Is the 'n = 30 rule of thumb' of ecological field studies reliable? A call for greater attention to the variability in our data

Abstract

A common practice of experimental design in field ecology, which relies on the Central Limit Theorem, is the use of the ‘n = 30 rule of thumb’. I show here that papers published in Animal Biodiversity and Conservation during the period 2010–2013 adjust to this rule. Samples collected around this relatively small size have the advantage of coupling statistically–significant results with large effect sizes, which is positive because field researchers are commonly interested in large ecological effects. However, the power to detect a large effect size depends not only on sample size but, importantly, also on between–population variability. By means of a hypothetical example, I show here that the statistical power is little affected by small–medium variance changes between populations. However, power decreases abruptly beyond a certain threshold, which I identify roughly around a five–fold difference in variance between populations. Hence, researchers should explore variance profiles of their study populations to make sure beforehand that their study populations lies within the safe zone to use the ‘n = 30 rule of thumb’. Otherwise, sample size should be increased beyond 30, even to detect large effect sizes.