Table 2.
Mediation Analyses Between Social Engagement Variables and Depressive Symptoms by Gender
| Not living alone | Social network | Activity participation | ||||
|---|---|---|---|---|---|---|
| Men | Women | Men | Women | Men | Women | |
| IV/loneliness (path a) | ||||||
| B (SE) | −.69 (.20)*** | −.24 (.10)* | −.01 (.01) | −.02 (.01)*** | −.04 (.02)* | −.04 (.01)** |
| Beta | −.21*** | −.12* | −.13 | −.17*** | −.15* | −.14** |
| Loneliness/Depression (path b) | ||||||
| B (SE) | 1.47 (.25)*** | 1.77 (.20)*** | 1.56 (.25)*** | 1.69 (.20)*** | 1.53 (.25)*** | 1.73 (.20)*** |
| Beta | .31*** | .41*** | .33*** | .39*** | .33*** | .40*** |
| IV/Depression (path c) | ||||||
| B (SE) | −3.24 (.86)*** | −1.05 (.40)** | −.08 (.03)* | −.13 (.03)*** | −.16 (.07)* | .23 (.06)*** |
| Beta | −.21*** | −.13** | −.14* | −.22*** | −.14* | −.18*** |
| IV/Depression controlling for loneliness (path c′) | ||||||
| B (SE) | −2.22 (.82)** | −.62 (.36) | −.06 (.03)* | −.09 (.03)** | −.11 (.06) | .16 (.05)** |
| Beta | −.14** | −.08 | −.11* | −.15** | −.10 | −.13** |
| Indirect effect (a × b) with 95% CI | −1.02(−2.27; −.14) | −.43(−.89; −.03) | −.02(−.05;.01) | −.04(−.07; −.02) | −.05(−.10; −.01) | −.07(−.13; −.02) |
| Sobel’s z valuea | −3.10** | −2.46* | −1.59 | −3.18** | −2.18* | −2.70** |
Notes. Notations (a, b, c, and c′) can be found in Figure 1. All analyses were conducted with a control of background variables (age, education, length of stay in the United States, and self-rated health).
Both unstandardized (B) and standardized (Beta) coefficients are presented. It is important to note that unstandardized coefficients are used for testing mediation and moderated mediation models using the bootstrap method and Sobel tests. Because coefficients are standardized within each sample, standardized coefficients may help compare the strengths of effects within sample.
aSobel test was separately conducted to confirm the results from the bootstrapping method.
*p < .05. **p < .01. *p < .001.