Table 4. Robustness model: Effect of air pollution on mental health, mediated by cognitive and non-cognitive ability using logistic/probit model, 2016†.
| Logistic Model | Probit Model | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| IV_Mandarin usage | Depression | Cognitive ability | IV_Mandarin usage | Depression | IV_Mandarin usage | Depression | Cognitive ability | IV_Mandarin usage | Depression | |
| PM2.5 | 0.038*** (0.015) |
-0.059*** (0.011) |
-0.169*** (0.063) |
0.030** (0.015) |
0.038*** (0.015) |
-0.059*** (0.011) |
-0.087*** (0.030) |
0.030** (0.015) |
||
| Cognitive ability | -0.119*** (0.011) |
-0.119*** (0.011) |
||||||||
| IV_Mandarin usage | -0.019 (0.025) |
-0.019 (0.025) |
||||||||
| Non-cognitive ability | 1.293*** (0.046) |
0.711*** (0.025) |
||||||||
| Covariates | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| County/District Fixed Effects | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Observations | 10214 | 13761 | 13761 | 10214 | 13761 | 10214 | 13761 | 13761 | 10214 | 13761 |
| Adjusted R2 | 0.134 | 0.461 | 0.141 | 0.134 | 0.461 | 0.141 | ||||
| Pseudo R2 | 0.496 | 0.427 | 0.495 | 0.428 | ||||||
†Notes: Standard errors in parentheses.
* p < 0.1
** p < 0.05
*** p < 0.01. The explanation of variables is the same with those in Table 2. The first and sixth columns are run with the instrument (Mandarin used in interview) regressed on non-cognitive ability; the second and seventh columns with depression regressed on PM2.5; the third and eighth columns with the first mediator (cognitive ability) regressed on PM2.5; the fourth and ninth columns with the second mediator (Mandarin used in interview) regressed on PM2.5; and the fifth and tenth columns with depression regressed on PM2.5 and both mediators (cognitive ability and Mandarin used in interview). Both the first and third columns are employing Logistic model, while the sixth and eighth columns are using Probit model.