Reading Sample Size in a Study You Read About
A summary of a study appears in a news feed, and the first question worth asking is not what the finding was but who produced it and how many people were involved. Sample size is the part of methodology most often left unstated, and it changes how much weight a number can carry. This annotated example walks through a short study abstract line by line, showing which phrases deserve scrutiny and why.
The artifact, as it typically appears
Consider a familiar summary: "In a randomised trial of 42 adults, six weeks of supervised resistance training three times weekly was associated with a modest rise in grip strength. Authors noted that daytime energy after a midday meal was reported as more stable by participants, though no difference was observed in an estimated VO2 figure or resting pulse."
On its own the paragraph reads cleanly. The number 42, the six weeks, the three sessions per week, the grip strength result — all of these are concrete. What the paragraph does not tell a casual reader is whether 42 was chosen deliberately or simply what the recruiting window allowed, and whether the trial was powered to detect changes in anything other than a single handgrip measurement. Those two omissions do most of the work in deciding how seriously the finding should be taken.
Why the number 42 appears and why it matters
Sample size is usually the result of a calculation made before recruitment, not a figure arrived at after the fact. Investigators estimate how large a difference they expect, how much noise the measurements carry, and how much uncertainty they are willing to accept — then work backwards to a target number. If 42 is a power calculation result, it is defensible for the main outcome and nothing else. If 42 is simply the number who happened to complete, the study is descriptive rather than conclusive.
- The stated outcome should match the calculation, not just the reported finding.
- Dropouts and incomplete data often reduce a planned cohort well below its target.
- Small trials can still be useful as pilot work but not as final evidence.
- Secondary outcomes such as energy after meals are usually underpowered by design.
- A named registry entry or protocol document should confirm the original target.
What the surrounding language reveals
The phrase "supervised" matters because it describes both the setting and the level of adherence, which is difficult to reproduce outside that setting. "Modest rise" is careful wording — it signals a small effect size rather than a decisive one. The absence of any change in estimated VO2 or resting pulse is worth noting: six weeks is short for measurable cardiovascular adaptation, so a null result there is unsurprising rather than meaningful. The energy-after-meals comment sits at the level of self-report, not measurement.
A number in a study is not the size of the effect — it is the resolution of the instrument the study used to look at it.
Readers who feel they do not know which numbers matter are usually reacting to this gap. The relevant numbers here are the planned sample, the achieved sample, and the outcome the trial was designed to test. Everything else is a footnote. A lab results glossary or a question list for a doctor prepared in advance helps shift attention toward those three figures rather than the headline sentence.
When to accept the wording and when to rewrite it
The published summary should be quoted as written when it is being cited as a source. If it is being paraphrased for a patient-facing note or a personal health file, the safer forms are: "a small trial of 42 adults over six weeks found a modest change in one strength measure" and "no changes were detected in fitness or resting pulse." Both keep the limitation visible. Anyone keeping a habit journal or logging a wearable's VO2 estimate alongside such reading should note that self-recorded numbers and trial endpoints are not interchangeable, and that a six-week window cannot say much about trends that develop over years.
