Mediation effects of subjective well-being and subjective social status on the relationship between social trust and emotional health of older adults
Table 1 presents the descriptive statistics of all the variables used in the analysis. Table 2 shows the regression results on the relationship of social trust and emotional health and the mediation effect of subjective well-being. The dependent variable in Models 1, 3, 4, 6, 7, and 9 was emotional health, and the dependent variable in Models 2, 5, and 8 was subjective well-being. Model 1 shows that trust in family members is significantly and positively associated with respondents’ emotional health (coefficient = 2.854, P < 0.01). Elderly living in urban neighbourhoods showed better emotional health than those living in rural neighbourhoods (coefficient = 0.984, P < 0.05). Compared with male respondents, female respondents had significantly poorer emotional health (coefficient=-1.450, P < 0.01). Further, the higher the level of education, the better was the reported emotional health of the respondents (coefficient = 0.348, P < 0.01). The worse the self-rated health of the respondents, the more likely they were to suffer from poor emotional health (coefficient=-3.078, P < 0.01). Model 2 shows that trust in family members had a significant positive effect on the subjective well-being of the respondents (coefficient = 0.189, P < 0.01). Older, female, and well-educated respondents were more likely to report higher subjective well-being than younger, male, and less-educated respondents (coefficient = 0.014, P < 0.01; coefficient = 0.074, P < 0.05; coefficient = 0.031, P < 0.01; respectively). Respondents with poor health were more likely to have lower subjective well-being (coefficient=-0.210, P < 0.01).
Table 1
Descriptive statistics of the variables used in the analysis (n = 3767)
variables
|
Mean (SD)/Percentage
|
Emotional health (20–80)
|
71.87 (10.12)
|
Trust in family members (1–5)
|
4.80 (0.50)
|
Trust in friends (1–5)
|
4.32 (0.72)
|
Trust in neighbors (1–5)
|
3.88 (0.79)
|
subjective well-being (1–5)
|
3.78 (0.92)
|
Subjective social status (1–10)
|
4.35 (1.80)
|
Age ( > = 60 years old)
|
64.24 (4.27)
|
Neighborhood types
|
|
Rural neighborhoods)
|
75.07
|
Urban neighborhoods
|
24.93
|
Gender
|
|
Male
|
53.20
|
Female
|
46.80
|
Educational level (1–9)
|
2.39 (1.52)
|
Self-rated health (1–5)
|
2.82 (1.03)
|
Table 2
The relationship of social trust and emotional health and the mediation effects of subjective well-being
|
Trust in family members
|
Trust in friends
|
Trust in neighbors
|
|
Model 1: emotional health
|
Model 2: subjective well-being
|
Model 3: emotional health
|
Model 4: emotional health
|
Model 5: subjective well-being
|
Model 6: emotional health
|
Model 7: emotional health
|
Model 8: subjective well-being
|
Model 9: emotional health
|
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Trust in family members
|
2.854***
|
(0.309)
|
0.189***
|
(0.029)
|
2.476***
|
(0.306)
|
|
|
|
|
|
|
|
|
|
|
|
|
Trust in friends
|
|
|
|
|
|
|
1.703***
|
(0.216)
|
0.171***
|
(0.020)
|
1.362***
|
(0.214)
|
|
|
|
|
|
|
Trust in neighbors
|
|
|
|
|
|
|
|
|
|
|
|
|
1.461***
|
(0.197)
|
0.186***
|
(0.018)
|
1.091***
|
(0.197)
|
subjective well-being
|
|
|
|
|
1.999***
|
(0.170)
|
|
|
|
|
1.995***
|
(0.171)
|
|
|
|
|
1.988***
|
(0.172)
|
Age
|
0.004
|
(0.037)
|
0.014***
|
(0.003)
|
-0.025
|
(0.036)
|
0.002
|
(0.037)
|
0.014***
|
(0.003)
|
-0.026
|
(0.036)
|
0.003
|
(0.037)
|
0.014***
|
(0.003)
|
-0.025
|
(0.036)
|
Urban neighborhoods (ref: Rural neighborhoods)
|
0.984**
|
(0.402)
|
-0.020
|
(0.038)
|
1.025***
|
(0.395)
|
1.023**
|
(0.404)
|
-0.009
|
(0.038)
|
1.040***
|
(0.397)
|
1.198***
|
(0.408)
|
0.021
|
(0.038)
|
1.156***
|
(0.401)
|
Female (ref: male)
|
-1.450***
|
(0.322)
|
0.074**
|
(0.030)
|
-1.597***
|
(0.317)
|
-1.403***
|
(0.323)
|
0.076**
|
(0.030)
|
-1.554***
|
(0.318)
|
-1.329***
|
(0.323)
|
0.084***
|
(0.030)
|
-1.497***
|
(0.318)
|
Educational level
|
0.348***
|
(0.116)
|
0.031***
|
(0.011)
|
0.286**
|
(0.114)
|
0.326***
|
(0.116)
|
0.029***
|
(0.011)
|
0.268**
|
(0.114)
|
0.351***
|
(0.116)
|
0.032***
|
(0.011)
|
0.287**
|
(0.114)
|
Self-rated health
|
-3.078***
|
(0.152)
|
-0.210***
|
(0.014)
|
-2.659***
|
(0.153)
|
-3.019***
|
(0.153)
|
-0.200***
|
(0.014)
|
-2.620***
|
(0.154)
|
-3.067***
|
(0.153)
|
-0.201***
|
(0.014)
|
-2.667***
|
(0.154)
|
Constants
|
66.228***
|
(2.918)
|
2.456***
|
(0.275)
|
61.317***
|
(2.896)
|
72.494***
|
(2.684)
|
2.604***
|
(0.251)
|
67.298***
|
(2.675)
|
74.154***
|
(2.631)
|
2.604***
|
(0.245)
|
68.979***
|
(2.624)
|
N
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
R-sq
|
0.147
|
|
0.075
|
|
0.177
|
|
0.142
|
|
0.082
|
|
0.172
|
|
0.140
|
|
0.089
|
|
0.170
|
|
adj. R-sq
|
0.146
|
|
0.073
|
|
0.176
|
|
0.141
|
|
0.081
|
|
0.170
|
|
0.139
|
|
0.088
|
|
0.168
|
|
Log lik.
|
-13764.447
|
|
-4866.960
|
|
-13696.387
|
|
-13775.740
|
|
-4852.439
|
|
-13708.882
|
|
-13779.302
|
|
-4836.975
|
|
-13713.632
|
|
Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. |
Models 4 and 5 show that trust in friends was significantly and positively associated with respondents’ emotional health and their subjective well-being (coefficient = 1.703, P < 0.01; coefficient = 0.171, P < 0.01; respectively). Models 7 and 8 show that trust in neighbours was significantly and positively associated with respondents’ emotional health and their subjective well-being (coefficient = 1.461, P < 0.01; coefficient = 0.186, P < 0.01; respectively). In terms of the influence of demographic factors, the regression results of Models 4 and 7 are similar to those of Model 1, and the regression results of Models 5 and 8 are similar to those of Model 2.
We followed Baron and Kenny [25] to test the presence of mediation effects of subjective well-being. Figure 2 shows the results of Model 3, indicating that adding the variable of subjective well-being to Model 1 weakened the effect of trust in family members on emotional health from coefficient = 2.854 to coefficient = 2.476. Figure 3 shows that after adding the subjective well-being variable into Model 4, the impact of trust in friends on the emotional health of the respondents decreased by 0.341. Figure 4 shows that after adding the subjective well-being variable into model 7, the impact of trust in neighbours on the emotional health of the respondents decreased by 0.370. These results suggest that subjective well-being can effectively reduce the impact of social trust (in family, friends, and neighbours) on the emotional health of older respondents.
Table 3 shows the regression results of the mediation effect of subjective social status on the relationship between social trust and emotional health. Models 10, 12, and 14 show that trust in family members, trust in friends, and trust in neighbours had significant positive effects on respondents’ subjective social status (coefficient = 0.115, P < 0.05; coefficient = 0.095, P < 0.05; coefficient = 0.121, P < 0.01; respectively).
Table 3
The regression results of mediation effects of subjective social status on the relationship between social trust and emotional health
|
Trust in family members
|
Trust in friends
|
Trust in neighbors
|
|
Model 10: subjective social status
|
Model 11: emotional health
|
Model 12: subjective social status
|
Model 13: emotional health
|
Model 14: subjective social status
|
Model 15: emotional health
|
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Trust in family members
|
0.115**
|
(0.057)
|
2.762***
|
(0.306)
|
|
|
|
|
|
|
|
|
Trust in friends
|
|
|
|
|
0.095**
|
(0.040)
|
1.627***
|
(0.214)
|
|
|
|
|
Trust in neighbors
|
|
|
|
|
|
|
|
|
0.121***
|
(0.036)
|
1.365***
|
(0.195)
|
subjective social status
|
|
|
0.804***
|
(0.087)
|
|
|
0.804***
|
(0.087)
|
|
|
0.797***
|
(0.087)
|
Age
|
0.020***
|
(0.007)
|
-0.013
|
(0.037)
|
0.020***
|
(0.007)
|
-0.014
|
(0.037)
|
0.020***
|
(0.007)
|
-0.013
|
(0.037)
|
Urban neighborhoods (ref: Rural neighborhoods)
|
0.208***
|
(0.075)
|
0.817**
|
(0.398)
|
0.213***
|
(0.075)
|
0.851**
|
(0.400)
|
0.235***
|
(0.075)
|
1.011**
|
(0.404)
|
Female (ref: male)
|
0.074
|
(0.060)
|
-1.509***
|
(0.319)
|
0.075
|
(0.060)
|
-1.463***
|
(0.319)
|
0.080
|
(0.060)
|
-1.393***
|
(0.320)
|
Educational level
|
0.128***
|
(0.022)
|
0.245**
|
(0.115)
|
0.127***
|
(0.021)
|
0.224*
|
(0.115)
|
0.129***
|
(0.021)
|
0.248**
|
(0.116)
|
Self-rated health
|
-0.337***
|
(0.028)
|
-2.807***
|
(0.153)
|
-0.332***
|
(0.028)
|
-2.752***
|
(0.154)
|
-0.331***
|
(0.028)
|
-2.803***
|
(0.154)
|
Constants
|
3.050***
|
(0.542)
|
63.776***
|
(2.897)
|
3.180***
|
(0.497)
|
69.938***
|
(2.669)
|
3.106***
|
(0.486)
|
71.680***
|
(2.617)
|
N
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
3767
|
|
R-sq
|
0.064
|
|
0.166
|
|
0.064
|
|
0.161
|
|
0.066
|
|
0.159
|
|
adj. R-sq
|
0.063
|
|
0.165
|
|
0.063
|
|
0.159
|
|
0.064
|
|
0.157
|
|
Log lik.
|
-7423.296
|
|
-13721.965
|
|
-7422.476
|
|
-13733.546
|
|
-7419.775
|
|
-13738.024
|
|
Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. |
We also followed Baron and Kenny [25] to test the presence of mediation effects of subjective social status. Figure 5 shows the results of Model 11, indicating that adding the variable of subjective social status to Model 1 weakened the effect of trust in family members on emotional health from coefficient = 2.854 to coefficient = 2.762. The impact of trust in family members on the emotional health of the respondents decreased by 0.092. Figure 6 shows that after adding the subjective social status variable into Model 4, the impact of trust in friends on the emotional health of the respondents decreased by 0.076. Figure 7 shows that after adding the subjective social status variable into Model 7, the impact of trust in neighbours on the emotional health of the respondents decreased by 0.096. These results suggest that subjective social status can effectively reduce the impact of social trust (in family, friends, and neighbours) on the emotional health of the elderly.
In terms of demographic factors, Model 10 shows that age, neighbourhood type, education level, and self-rated health had significant influence on the subjective social status of the respondents. Among them, the older the respondents, the higher was their subjective social status (coefficient = 0.020, P < 0.01). Compared with the respondents in rural neighbourhoods, the respondents in urban neighbourhoods had significantly higher subjective social status (coefficient = 0.208, P < 0.01). Respondents with a higher level of education also had a higher subjective social status (coefficient = 0.128, P < 0.01), while respondents with poor health had relatively poor subjective social status (coefficient=-0.337, P < 0.01). Models 15 and 17 show similar results.
Moderation effects of subjective well-being and subjective social status on the relationship between social trust and emotional health of older adults
Table 4 shows the regression results of the moderation effects of subjective well-being and subjective social status on the relationship between social trust and emotional health. The results from model 16 show that subjective well-being moderated the relationship between trust in family members and emotional health. The positive relationship between trust in family members and emotional health was weakened by subjective well-being. Model 17 shows that the positive relationship between trust in neighbours and emotional health was weakened by subjective social status.
Table 4
The regression results of moderation effects of subjective well-being and subjective social status on the relationship between social trust and emotional health
|
Model 16: emotional health
|
Model 17: emotional health
|
|
Coefficient
|
S.E.
|
Coefficient
|
S.E.
|
Subjective well-being
|
7.086***
|
(1.398)
|
|
|
Subjective social status
|
|
|
1.210
|
(0.857)
|
Trust in family members
|
4.552***
|
(1.123)
|
1.411*
|
(0.821)
|
Trust in friends
|
0.810
|
(1.084)
|
0.703
|
(0.679)
|
Trust in neighbors
|
1.702*
|
(0.926)
|
1.918***
|
(0.570)
|
Trust in family members # Subjective well-being
|
-0.749**
|
(0.310)
|
|
|
Trust in friends # Subjective well-being
|
-0.102
|
(0.284)
|
|
|
Trust in neighbors # Subjective well-being
|
-0.304
|
(0.239)
|
|
|
Trust in family members # Subjective social status
|
|
|
0.186
|
(0.185)
|
Trust in friends # Subjective social status
|
|
|
-0.047
|
(0.149)
|
Trust in neighbors # Subjective social status
|
|
|
-0.289**
|
(0.125)
|
Age
|
-0.019
|
(0.036)
|
-0.015
|
(0.036)
|
Urban neighborhoods (ref: Rural neighborhoods)
|
1.237***
|
(0.398)
|
1.053***
|
(0.401)
|
Female (ref: male)
|
-1.536***
|
(0.316)
|
-1.449***
|
(0.318)
|
Educational level
|
0.288**
|
(0.114)
|
0.244**
|
(0.115)
|
Self-rated health
|
-2.590***
|
(0.153)
|
-2.719***
|
(0.153)
|
Constants
|
41.202***
|
(5.709)
|
59.688***
|
(4.620)
|
N
|
3767
|
|
3767
|
|
R-sq
|
0.184
|
|
0.174
|
|
adj. R-sq
|
0.181
|
|
0.172
|
|
Log lik.
|
-13680.913
|
|
-13703.620
|
|
Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. |