Fig. 6

Increased power when using the Probability Ratio to reanalyse data from Huus et al. 2020. a Boxplots showing the Kau index scores at each age for all taxa that were significantly different by either diet or the interaction of age and diet in two-way ANOVAs (nominal p < 0.05). Each plot shows which of diet and the diet:age interaction were significant and their p-values. Boxes represent the median and IQR and the whiskers the largest and smallest values within one and half IQR of the upper and lower IQR limits. UC represents taxa that were unclassified below the given taxonomy. b Plot as in (a) but for the probability ratio scores. c Left: Scatterplot of the p-values for all the comparisons that were significant in either (a) or(b) showing the p value when using the Probability Ratio (y) or Kau Index (x). The red-line shows y = x, highlighting the lower p values in the probability ratio tests. Right: Scatterplot as for p values but showing the effect size (difference in means between the two diets) at each of the three ages for all of the significant comparisons in (a) and (b). Effect size is quantified as the strictly standardised mean difference (MAL-CON). d Boxplots comparing the absolute coefficient of variation of each taxon’s score within each experimental group (for all taxa detected in the Huus et al. dataset) when using either the Kau Index or probability ratio. Significance shown from Mann-Whitney U test