Table A3.
Mediation analysis.
| Donation |
Volunteering |
||||||
|---|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | |
| Pos. Private (T1) | 0.825 | 0.760 | 0.801 | ||||
| (1.462) | (1.446) | (1.443) | |||||
| Neg. Private (T3) | 2.720** | 2.233 | 2.069 | ||||
| (1.378) | (1.389) | (1.396) | |||||
| Pos. Public (T2) | 2.196* | 1.863 | 0.094* | 0.096** | |||
| (0.084) | (1.243) | (0.048) | (0.048) | ||||
| Neg. Public (T4) | 1.043 | 0.998 | 0.088** | 0.088** | |||
| (1.259) | (1.256) | (0.051) | (0.051) | ||||
| Mediator | No | Trust | Trust | No | Resp. | No | Resp. |
| Mediator Fn. Form | Linear | Non-par. | Linear | Linear | |||
| Controls | Y | Y | Y | Y | Y | Y | Y |
| Observations | 672 | 672 | 672 | 682 | 682 | 689 | 689 |
| Rsquare | 0.053 | 0.072 | 0.08 | 0.007 | 0.027 | 0.012 | 0.013 |
| T1 = T3 | 0.025 | 0.057 | 0.068 | ||||
| T2 = T4 | 0.038 | 0.068 | 0.003 | 0.003 | |||
Notes: This table reports how treatment coefficients change when we control for mediating variables. Regressions include the full set of treatment variables but for readability only treatment coefficients are reported for treatment arms that affect the respective mediator. We furthermore restrict regressions to specifications for which we observe significant treatment effects. For trust, we estimate linear and non-parametric models with dummy variables for each response option in the Likert scale. For responsibility, we estimate models with the binary mediator. All estimations are OLS. Robust standard errors are in parentheses. The bottom rows present p-values from a test of equal coefficients. * , ** , ***