Estimates of OLS regression with matching subsamples
| (1) | (2) | (3) | (4) | (5) | (6) | |
| Self-reported health | ADL limitations | Social integration | ||||
| Original | Matching | Original | Matching | Original | Matching | |
| Panel A: lower ranking | ||||||
| Coefficient (95% CI) | 0.1109** (0.019 to 0.203) | 0.0836* (− 0.012 to 0.179) | −0.0376 (−0.156 to 0.081) | −0.0353 (−0.170 to 0.099) | 0.0768 (−0.099 to 0.252) | 0.0495 (−0.133 to 0.232) |
| Other controls | Yes | Yes | Yes | Yes | Yes | Yes |
| Province dummies | Yes | Yes | Yes | Yes | Yes | Yes |
| Observations | 12 258 | 3145 | 15 753 | 4080 | 15 753 | 4080 |
| R-squared | 0.243 | 0.246 | 0.139 | 0.165 | 0.069 | 0.101 |
| Sobel test of mediation for social integration | ||||||
| Total effect | Direct effect | Indirect effect | Proportion of total effect that is mediated | |||
| Self-reported health | 0.088* | 0.09* | −0.002 | −2.7% | ||
| ADL limitations | −0.038 | −0.035 | −0.004 | 9.6% | ||
| Panel B: higher ranking | ||||||
| Coefficient (95% CI) | 0.1558*** (0.056 to 0.256) | 0.1466*** (0.042 to 0.251) | −0.2506*** (−0.377 to −0.124) | −0.1829** (−0.347 to −0.019) | 0.4091*** (0.221 to 0.597) | 0.3797*** (0.181 to 0.579) |
| Other controls | Yes | Yes | Yes | Yes | Yes | Yes |
| Province dummies | Yes | Yes | Yes | Yes | Yes | Yes |
| Observations | 12 228 | 3145 | 15 721 | 4080 | 15 721 | 4080 |
| R-squared | 0.245 | 0.246 | 0.137 | 0.165 | 0.074 | 0.101 |
| Sobel test of mediation for social integration | ||||||
| Total effect | Direct effect | Indirect effect | Proportion of total effect that is mediated | |||
| Self-reported health | 0.147*** | 0.128** | 0.019*** | 12.6% | ||
| ADL limitations | −0.19** | −0.148* | −0.041*** | 21.9% | ||
95% CI shown in parentheses.
***p<0.01, **p<0.05, *p<0.1.
ADL, activities of daily living; OLS, Ordinary Least Squares.