Parameter of OR and test of goodness of fit for multilevel models, Benishangul Gumuz Region, Northwest Ethiopia, 2021
| Models | Fixed intercept (−cons (95% CI)) | Random effect as level 2 variance var (−cons (95% CI)) | Intraclass correlation coefficient (ρ) | Log-likelihood (LR) deviance | Significance of LR test versus logistic regression (P value) |
| ANC fourth and more* | |||||
| Empty model | 2.54 (1.7 to 3.78) | 1.92 (1.23 to 3.04) | 0.37=37% | −1161.48 | <0.0001 |
| Full model | 0.26 (0.04 to 1.67) | 2.76 (1.54 to 4.96) | 0.46=46% | −633.06 | <0.00001 |
| Skilled delivery* | |||||
| Empty model | 1.48 (1.17 to 1.87) | 0.59 (0.36 to 0.96) | 0.15=15% | −1402.31 | <0.0001 |
| Full model | 1.18 (0.3 to 4.66) | 1.1 (0.63 to 1.92) | 0.25=25% | −706.65 | <0.00001 |
| PNC fourth* | |||||
| Empty model | 1.59 (0.99 to 2.54) | 2.72 (1.72 to 4.33) | 0.45=45% | −1099.36 | <0.0001 |
| Full model | 0.08 (0.01 to 0.71) | 2.25 (1.27 to 3.98) | 0.40=40% | −523.25 | <0.00001 |
A p value less than 0.05 is statistically significant and the data fit for the multilevel model.
*Multilevel regression model applied to measure the effect of factors on outcome.
ANC, antenatal care; CI, Confidence Interval; LR, log-likelihood ratio; OR, Odds ratio; PNC, postnatal care.