Fig. 3: Interpretation requires nuanced reflection considering both statistical and biological significance.

In the following scenario, the need for nuanced interpretation that focuses on the biology of the effect is demonstrated. Graphs plot the marginal means (the average value of a particular group when all other variables are held constant) and the 95% confidence intervals. A Consider a model induction scenario, where the inducing agent leads to a change in a critical parameter such that both males and females are in the diseased state (above the blue dotted line) however, the effect size of that effect differs between the two sexes studied. B The resulting animals are then treated with treatment 1, which had a statistically significant interaction, such that sex explained variation in the treatment effect. However, in this scenario whilst the effect size is different, the biological impact is the same – both sexes studied are no longer in the disease state. C Whilst when the model induced animals are treated with treatment 2, the statistical analysis finds that no significant interaction (the lines are parallel) and estimates a main effect of the treatment. However, in this scenario, whilst the effect size is equivalent the resulting biological state is that the females are still in the disease state whilst the males are not. These hypothetical scenarios demonstrate how important it is to consider the biology in addition to the statistical significance.