data” can mean one and only one thing. Such headlines encourage passive acceptance of authority, instead of helping audiences think critically about data and the people who produce it and analyze it. That is why it’s important to be specific about who is making a claim, or where this comes from. Even a simple phrase like “Here’s What Experts Think” achieves this goal. Other options here include:

3) Differentiate between causation and correlation


"Being Around Birds ‘Linked to Mental Wellbeing Boost'"


"The Presence of Birds Can Improve Mental Health"


Headlines are one of the most common places to conflate causation and correlation. The study cited in the above examples found that “everyday encounters with birdlife were associated with time-lasting improvements in mental wellbeing.” Elsewhere, the study uses incredibly cautious language, pointing to (for example) a “possible causal link effect of birdlife on mental wellbeing.” Even when acknowledging the lack of a clear cause-and-effect relationship elsewhere in the reporting, the headline in the second example conflates causation with correlation. When in doubt, avoid causal language. Try one or more of these word swaps:

  • Instead of “causes” - linked to, tied to, associated with
  • Instead of “study shows” or “proves” - study suggests, study claims
  • Instead of “can” - could, may

4) Be cautious about inference from research on animal models.


An article with the headline “The Effects of Niacin on Alzheimer’s in Mice” informs readers about a medical study done on mice, and notes that “whether this research can translate to humans is unclear.”


"Nicotinamide Riboside May Increase Breast Cancer Risk, Metastasis." Neither the headline or the article ever mentions this was a study done on mice. In fact, it refers to a “person’s risks” and includes bites from the study’s author on supplements that people take and how individualized cancers may be.


Medical research relies heavily on model organisms (mice, fruit flies, fish, etc.) with simpler bodily systems than humans before conducting research on people. While this research may have implications for people, it can be tempting to make a leap to those implications without sufficient data.

Press releases often gloss over those technical distinctions, so it's important not to rely solely on press releases. If you don’t have time to puzzle through a whole paper, read the abstract (which you can see even if the paper is paywalled) and make sure it lines up with what you’ve heard from other sources. The abstract of the paper cited in the University of Missouri press release mentions “the role of NR uptake in cancer prevalence and metastases formation in triple negative breast cancer (TNBC) animal model” - and a quick search for “TNBC animal model” brings up mice bred to study this type of cancer. Reading only the press release, it would be easy to conclude this was a human study.

5) Be clear when scientific consensus exists


"Masks Cut Covid Spread in Schools, Study Finds"


“Doctors Debate Study That Found Masks in Boston-Area Schools Cut COVID Rates”


These headlines reflect the same study. The second headline is misleading -- in fact, the doctors unanimously found the study strong, and merely disagreed about the practical implications. These inconsistencies can make it difficult for people to gauge (and trust) the information they receive and the risk calculations that may lie within.

Provide feedback.