Figure 3.

Distribution of $X_{\text{sim}}$ as a result of permutations of animal exposures. X evaluates whether the change from 25BPA to 250BPA is different from changes between other consecutive concentrations in data sets from the chronic study. To evaluate the significance of results concerning X, we used the permutation test. The measurements performed in each animal were not rearranged; only the exposure labels were permutated. We computed X for 10,000 permutations and performed a convolution for the contribution of each data set. We generated the distribution $X_{\text{sim}}$, which approximates the one of X. We marked the thresholds for values higher than 95% and 99.5% of the distribution of $X_{\text{sim}}$. $X_{\text{observed}}$ above this threshold leads us to decide against the $H_0$ with $p<0.05$ and 0.005, respectively. (A) $X_{\text{sim}}$ for Criterion A(1.2) with 10,000 iterations: the ratio between consecutive values has to be at least 1.2 to be taken into account. (B) $X_{\text{sim}}$ for Criterion B(0.5) with 10,000 iterations: the p-value between consecutive values has to be at least 0.5 (t-test) to be taken into account. In both cases, we see that the simulation converges toward a smooth distribution. Number of animals per group $n=8–10$, number of groups: 6. Note: BPA, bisphenol A; $H_0$, null hypothesis.