Fig. 3

The difference in growth rate that can be detected is data-dependent. The results of comparing two curves and calculating the log Bayes factor, \(\ln \mathcal {B}_{23}\), for hypothesis 2 (H ₂ following previous convention; the curves have the same growth rate) versus hypothesis 3 (H ₃ following previous convention; the curves have different growth rates) as the difference in growth rate, Δ μ _max, between the two is increased. Here, the first curve is fixed, whilst the growth rate for the second is increased and n replicates with computationally random noise (with level σ) are generated for each difference in growth rate. The solid lines and shading represent the mean and standard deviation respectively for each case. The dashed line represents the line below which we favour hypothesis 3 over hypothesis 2 based on Jeffreys’ scale. The effect of using curves with different numbers of points and noise levels is explored, as well as using different numbers of replicates