Reading path: Coastal water observation →
A single line can hide variation
A plotted mean compresses a group of observations into one value. That can make comparisons easier, but it conceals how closely the observations cluster. Two sites might have the same average temperature even though one has stable readings and the other varies widely among samples. Showing spread gives readers context for the summary.
For example, if three replicate samples at a location measure 7, 8, and 9 units, their mean is 8. A second location with replicates of 3, 8, and 13 also averages 8. A plot of means alone makes the sites appear identical; the individual values reveal that the second sample set is more dispersed. The example illustrates why a central value needs supporting context.
Choose a display that matches the data
For a small number of observations, plot the individual points. Readers can see the number of samples, clusters, and unusual values directly. For repeated measurements, a line with a clearly defined range or interval may be more legible, but the caption should say what the band represents. A standard deviation, range, standard error, or confidence interval answers a different question; do not use these labels as if they were synonyms.
When observations are paired or grouped by station, show that structure if it matters. Connecting repeated measurements from the same location may reveal a pattern hidden by pooled summaries. Conversely, a shaded region without a clear definition can create an impression of precision without explaining its basis.
Separate observed spread from uncertainty
Replicate spread describes variation among the observed measurements. An uncertainty interval may instead express uncertainty around an estimate, depending on the method. Neither automatically represents every source of error, such as calibration, sampling choices, or differences in deployment. State what was included and what was not when the distinction affects interpretation.
Suppose a seasonal average is calculated from several stations. A narrow interval around that average does not necessarily mean each beach had similar conditions; aggregation can mask local variation. Include the station-level observations or a separate panel when the question concerns differences among places.
Write a readable caption
A useful caption names the summary, the spread or interval, the sample count, and the unit. If some times or stations have fewer replicates, say so or make the varying counts visible. Readers should not have to infer whether a band represents raw minimum and maximum values or a model-based interval.
Showing spread does not make a result weaker. It makes the evidence easier to evaluate and helps distinguish a consistent pattern from one that depends on a few variable observations.
Have a correction or a source to suggest? Send a note →