The house rules. Mostly these exist to stop research reporting doing the two things it usually does: overstating the finding, and hiding the method.
Say what the study actually found
The finding goes near the top, in plain words, with the numbers attached. Not “researchers have uncovered a surprising link” — the link, what it was between, and how big it was.
Match the verb to the design
This is the rule that matters most, and the one most psychology coverage breaks.
- Cross-sectional survey — “was associated with”, “went together with”. Never “causes”, “leads to”, “makes you”, “triggers”.
- Longitudinal cohort — “predicted”, “preceded”, “was followed by”.
- Randomised controlled trial — “caused”, “led to”, “improved”, within the trial population.
- Meta-analysis or systematic review — “across studies”, “evidence indicates”.
- Lab experiment — “under lab conditions”, “in this experiment”.
- Animal study — no extrapolation to human advice at all.
Translate the effect size
“Statistically significant” is not a size and we do not use it as one. Where the paper reports a number, we give it and then say what it means:
- A correlation of 0.3 explains roughly nine per cent of the variation. We say that rather than “strongly linked”.
- An odds ratio is converted to absolute risk. “Doubles your risk” means nothing without the base rate — two per cent to four per cent is a doubling and also a two-percentage-point change, and we report both.
- A percentage change and a percentage-point change are different things and we distinguish them every time.
Name the sample
Not “participants” but “1,043 adults in Norway”. Not “young people” but “212 university students”. If the sample was students, one country, or a self-selected online panel, that goes in the body of the article, not a footnote. Most psychology samples generalise poorly and readers are entitled to know.
Give statistics a year and a population
Prevalence figures age badly and get repeated long past their shelf life. Every statistic on this site carries its year and who it describes.
Label preprints
If a paper has not been peer reviewed, the article says so in the text — “in a preprint that has not yet been peer reviewed” — not in small print at the end.
Quotes are real or absent
We only publish a quotation that a named person actually said, with a link to where they said it. Not a composite, not an unnamed “expert”, not a paraphrase in quotation marks. If no genuine quote is available, the article runs without one and uses attributed paraphrase instead: “the authors write that…”.
We do not open articles with invented anecdotes. No fictional patient, no composite reader, no friend-of-the-writer who does not exist.
State the limits
Every article carries a section on what the study cannot tell you: the design, the sample, what was not measured, what remains unknown. It is the section most competing coverage leaves out and the reason we think ours is worth reading.
Plain language, no filler
Short sentences next to long ones. Jargon explained the first time or not used. No wellness-industry vocabulary, no “rewire your brain”, no “science-backed”. Where we are uncertain, we sound uncertain.
Practical advice, hedged honestly
Where a finding genuinely supports something a reader could do, we say so, with the hedge the evidence deserves. Where it does not, we say that instead of manufacturing tips. A study that explains why something happens does not automatically tell you how to fix it.