3. Results
The mean age of the participants was 20.69 ± 1.61 years old. There were 336 (34.4%) males and 615 (62.9%) females (with 27 (2.8%) missing); 917 (93.8%) local students and 61 (6.2%) international students (including those from mainland China). There were 418 (42.7%) students in year two, 322 (32.9%) students in year three, and 238 (24.3%) students in year four.
Table 1 shows that the measures used in this study were internally consistent. Cronbach’s α values ranged from 0.65 to 0.96, and inter-item correlations ranged from 0.27 to 0.63.
Table 2 presents the results of the correlation analyses. STRESS variables (COVID stress, DASS-stress, difficulties, fatigue) were positively associated with psychological morbidity indicators (PTSD, depression, and suicidal behavior) and IA (
ps < 0.01), and the aforementioned variables were all negatively correlated with PP indicators (life satisfaction, adversity beliefs, family functioning, flourishing, resilience, and emotional competence) (
ps < 0.05). STRESS, psychological morbidity, and IA were positively correlated with each other, and they were negatively correlated with PP (
ps < 0.001).
Table 3 and
Table 4 present results of two mediation analyses, respectively. In
Table 3, STRESS positively predicted IA with a high total effect (
B = 0.98,
p < 0.001). Psychological morbidity also positively predicted IA (
B = 0.10,
p < 0.001). Psychological morbidity partially mediated the association between STRESS and IA (
Estimate = 0.31, BC 95%CI = [0.16, 0.47],
p < 0.001) with a significant direct effect (
B = 0.66,
p < 0.001).
Table 4 shows the mediating effect of PP on the association between STRESS and IA. The total effect (
B = 0.98,
p < 0.001) and the direct effect (
B = 0.89,
p < 0.001) of STRESS on IA were significant. PP negatively predicted IA (
B = −0.37,
p < 0.01) and partially mediated the association between STRESS and IA (
Estimate = 0.08, BC 95%CI = [0.02, 0.05],
p < 0.01).
Table 5 and
Table 6 show results of the moderated mediation analyses (
Figure 4 and
Figure 5), respectively. STRESS significantly predicted IA through psychological morbidity as a mediator, and PP moderated the mediating effect by mitigating the predictive effect of STRESS on psychological morbidity. The interactive effect between STRESS and PP on psychological morbidity was significant (
B = −0.55,
p < 0.001). The predictive effect of STRESS on psychological morbidity was weaker (
B = 2.22,
p < 0.001) when PP was high (M + 1 SD), and it was stronger (
B = 2.92,
p < 0.001) when PP was low (M–1 SD). Furthermore, the index of moderated mediation was significant (95%CI did not cross 0) (
Index = −0.06, BC 95%CI = [−0.10, −0.03]), indicating that PP negatively moderated the mediation effect of psychological morbidity. Specifically, the mediation effect of psychological morbidity under high PP was smaller (
Estimate = 0.23, BC 95%CI = [0.12, 0.34]) than that under low PP (
Estimate = 0.30, BC 95%CI = [0.16, 0.45]). The contrast between the mediation effects under high PP and low PP was significant (95% CI did not cross 0) (
Contrast = −0.07, BC 95%CI = [−0.12, −0.03]).
4. Discussion
The present study investigated the association between COVID-19 related stress and IA in university students in Hong Kong, with psychological morbidity as a risk factor and positive psychological attributes as protective factors. The study is significant as there is scant literature on this research area in higher education in Hong Kong [
13,
78]. As IA is a common problem in university students during the pandemic, the study contributes to our understanding of the risk and protective factors of IA and provides direction for effective intervention and prevention programs.
Regarding risk factors, the present study showed that COVID-19 related stress (indexed by COVID stress, general stress, difficulties encountered, and pandemic fatigue) positively predicted IA, which supports Hypothesis 1a. The result is in line with the general literature suggesting a negative association between general stress and IA. For example, higher perceived stress positively predicted higher IA in Indian college students [
79]. Perceived stress also positively predicted IA in Malaysian youth [
80]. The result was also consistent with the few existing empirical studies on specific COVID-19-related stress and IA. For example, a study on Indonesian college students showed that perceived COVID-19 stress was positively correlated with IA [
7]. Based on 6061 medical students in Chinese universities, elevated life stress and uncertainty stress during the pandemic also positively predicted IA [
9]. According to compensation theory and the cognitive-behavioral model of IA, students’ engaging in Internet addictive behaviors is a form of “compensation” for perceived stress, which involves a misconception that using the Internet could help them avoid or escape from stress [
81,
82]. As there is limited research on the predictive effect of COVID-19-related stress on IA in university students, the result of this study contributes to this research area.
Results of this study also showed that psychological morbidity (indexed by depression, PTSD, and suicidal behavior) positively predicted IA in university students, supporting Hypothesis 1b. This is in line with the extant literature. Existing studies showed a positive association between depression and IA in university students [
17,
18,
19]. Depression also predicted a latent profile group of IA in a study on first-year male undergraduate students [
23]. In addition, Gavurova et al. [
10] found that depression was a significant predictor of IA in Czech and Slovak college students during the pandemic. For PTSD, it positively predicted IA in university students [
21] and adolescents [
20]. PTSD also predicted IA in university students in Hong Kong during the pandemic [
8]. For suicidal behavior, research found that suicidal thoughts and plans positively predicted IA [
83]. Taken together, findings of the present study suggest that psychological morbidity is an important risk factor of IA in university students during the pandemic. Theoretically, IA is a problematic behavior developed when individuals use the Internet as a way to temporarily escape from external or internal stressors [
24,
25]. Zhao et al. [
24] also found that escape or avoidance is a central symptom of IA. In short, increased psychological morbidity may constitute a major internal stress which then triggers IA in university students under the pandemic.
In addition, this study showed that psychological morbidity mediated the connection between stress related to COVID-19 and IA, which supports Hypothesis 1c. The findings enrich the existing scientific literature on the underlying mechanisms for the relationship between stress and IA under the pandemic. The relationship between stress and depression has been extensively examined in the scientific literature and the pathway from stress to depression has been theoretically proposed and empirically investigated by scholars in psychology and neuroscience [
84,
85]. Scientific studies found that stress could lead to brain disturbance which is a central feature of depression [
85]. Against the existing literature, it is reasonable to argue that the prolonged stress related to COVID-19 may lead to psychological symptoms which then trigger students’ Internet addictive behaviors as a way to avoid or escape from the psychological symptoms, such as depressive moods.
Regarding the protective factors of IA, positive psychological qualities based on several indicators negatively predicted IA in university students during the pandemic, supporting Hypothesis 2a. The result is consistent with the extant literature. Conceptually, the paradigm of positive psychology highlights the protective function of positive psychological qualities in preventing mental illness and addictive behaviors in youth [
27,
28]. Particularly, the protective role of important positive constructs, including life satisfaction, flourishing, positive beliefs, resilience, emotional competence, and positive family functioning, were highlighted and were also supported by the empirical literature [
31,
34,
38]. Life satisfaction refers to “the cognitive assessment of one’s life as a whole”, which is not only regarded as an outcome variable but also an important indicator in positive psychology [
86] (p. 129). Empirical studies conducted in different contexts suggest the predictive role of life satisfaction in IA [
30,
31]. Flourishing is defined as an optimal developmental state which comprises “competence, engagement, meaning and purpose, optimism, self-acceptance, supportive relationships” [
87] (p. 1). Literature showed that flourishing was negatively associated with addictive behaviors, such as substance abuse and Facebook addiction [
32,
33]. Resilience and emotional competence, negatively predicted IA in university students in previous studies [
34,
35]. As far as positive beliefs about adversity are concerned, they are important cultural values about the “nature of adversity such as its causes, consequences and the proper coping behavior” [
74] (p. 64). This positive attribute also plays an important role in protecting university students from developing addictive behaviors when experiencing adversities or stress [
36,
74]. Finally, as an important external asset, positive family functioning was found to be negatively associated with IA in university students [
37,
38]. Taken as a whole, the findings of this study contribute to the existing theories and literature on positive roles of these psychological attributes. They also echo the recent development that positive psychological attributes are to be emphasized in the higher education sector [
71,
81,
88]. Practically speaking, positive psychological attributes, such as resilience, emotional competence, positive values, and self-awareness, have been incorporated in the graduate attributes of many higher education institutions [
73,
75].
In addition, results of this study showed that positive psychological attributes mediated the linkage between stress related to COVID-19 and IA in university students, providing support to Hypothesis 2b. The mediating role of positive psychological attributes was also reported in the literature. For example, life satisfaction mediated the association between pandemic anxiety and social media addiction [
39]. It also mediated the relationship between life stressors and IA [
40]. Resilience was found to be a mediator of the association between COVID-19 fear and smartphone addiction [
41], and emotional intelligence mediated the relationship between different stressors and addictive behaviors, such as smoking and smartphone addiction [
42,
43]. Based on ecological system theory and the positive youth development (PYD) perspective, an individual’s positive functioning and development would be influenced by different environmental factors [
89,
90]. The increased environmental stress may undermine individuals’ positive psychological attributes [
91], hence reducing their protective role in IA.
Furthermore, the present study revealed that positive psychological factors moderated the mediating effect of psychological morbidity on the relationship between pandemic-related stress and IA by moderating the association between pandemic stress and psychological morbidity. While there is richer literature on the predictive effect of positive psychological attributes on IA, the way in which it would protect university students from IA by altering the relationship between stress and IA has not been investigated. This protective role is consistent with the literature suggesting that positive psychological attributes could mitigate the negative effect of stress/stressor on psychological morbidity. For example, life satisfaction moderated the positive effects of academic stress on youth depressive disorder [
92]. Emotional intelligence also moderated the positive linkage between peer victimization and IA in primary school students [
93]. Based on university students in Hong Kong, a study found that PYD attributes moderated the linkage between need dissatisfaction and depressive disorder during the pandemic [
53]. Theoretically, while pandemic-related stress may harm students’ mental health, which increases the risk of IA, students with higher positive psychological attributes, such as resilience, adversity beliefs, and emotional competence, would possess more positive resources to cope with stress and adversity [
74] and have better emotional management under stress [
35]. This would reduce their risk for developing psychological morbidity, thus reducing their risk for IA. The result suggests a possible pathway through which positive psychological attributes may have protected university students from IA during the pandemic period, which has not been explored in the existing studies.
The present findings have significant implications for assessment, prevention, and treatment of IA. First, as psychological morbidity was identified as a risk factor of IA, it is necessary to understand the psychological morbidity of people suffering from IA in clinical assessment [
94,
95]. Hence, a more comprehensive and all-round assessment approach should be adopted and “configuring a complete image of the patient’s actual psychopathology” is necessary for assessing IA and devising an effective intervention plan [
96] (p. 80). Prevention and treatment of IA might be more effective if the prevention and treatment of related psychological morbidity is involved [
97]. As comorbidity is very common in addiction, there is a need to take psychological morbidity of those who have problems of IA into account.
Second, the protective role of positive psychological qualities in IA indicates that the assessment of IA should also consider assessing the positive psychological attributes of the person. In addition, the prevention and intervention of IA should consider promoting the important positive psychological attributes to strengthen protective factors of IA. For example, a study reported the effectiveness of adopting multi-family group therapy (focusing on enhancing parent–child communication and bonding) in reducing IA in adolescents [
98]. The project P.A.T.H.S focusing on promoting positive youth development in Hong Kong adolescents was evidenced to be highly effective in reducing IA in adolescents by promoting positive psychological attributes as protective factors in high school and university students [
99,
100]. In addition, a study on a group-based positive psychology intervention for IA showed effectiveness of the intervention with reduction in IA intensity in the experiment group [
101]. However, education on positive psychological attributes is in deficit in Hong Kong [
102], and inclusion of assessment of positive psychological attributes in the context of IA is not common.
Despite the insightful nature of the study, several limitations should be noted. First, the research was cross-sectional in nature which cannot give definitive evidence on the causal relationships between the variables. Hence, research based on longitudinal design should be conducted in the future. Second, the study used quota sampling which has the limitation of generalization, although it is commonly adopted in studies under COVID-19 [
53]. Third, the participants of the study were recruited from one university in Hong Kong. Further research should include participants from other universities within and outside Hong Kong. Fourth, data were collected at the later stage of Wave 5 of COVID-19 in summer of 2022 during which the perceived stress may not have been as severe as that of the previous waves. Fifth, as students in different study years encounter different situations and tasks, it is interesting to examine the differences in risk and protective factors among students in different study years in future research. Sixth, it is also interesting to examine the relative roles of different risk and protective factors in IA in future research.