Table 3.
Parameters of the dual Gauss, Gauss-Weibull, and Weibull-Gauss distribution models for the overall sample
| Parameter | Result* | Mean* | Bootstrap: 95% confidence interval | |
|---|---|---|---|---|
| Lower bound | Upper bound | |||
|
Model 4: Gauss + Gauss** (N = 4717) -Log-likelihood = 17,272.30 | ||||
| Mean (μ1) | 16.14 | 16.12 | 15.18 | 17.07 |
| Std. dev. (σ1) | 6.06 | 6.06 | 5.57 | 6.60 |
| Weight (w) | 0.74 | 0.74 | 0.65 | 0.81 |
| Threshold (θ) | 27.30 | 27.30 | 24.40 | 30.35 |
| Mean (μ2) | 34.19 | 34.17 | 30.87 | 37.20 |
| Std. dev. (σ2) | 7.50 | 7.48 | 6.11 | 8.74 |
|
Model 5: Gauss + Weibull (N = 4717) -Log-likelihood = 17,261.31 | ||||
| Scale (μ) | 16.24 | 16.18 | 15.10 | 16.91 |
| Shape (σ) | 6.15 | 6.13 | 5.56 | 6.57 |
| Weight (w) | 0.26 | 0.27 | 0.20 | 0.37 |
| Threshold (θ) | 27.70 | 27.56 | 24.10 | 30.05 |
| Scale (η) | 38.08 | 37.93 | 34.48 | 40.15 |
| Shape (β) | 5.18 | 5.19 | 3.88 | 6.41 |
|
Model 6: Weibull + Gauss -Log-likelihood = 17,229.37 | ||||
| Scale (η) | 20.92 | 20.88 | 20.05 | 21.57 |
| Shape (β) | 2.85 | 2.86 | 2.74 | 3.00 |
| Weight (w) | 0.85 | 0.85 | 0.80 | 0.88 |
| Threshold (θ) | 32.70 | 32.60 | 30.00 | 34.70 |
| Mean (μ) | 38.83 | 38.73 | 36.58 | 40.27 |
| Std. dev. (σ) | 5.25 | 5.28 | 4.44 | 6.32 |
*The result column shows the parameter values from the sample with the highest likelihood in the original data. The mean column shows the average parameter value from the 1000 resamples
**μ1 and σ1, and μ2 and σ2 correspond to the parameters of Guass1 and Gauss2 distributions depicted in Fig. 3b