Reading path: Underwater images and evidence →
Count observations and define the unit
Sample size is the number of units observed or measured. Sampling design is the plan for choosing those units. Both affect interpretation, but they answer different questions. A study can have many measurements collected in a narrow or biased way, or fewer measurements selected to represent a broader range of conditions.
Before reading a reported sample size, identify the unit. Does “n = 60” mean organisms, water samples, camera frames, transects, or independent sites? Sixty frames from one dive do not necessarily provide the same independent coverage as frames taken during sixty separate dives. The unit matters because observations gathered close together may share conditions.
Ask how the sample was selected
Sampling design describes where, when, and how observations entered the study. Researchers might choose sites randomly, sample fixed stations repeatedly, select accessible locations, or target places where an organism is known to occur. Each choice suits different questions and creates different limits. A fixed-station design can be useful for tracking change at those stations, while a random design may help estimate conditions across a defined area.
Imagine a survey of rocky shore organisms that records 200 quadrats, all placed along one convenient stretch of beach. The count sounds substantial, but the sites may not represent other shores or even less accessible parts of that beach. The design reveals a local pattern; adding more nearby quadrats improves detail there without automatically broadening geographic coverage.
Avoid treating more as automatically better
Larger samples can improve precision and reveal variation, but they cannot fix every design problem. If observations systematically exclude stormy days, hidden animals, or certain habitats, collecting more observations under the same exclusion may reinforce a limited picture. Conversely, a carefully designed smaller sample may provide a useful comparison, though its estimates could be less precise.
Also check replication and independence. Repeated measurements at one site can document temporal change there. They do not equal independent sites for a question about geographic variation. Statistical methods may account for repeated or clustered data, but the reader should understand what was repeated and at what level.
Read the two together
When evaluating a paper, record both sample size and selection process. A compact note might say: “48 net tows; six stations; two visits per station; stations chosen along an accessible coastline.” That is much more informative than “sample size 48.” Then ask whether the design matches the population named in the conclusion and whether the analysis reflects the clustering of observations.
This distinction also helps compare studies. One paper may have fewer samples spread across more habitats, while another has many samples at a single site. They answer different questions. Looking at number and design together lets you judge not just how much data were collected, but what those data can reasonably represent.
Have a correction or a source to suggest? Send a note →