Multilevel logit regression estimates based on a four level binomial logit model for a null model with no predictor variables in the fixed part and a random variance component for households, communities, and regions (model 1); and for a model with individual and household predictor variables in the fixed part model and a random variance component for households, communities, and regions (model 2)
| Fixed parameters | Model 1 | Model 2 |
|---|---|---|
| Figures in the parentheses represent the standard errors. All estimates are significant at 0.05 probability level or smaller, except those marked by *. | ||
| Constant | −2.39 | −4.412 |
| Age | 0.042 (0.000) | |
| Female | 0.252 (0.028) | |
| Single | *−0.030 (0.036) | |
| Separated/divorced | 0.279 (0.052) | |
| Widowed | −0.253 (0.040) | |
| Ethnic | *−0.070 (0.056) | |
| Years of education | −0.073 (0.004) | |
| Very poor | 1.080 (0.143) | |
| Poor | 1.027 (0.129) | |
| Low income | 0.725 (0.116) | |
| Middle income | 0.553 (0.115) | |
| High income | 0.324 (0.124) | |
| Public health insurance | 0.559 (0.074) | |
| Other health insurance | 0.274 (0.083) | |
| Unemployed | 0.328 (0.070) | |
| Inactive | 0.489 (0.032) | |
| Urban | 0.145 (0.032) | |
| Random parameters | ||
| Level 4: between regions | 0.066 (0.030) | 0.023 (0.012) |
| Level 3: between communities | 0.083 (0.011) | 0.067 (0.010) |
| Level 2: between households | 3.052 (0.005) | 2.332 (0.061) |