Table 2.
Bayesian log-normal mixed-effects model of the relationship between environmental factors and geometric mean E. coli concentration at 15 beaches in the Metro Vancouver Region, 2013–2021 (other parameters, including the correlation effects of the varying slopes and the coefficients for year, are included in Table C in the Supplementary Material)
| Outcome/parameter a | Estimate b | 95% credible interval | R-hat c | Bulk ESS c | Tail ESS c |
|---|---|---|---|---|---|
| Fixed-level effects | |||||
| Intercept | 3.17 | (2.79, 3.54) | 1.01 | 439 | 821 |
| Previous sample day log E. coli geometric mean | 0.23 | (0.13, 0.34) | 1.00 | 901 | 1,496 |
| 48-h total rainfall | 0.20 | (0.16, 0.24) | 1.00 | 2,998 | 2,692 |
| Mean salinity | −0.28 | (−0.41, −0.13) | 1.00 | 1,212 | 1,674 |
| Antecedent dry days | 0.01 | (−0.03, 0.04) | 1.00 | 4,860 | 2,965 |
| 2-h mean air temperature | 0.31 | (0.21, 0.41) | 1.00 | 1,521 | 2,415 |
| 24-h mean UV | −0.13 | (−0.22, −0.05) | 1.00 | 1,450 | 2,167 |
| 24-h mean air temperature * 24-h mean UV (interaction term) | 0.19 | (0.11, 0.26) | 1.00 | 2,167 | 2,443 |
| Group-level effects for beach (SD) | 1.02 | (1.00, 1.04) | 1.00 | 6,875 | 2,838 |
a
Models conditioned on the study year as a fixed effect.
b
All the fixed-effects estimates and credible intervals are shown here on the mean-centred and standardized scale.
c
R-hat values indicate model convergence, with values closer to 1 indicating convergence. Bulk and tail effective sample size (ESS) are indicators of Markov chain sampling efficiency, with higher numbers showing more reliable results.