AbstractsBiology & Animal Science

The newly proposed toxicity equivalence approach using percent effect endpoint in WET test can address the limitations of current statistical endpoints NOEC and IC25. The statistical implementation of toxicity equivalence approach is demonstrated in this thesis. One asymptotic confidence interval and two bootstrap based confidence intervals are compared by using simulation study based on their coverage probabilities. The performance of asymptotic CI is the best among them and is recommended with the proper sample size. In formulating the hypothesis test for toxicity equivalence approach in WET test, the non-inferiority hypothesis test is proposed to replace the use of the traditional hypothesis test. The null hypothesis is set as toxicity of effluent (H0: PE=<E, where E is the non-inferiority margin) while the alternative hypothesis is that the effluent is non-toxic (within the non-inferiority margin) compared with control. Furthermore, the power and sample size of the recommended asymptotic CI are assessed by using simulation study. This proposed statistical implementation of toxicity equivalence approach is also applied to WET dataset.