Table 2.
Bias and standard deviation (shown in brackets) in estimating σ for the 4 estimation methods in 9 different scenarios. The true values of α and σ in each scenario are shown in the first 2 columns, and μ = 100 for all scenarios. All values are given to 2 significant figures. ∞a and ∞b indicate, respectively, where 32.4% and 0.3% of replicates yielded infinite estimates
| σ | α | MLE | Saddle | RMA-75 | RMA |
| 5 | 102 | 0.00059 (0.20) | – 0.40 (0.19) | 1.5 (0.71) | 7.0 (1.9) |
| 20 | 102 | – 0.0069 (0.40) | – 0.46 (0.43) | 5.6 (1.0) | 15.0 (1.6) |
| 100 | 102 | 0.003 (1.0) | 7.3 (0.99) | 25.0 (5.0) | 45.0 (5.1) |
| 5 | 103 | – 0.067 (0.62) | – 0.56 (0.56) | 3.2 (4.7) | 32.0 (19) |
| 20 | 103 | – 0.11 (1.2) | – 1.9 (1.1) | 6.0 (5.3) | 44.0 (19) |
| 100 | 103 | – 0.00048 (2.8) | – 5.9 (2.7) | 27.0 (7.8) | 100.0 (16) |
| 5 | 104 | – 0.72 (2.4) | – 1.2 (2.1) | ∞a (∞a) | 310.0 (190) |
| 20 | 104 | – 0.40 (4.0) | – 2.5 (3.6) | ∞b (∞b) | 300.0 (180) |
| 100 | 104 | – 0.52 (8.5) | – 10.0 (7.8) | 36.0 (46) | 360.0 (190) |