How to Understand What a Study Actually Proved
In the morning, a news alert pops up in your feed: «Scientists prove coffee protects the brain.» A few hours later, another channel reports that coffee is harmful to the heart. It seems as if science is contradicting itself. However, the issue is often not a contradiction, but rather that the short headline omitted crucial details: who was studied, what was measured, and how long the research lasted.
«Science Humanly» is a Telegram channel and mailing list for those who want to read scientific news critically but are not ready to spend an entire evening on every article. In our weekly edition, we break down three or four studies in about three minutes: we explain the results in plain language, assess the reliability of the findings, and provide a link to the original source.
Why Titles Often Promise More Than the Research Delivers
Research typically examines a narrow question. For example: will energy consumption change in overweight adults if they are helped to sleep longer? The title might transform this result into a general promise: «Sleep aids weight loss.»
There is a distinction between these two formulations. The study might have only involved people of a specific age, with a particular sleep characteristic, and for a short duration. A result from such a group does not yet mean it will work equally well for everyone.
To evaluate the news, it is helpful to ask five questions:
1. Who was studied? Were the participants healthy volunteers, patients with a specific diagnosis, students, or people of a certain age?
2. What exactly did the researchers do? Did they compare groups, observe habits, or analyze existing data?
3. What was measured? Well-being, blood tests, weight change, brain imaging, or another indicator?
4. How long did the study last, and how many participants were there?
5. Does the sensational headline match the authors’ cautious conclusion?
Connection is not yet a cause
Imagine that in a large group of people, those who walk more tend to feel better, on average. This is an interesting observation, but it doesn’t inherently prove that walking is the cause of the improvement. Perhaps the people who walk more already have more free time, better health, or live in a more convenient urban environment.
Observational studies help reveal correlations and formulate hypotheses. To state causation with greater certainty, however, more data and an appropriate research design are necessary. For instance, in a randomized controlled trial, participants are randomly assigned to groups, and the intervention results are compared against a control group. While this design doesn’t resolve every question, it generally does a better job of separating the effect of the intervention from other factors.
Therefore, in this channel, we pay attention to verbs. «Is related to» cannot always be substituted with «causes.» «Researchers found» does not mean the result is already applicable to everyone. And «improved among participants» does not always mean one group improved more than another.
One number—only with context
A number can sound convincing yet still be misleading. «The risk increased by 50 percent”—but how high was it originally? „Participants lost three kilograms”—over what period and compared to whom? „The effect was discovered”—how accurately was it measured?
In one randomized study, 80 adults who typically slept less than 6.5 hours attempted to increase their sleep time. The group that received recommendations for extending sleep slept approximately 1.2 hours longer per night, and their energy consumption decreased by an average of about 270 kilocalories per day compared to the control group. This is a beneficial result, but it applies only to specific participants and conditions. It does not prove that simply sleeping longer is sufficient for everyone to lose weight. The original source describes the methodology, the participant groups, and the study’s limitations.
What does «proven / not yet» mean?
In this channel, we will use the «Proven / Not Yet» scale, but it doesn’t label an entire topic. It refers to a specific claim.
For example, a correct conclusion might read: «In this trial, the limited-time eating window did not result in greater weight loss than the standard regimen, despite having the same caloric intake.» This does not mean that all eating patterns are equivalent or that timing never matters. It simply means that this specific study did not demonstrate an added benefit under the conditions tested.
For each material, we will specify:
What was investigated: the question and participants. What resulted: the main findings without exaggeration. How reliable the conclusion is: what the design allows us to say and the degree of confidence in the data. What the research does not prove: the boundaries of the result’s applicability. Where to check: a link to the article or primary source.How Editorial Work is Organized
The editor selects the research topics. The model helps prepare a draft in simple language, but it is not responsible for the final text. Before publication, the editor cross-references the phrasing with the original: they verify the sample size and composition, figures, methods, authors' conclusions, and limitations.
Next, the issue is formatted according to the template. The illustration must explain the concept, but it should not substitute for the data with an impressive picture. Publications are prepared in advance, but a person presses the final button.
This process allows automation to be used for speed while preserving human review where errors are most likely—in the numbers and the interpretation of the results.
Where to read «Science Made Human»
The main channel—Telegram:—features short posts, headline analyses, polls, and reader reactions. The second format is a newsletter. This allows you to easily save, reread the weekly issue, and access source links.
Publication Rhythm:
One weekly issue: three to four studies, a brief analysis, and a link to the original source.
Two short posts: one figure of the week, a headline analysis, or a question for the audience.
Newsletter: the same issue compiled with a convenient set of links.
«Science Humanly» does not promise that the reader will find a definitive answer after every piece of material. The task is more modest and practical: to help understand exactly what the data reveals, where a confident conclusion ends, and what warrants further personal investigation.
2. Six-Month Thematic Plan
One main issue is released weekly. Two short posts support the main theme of the issue: for example, one breaks down the headline, and the other explains an important statistic or asks the reader a question. Research for future issues is selected closer to publication to keep the topics timely.
