Table 3. Multilevel logistic regression analysis of choosing a private versus a public specialist in the 35–65 year-old population of Malmö, 2006, Values are odds ratios (OR) and 95% confidence interval (CI) unless stated otherwise.
| Simple logistic regression analysis | Multilevel logistic regression analysis | ||
|---|---|---|---|
| Model 1 | Model 2 | Model 3 | |
| Specific individual average effects | |||
| Men vs. women | 0.96 (0.92–1.01) | 0.94 (0.88–1.01) | 0.94 (0.88–1.01) |
| Age groups | |||
| 35–39 | Reference | ||
| 40–44 | 1.01 (0.93–1.09) | 1.07 (0.94–1.20) | 1.07 (0.94–1.20) |
| 45–49 | 1.02 (0.94–1.11) | 1.22 (1.07–1.37) | 1.22 (1.07–1.37) |
| 50–54 | 1.08 (1.00–1.17) | 1.25 (1.10–1.41) | 1.25 (1.10–1.41) |
| 55–59 | 1.21 (1.12–1.31) | 1.30 (1.16–1.46) | 1.30 (1.16–1.46) |
| 60–64 | 1.20 (1.10–1.30) | 1.20 (1.06–1.35) | 1.20 (1.06–1.35) |
| High vs. low income | 2.13 (2.02–2.24) | 1.14 (1.06–1.22) | 1.13 (1.04–1.22) |
| Specific contextual average effects | |||
| High vs. low neighbourhood income | 3.50 (2.13–5.78) | ||
| 80% IOR | 0.09–130.28 | ||
| POOR (%) | 33 | ||
| General contextual effects* | |||
| Neighbourhood variance | 4.479 (3.699–5.502) | 3.980 (3.277–4.892) | |
| PCV (%) | 11 | ||
| ICC (%) | 57.8 (53.1–62.7) | 54.9 (50.1–59.9) | |
| MOR | 7.53 (6.42–9.37) | 6.71 (5.62–8.25) | |
| AUC | 0.600 (0.593–0.606) | 0.895 (0.891–0.899) | 0.895 (0.891–0.899) |
| AUC change* | 0.295 | 0.000 | |
| Goodness of fit | |||
| DIC | 44726 | 24647 | 24648 |
| DIC change* | -20079 | 1.28 | |
IOR: interval odds ratio. POOR: proportion of opposed odds ratios. PCV: proportional change in the variance. ICC: intra-class correlation coefficient. MOR: median odds ratio. AUC: area under the receiver operating characteristic curve. DIC: Bayesian diagnostic information criterion.
*: change in relation to the previous model