Figure 17. Scatterplots with linear model fits showing how relative error (standard error relative to the mean), is reduced by sample size (see legend), but also by mean louse density (left panel) and louse prevalence (mid panel). The coefficient of variation had a weak and inconsistent effect on relative error at a given sample size (right panel). Each point represents the cage mean from a standardised sampling event, calculated after 10, 20 or 30 fish.
