Table 1.
Community-level clustering (random effect) and model selection in the use of unimproved sanitation services among households in Ethiopia, 2019 EMDHS.
| Variable | Null model | Model I | Model II | Model III |
|---|---|---|---|---|
| Random effect | ||||
| Variance | 10.44 [8.19, 13.3] | 4.95 [3.8, 6.43] | 2.33 [1.79, 3.02] | 2.2 [1.69, 2.85] |
| ICC | 0.76 [0.71, 0.80] | 0.6 [0.54, 0.66] | 0.41 [0.35, 0.48] | 0.4 [0.34, 0.46] |
| MOR | 21.5 | 8.24 | 4.35 | 4.05 |
| PCV | Ref | 0.53 | 0.78 | 0.79 |
| Model comparison | ||||
| Log-likelihood | − 2914.6 | − 2768.7 | − 2741.9 | − 2656.5 |
| Deviance | 5829.2 | 5537.4 | 5483.8 | 5312 |
| Mean VIF | – | 2.07 | 3.96 | 4.17 |
N.B: ICC = Inter Cluster Correlation coefficient, MOR = Medians Odds Ratio, PCV = Proportional Change in Variance, and VIF = Variance Inflation Factor.