Social demographic of the participants (N= 10175)
The social demographic characteristics of 10175 female workers and the distributions of LS were presented in Table 1. More than half of (60.7%) women aged below 44, and 39.3% women aged between 45 and 59. As for marital status, the majority of participants (79.9%) were married, follow by unmarried (12.2%). 49.9% women had an educational degree below undergraduate, and over half of the participants (61.9%) with annual income between 30,001 and 60,000. Most participants (83%) were pregnant or had children. The average LS score of the women were 20.73±7.08. Different distributions of demographics were found regarding LS scores for age, marital status, educational degree, annual income, occupation and pregnancy or childbirth. Women older than 45 years, receiving good education, and having higher annual income tended to have higher levels LS(P<0.001). Regarding marital status, women who were married were higher than those of other groups(P<0.001). Women have been pregnant or had children showed higher levels LS(P<0.001).More detailed information was shown in Table 1.
Table 1 Mean scores of LS according to social demographic characteristics (N= 10175)
Variables
|
N (%)
|
LS(Mean ± SD)
|
F/t
|
P
|
Age(years)
|
|
|
-8.545
|
<0.001
|
≤44
|
6172(60.7)
|
20.25±7.06
|
|
|
45-59
|
4003(39.3)
|
22.47±7.05
|
|
|
Marital status
|
|
|
80.594
|
<0.001
|
Married
|
8128(79.9)
|
21.24±6.95
|
|
|
Unmarried
|
1239(12.2)
|
19.33±7.09
|
|
|
Divorced or widowed
|
746(7.3)
|
17.75±7.11
|
|
|
other
|
62(0.6)
|
17.44±9.30
|
|
|
Educational degree
|
|
|
9.888
|
<0.001
|
Below undergraduate
|
5079(49.9)
|
20.14±7.08
|
|
|
Bachelor
|
4913(48.3)
|
21.04±7.06
|
|
|
Master and above
|
183(1.8)
|
21.07±7.26
|
|
|
Annual income
|
|
|
87.988
|
<0.001
|
≤30,000 yuan
|
2694(26.5)
|
19.11±7.13
|
|
|
30,001-60,000 yuan
|
6299(61.9)
|
21.02±6.93
|
|
|
60,001-100,000 yuan
|
1104(10.8)
|
22.77±6.95
|
|
|
>100,000 yuan
|
78(0.8)
|
23.76±7.29
|
|
|
Occupation
|
|
|
18.597
|
<0.001
|
Medical hygiene
|
1990(19.6)
|
20.24±7.06
|
|
|
Education
|
3485(34.3)
|
21.48±6.798
|
|
|
Financial industry
|
289(2.8)
|
20.08±7.08
|
|
|
Electronic industry
|
329(3.2)
|
19.06±7.35
|
|
|
other
|
4082(40.1)
|
20.45±7.24
|
|
|
Pregnancy or childbirth
|
|
|
5.783
|
<0.001
|
Yes
|
8444(83.0)
|
20.91±7.04
|
|
|
No
|
1731(17.0)
|
19.82±7.20
|
|
|
Correlations among continuous variables
Table 2 displayed the results of Pearson’s correlation analysis. Age, stress, anxiety symptoms, and LS were significantly correlated with each other in the expected direction. And the results indicated that both family stress and anxiety symptoms had significant negatively correlation with LS.
Table 2 Correlations among family stress, anxiety symptoms and LS
Variables
|
1
|
2
|
3
|
4
|
|
1.Age
|
1
|
|
|
|
|
2.Family stress
|
0.047**
|
1
|
|
|
|
3.Anxiety symptoms
|
-0.016
|
0.513**
|
1
|
|
|
4.Life satisfaction
|
0.082**
|
-0.500**
|
-0.374**
|
1
|
|
Note: *P < 0.05, **P < 0.01
Testing for the Mediation Model
Hypothesis 1 stated that anxiety symptoms will mediate the association between family stress and LS. The multiple liner regression analysis [40] was adopted to test the mediation effect. As presented in Table 3. The first step indicated that the regression of family stress on LS was significant (B =-1.13, p <0.001), and the regression of family stress on anxiety symptoms was also significant in the second step (B =0.74, p <0.001). The last step demonstrated that the regression of anxiety symptoms on LS was significant (B =-0.24, p <0.001), and when the family stress and mediator (anxiety symptoms) were both included in the model as independents, the path coefficient of the family stress on LS remained significant (B =-0.95, p <0.001).
Besides, the PROCESS macro (Model 4) was used and a bootstrap procedure conducted to examine the mediating effect. 5000 bootstrapping samples was generated from the original dataset(N=10175) via random sampling. The indirect effect of family stress on LS through anxiety symptoms was significant (SE=0.01, 95%= [-0.03, -0.02]). The 95% biased-corrected confidence interval did not contain zero, which confirmed that anxiety symptoms played a mediating role on the association between family stress and LS.
Table 3 Mediated regression analysis for family stress and anxiety symptoms on LS
Variables
|
B
|
SE
|
β
|
t
|
F
|
Step 1.
|
|
|
|
|
|
Independent: Family stress
|
-1.13
|
0.02
|
-0.51
|
-59.25***
|
1801.09***
|
Dependent: LS
|
|
|
|
|
|
Step 2.
|
|
|
|
|
|
Independent: Family stress
|
0.74
|
0.01
|
0.52
|
60.49***
|
1831.29***
|
Dependent: Anxiety symptoms
|
|
|
|
|
|
Step 3.
|
|
|
|
|
|
Independent: Family stress
|
-0.95
|
0.02
|
-0.43
|
-43.39***
|
1310.60***
|
Mediator: Anxiety symptoms
|
-0.24
|
0.02
|
-0.15
|
-15.61***
|
|
Dependent: LS
|
|
|
|
|
|
Note: *P < 0.05, **P < 0.01,***P < 0.001
Testing for the Moderated Mediation Model
The results of the moderated mediation analyses for family stress, anxiety symptoms, age and LS were showed in Table 4. The PROCESS macro (Model 59) was used to test the moderated mediation, as is recommended by Bolin[39]. In Model 1, family stress was negatively related to LS (β=-0.51, p <0.001), and the interaction term between family stress and age was negatively related to LS (β=-0.03, p <0.01), which indicated that the age could moderated the association between family stress and LS. Model 2 was significant, family stress was positively related to anxiety symptoms (β=0.51, p <0.001), and the interaction term between family stress and age was not related to anxiety symptoms, which indicated that the relationship between family stress and anxiety symptoms could not be moderated by age. Model 3 was also significant, anxiety symptoms was negatively related to LS (β=-0.15, p <0.001), and the interaction term between anxiety symptoms and age was positively related to LS (β=0.03, p <0.1), which demonstrated that the age moderated the association between anxiety symptoms and LS.
These results indicated that the relationships between both family stress and LS as well anxiety symptoms and family stress are moderated by age. To more thoroughly understand the moderating effect of age, the present study plotted the association between family and LS, separately for younger and middle-aged groups. Simple slope test presented the association between family stress and LS (Fig 1). For women from younger group, the family stress was significantly and negatively associated with LS (β=-0.39, t=-28.00, p <0.001). For women from middle-aged group, family stress was also negatively related to LS but considerably weaker (β=-0.47, t=-33.36, p <0.001). And Fig 2 showed the simple slope test results of the association between anxiety symptoms and LS scores. As for women from young group, anxiety symptoms was significantly and negatively associated with LS (β=-0.18, t=-12.70, p <0.001), and for middle-aged women, resilience was still negatively correlated with LS, but much strengthener (β=-0.13, t=-9.40, p <0.001).
The index of the moderated mediation was 0.01, SE=0.006, and the biased-corrected 95% confidence interval for index of moderated mediation was from 0.002 to 0.025, which did not contain zero. For younger workers, family stress negatively predicted LS (effect=-0.09, t=0.008,95%CI [ -0.108, -0.077]), while among those with higher age, family stress had a significant but weaker prediction on LS (effect=-0.07, t=0.008,95%CI [ -0.082, -0.050]).
Table 4 Moderated mediation effect of anxiety symptoms on LS
Variables
|
Model 1
(Life satisfaction)
|
Model 2
(Anxiety symptoms)
|
Model 3
(Life satisfaction)
|
|
β
|
t
|
β
|
t
|
β
|
t
|
Family stress
|
-0.51
|
-59.36***
|
0.51
|
60.31***
|
-0.43
|
-43.45***
|
Age
|
0.10
|
11.64***
|
-0.04
|
-4.88***
|
0.09
|
10.81***
|
Family stress × Age
|
-0.03
|
-3.19**
|
-0.01
|
-1.53
|
-0.04
|
-4.29***
|
Anxiety symptoms
|
|
|
|
|
-0.15
|
-15.74***
|
Anxiety symptoms × Age
|
|
|
|
|
0.03
|
2.55*
|
R2
|
0.26
|
|
0.26
|
|
0.28
|
|
F
|
1205.20***
|
|
1211.80***
|
|
791.34***
|
|
Note: *P < 0.05, **P < 0.01,***P < 0.001