Table 1 Type I error rate of testing non-inferiority for different fixed sample sizes under different sample allocation scenarios. \(p_{NC}\) is the probability of retreatment after beef cattle being treated with negative control. ENFO is enrofloxacin; Z is the new treatment. The log odds ratio of Z to ENFO is 0.2 which indicates Z is inferior to ENFO. Sample allocation is the number of samples assigned to each treatment. For example, (87, 1140, 1173) means the number of samples assigned to NC, ENFO, and Z are 87, 1140, and 1173
total sample size | \(p_{NC}\) | \(p_{ENFO}\) | \(p_{Z}\) | sample allocation (NC, ENRO, Z) | with existing network | non-inferiority type I error |
|---|---|---|---|---|---|---|
2400 | 0.681 | 0.2229 | 0.2613 | (800, 800, 800) | No | 4.24% |
2400 | 0.681 | 0.2229 | 0.2613 | (800, 800, 800) | Yes | 3.83% |
2400 | 0.681 | 0.2229 | 0.2613 | (87, 1140, 1173) | Yes | 4.02% |
3000 | 0.681 | 0.2229 | 0.2613 | (1000, 1000, 1000) | No | 3.87% |
3000 | 0.681 | 0.2229 | 0.2613 | (1000, 1000, 1000) | Yes | 3.32% |
3000 | 0.681 | 0.2229 | 0.2613 | (87, 1448, 1465) | Yes | 3.91% |
3600 | 0.681 | 0.2229 | 0.2613 | (1200, 1200, 1200) | No | 3.85% |
3600 | 0.681 | 0.2229 | 0.2613 | (1200, 1200, 1200) | Yes | 3.28% |
3600 | 0.681 | 0.2229 | 0.2613 | (87, 1757, 1756) | Yes | 3.66% |