Preliminary analyses
Means, standard deviations, and correlations for all variables are presented in Table 1. Correlation analyses showed that perceived stress was positively associated with smartphone addiction, r = 0.18, p < 0.01, indicating that perceived stress was a risk factor for smartphone addiction in medical college students. Psychological capital was negatively associated with smartphone addiction, r = − 0.29, p < 0.01. In addition, negative emotions was positively related to smartphone addiction, r = 0.31, p < 0.01, indicating that medical college students with high negative emotions were more likely to get addicted to smartphone. Finally, perceived stress was positive related to negative emotions, r = 0.20, p < 0.01.
Table 1
Means, standard deviations, and correlations of the main study variables
Variable
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
8
|
1.Gender
|
—
|
|
|
|
|
|
|
|
2.Age
|
−.13**
|
—
|
|
|
|
|
|
|
3.Only child
|
.11**
|
.01
|
—
|
|
|
|
|
|
4.Home location
|
.04
|
.13**
|
.46**
|
—
|
|
|
|
|
5.Perceived stress
|
−.01
|
−.07
|
.01
|
.01
|
—
|
|
|
|
6.Negative emotions
|
−.03
|
.06
|
−.02
|
.00
|
.20**
|
—
|
|
|
7.Psychological capital
|
.15**
|
−.05
|
−.01
|
−.04
|
.03
|
−.42**
|
—
|
|
8.Smartphone addiction
|
−.04
|
.05
|
.04
|
.05
|
.18**
|
.31**
|
−.29**
|
—
|
M
|
1.81
|
20.5
|
1.42
|
1.39
|
31.96
|
22.14
|
123.55
|
39.08
|
SD
|
0.40
|
1.40
|
0.50
|
0.49
|
3.45
|
6.31
|
16.73
|
9.89
|
Note. N = 769. SD: standard deviation. |
*p < 0.05. **p < 0.01. |
Testing for Mediation Effect
In Hypothesis 1, we anticipated that negative emotions would mediate the relationship between perceived stress and smartphone addiction in medical college students. To test this hypothesis, we followed MacKinnon’s four-step procedure to establish mediation effect[39], which requires (a) a significant relation between perceived stress and smartphone addiction in medical college students; (b) a significant association between perceived stress and negative emotions; (c)a significant relation between negative emotions and smartphone addiction after controlling for perceived stress; and (d) a significant coefficient for the indirect path between perceived stress and smartphone addiction through negative emotions. The bias-corrected percentile bootstrap approach determines whether the last condition is satisfied.
Regression analyses indicated that, in the first step, perceived stress positively predicted smartphone addiction in medical college students, b = 0.18, p < 0.01 (see Model 1 of Table 2). In the second step, perceived stress positively predicted negative emotions, b = 0.20, p < 0.01 (see Model 2 of Table 2). In the third step, when we controlled for perceived stress, negative emotions significantly and positively predicted smartphone addiction, b = 0.31, p < 0.01 (see Model 3 of Table 2). Finally, the bias-corrected percentile bootstrap method indicated that the indirect effect of perceived stress on smartphone addiction through negative emotions was significant, ab = 0.06, SE = 0.10, 95% CI = [0.10, 0.24]. The mediation effect accounted for 33.3% of the total effect. Overall, the above four criteria for establishing mediation effect were fully satisfied. Therefore, Hypothesis 1 was supported.
Table 2
Testing the mediation effect of perceived stress on smartphone addiction
Predictors
|
Model 1 (MCSSA)
|
Model 2(NE)
|
Model 3(MCSSA)
|
b
|
t
|
b
|
t
|
b
|
t
|
S
|
0.18
|
5.09**
|
0.20
|
5.63**
|
0.12
|
3.55**
|
NE
|
|
|
|
|
0.31
|
9.08**
|
R2
|
0.03
|
|
0.04
|
|
0.10
|
|
F
|
25.88**
|
|
31.73**
|
|
82.42**
|
|
Note. N = 769. Each column is a regression model that predicts the criterion at the top of the column. PS: perceived stress; NE: negative emotions; MCSSA: medical college students smartphone addiction. |
**p < 0.01. |
Testing for moderated mediation
As noted, Hypothesis 2 predicted that psychological capital would moderate the direct and/or indirect associations between perceived stress and smartphone addiction via negative emotions. To examine this hypothesis, we used the PROCESS macro (Model 59) developed by Hayes to test for moderated mediation[41]. Specially, we estimated the parameters for three regression models. In Model 1, we estimated the moderating effect of psychological capital on the relation between perceived stress and smartphone addiction in medical college students. In Model 2, we estimated the moderating effect of psychological capital on the relation between perceived stress and negative emotions. In Model 3, we estimated the moderating effect of psychological capital on the relation between negative emotions and smartphone addiction. The specifications of the three models are shown in Table 3.
Table 3
Testing the moderated mediation effect of perceived stress on medical college students smartphone addiction
Predictors
|
Model 1 (MCSSA)
|
Model 2(NE)
|
Model 3(MCSSA)
|
b
|
t
|
b
|
t
|
b
|
t
|
PS
|
0.18
|
5.43**
|
0.20
|
6.39**
|
0.14
|
4.18**
|
PC
|
−0.29
|
−8.65**
|
−0.42
|
−13.21**
|
−0.21
|
−5.77**
|
PS×PC
|
−0.04
|
−1.21
|
−0.07
|
−2.04*
|
−0.03
|
−0.75
|
NE
|
—
|
—
|
—
|
—
|
0.19
|
4.88**
|
NE×PC
|
—
|
—
|
—
|
—
|
−0.03
|
−0.72
|
R2
|
0.12
|
—
|
0.22
|
—
|
0.15
|
—
|
F
|
35.40**
|
—
|
73.80**
|
—
|
27.01**
|
—
|
Note. N = 769. Each column is a regression model that predicts the criterion at the top of the column. PS: perceived stress; NE: negative emotion; PC: psychological capital; MCSSA: medical college students smartphone addiction. |
*p < 0.05. **p < 0.01. |
Moderated mediation was established if either or both of two patterns existed[40, 41]: (a) the path between perceived stress and negative emotions was moderated by psychological capital (first stage moderation), and/or (b) the path between negative emotions and smartphone addiction was moderated by psychological capital (second stage moderation).
As Table 3 illustrates, in Model 1, there was a significant effect of perceived stress on smartphone addiction, b = 0.18, p < 0.01, but this effect was not moderated by psychological capital, b = − 0.04,p > 0.05. Model 2 showed that the effect of perceived stress on negative emotions was significant, b = 0.20, p < 0.01, and more importantly, this effect was moderated by psychological capital, b = − 0.07, p < 0.05. For descriptive purposes, we plotted predicted negative emotions against perceived stress, separately for low and high levels of psychological capital (1 SD below the mean and 1 SD above the mean, respectively) (Fig. 2). Simple slope tests indicated that for medical college students with high levels of psychological capital, perceived stress was not significantly associated with negative emotions, b simple = 0.145, p = 0.15. However, for medical college students with low levels of psychological capital, perceived stress was significantly associated with negative emotions, b simple = 0.22༌p < 0.05. That is, in the low psychological capital group, perceived stress has a significant positive predictive effect on negative emotions. This shows that the influence of perceived stress on negative emotions decreases with the increase of psychological capital. In other words, the indirect influence of perceived stress on smartphone addiction through negative emotions decreases with the increase of psychological capital. Model 3 indicated that there was a significant effect of negative emotions on smartphone addiction, b = 0.19, p < 0.01, but this effect was not moderated by psychological capital, b = − 0.03, p > 0.05.
The bias-corrected percentile bootstrap method further indicated that the indirect effect of perceived stress on smartphone addiction through negative emotions was moderated by psychological capital, with the index of moderated mediation = − 0.01, SE = 0.01, 95%CI = [− 0.01, − 0.00]. For medical college students low in psychological capital, perceived stress had an adverse impact on smartphone addiction through increased negative emotions, b = 0.16, SE = 0.05, 95%CI = [0.08, 0.27]. In contrast, the indirect effect was much weaker for medical college students high in psychological capital, b = 0.06, SE = 0.03, 95%CI = [0.02, 0.13]. Thus, Hypothesis 2 was partially supported.