“Screen time causes behaviour problems.” “Family dinners create better grades.” A strong verb can turn a complicated study into a simple command. Before changing family life, check what the researchers actually measured. Two things moving together is an association; it does not by itself show that one caused the other.
Begin with the original claim
Open the research paper or an official summary, not only the headline. Look for words such as associated with, linked to, correlated with or predicted. These can describe a real and useful pattern, but they are not the same as “caused.”
The UK Government Analysis Function advises checking original sources, target populations, assumptions and limitations. MedlinePlus similarly recommends asking who was studied, how many people took part, how long the research lasted, what type of study it was and who paid for it.
Ask which direction the story could run
Imagine a study finds that children with more homework support have higher grades. Support might help grades. But children struggling at school may also receive more support, which could produce a different pattern. Or a third factor—available caregiver time, school resources or household stability—may influence both.
Write three possibilities:
- A may influence B.
- B may influence A.
- Another factor may influence both.
This exercise does not prove the headline wrong. It prevents one explanation from masquerading as the only explanation.
Look at the design
An observational study measures what already happens; it can reveal patterns but often cannot rule out every alternative explanation. A randomized experiment assigns an intervention by chance and can give stronger evidence about cause when it is ethical, well designed and followed as planned. Many parenting questions cannot be randomized, so researchers use careful comparisons, repeated measurements and statistical adjustments. Those methods help, but they do not erase every limitation.
Also ask whether the outcome was measured objectively, reported by parents, reported by children or inferred from a questionnaire. Each method can be useful and imperfect.
Check who the result applies to
A study of 80 university students is not automatically evidence about toddlers. A finding from one city, one school system or families volunteering for an intensive programme may not transfer to every household. Note ages, location, sample selection and who was excluded.
Use the “I don’t know yet” script when a child asks about a dramatic claim. Honest uncertainty is more useful than a confident guess.
Translate the headline into a fair sentence
Replace “X causes Y” with: “In this group, researchers found that X and Y occurred together. The study can or cannot test cause because…” Add the main limitation. If the original study truly tested a causal intervention, describe exactly what was assigned and what outcome changed.
Decide whether any action is needed
Low-risk, already sensible practices do not require a perfect study. But major restrictions, costly products, health decisions or blame should not rest on one headline. Compare multiple reliable sources and discuss health decisions with a qualified professional. The child health appointment question cards can help you take a claim to an appointment.
A 60-second headline check
- What exactly was measured?
- Was the study observational or experimental?
- Could the direction run the other way?
- What third factors might matter?
- Who was studied, and does that resemble my child?
- Does the paper say “associated” while the headline says “causes”?
Notice when a stronger design is still imperfect
Even an experiment can lose participants, measure the wrong outcome or be too short to show lasting effects. Check whether families followed the assigned activity and whether researchers reported unintended harms as well as benefits. Good causal evidence comes from the whole design and analysis, not from one label. When the paper is hard to interpret, a cautious summary is more accurate than filling the gaps yourself.
