*5.3. Group Comparison*

Our findings partially supported the first hypothesis of our study: only female students showed an increase in depression and anxiety symptomatology and increasing feelings of inadequacy, inferiority, hostility, suspiciousness, and isolated lifestyle. Despite the significant effect, the effect size ranged from a small to medium, suggesting that these differences are not robust. For this reason, the results should be carefully interpreted. As a pandemic is an extraordinary event that cannot be replicated, in interpreting this finding, we should consider that other additional factors could explain the differences found between groups. For example, we have to keep in mind that every group is composed of different students asking to participate at different times of the year. For this reason, we should consider both variables related to students' university commitments, such as exams and examination sessions, and other variables, such as coping strategies to deal with the different containment measures. Even if only tentative interpretations can be suggested given the small to medium effect size, it is possible to hypothesize, looking at the results as a whole, that during the quarantine, the most affected areas were related to mood and the quality of interpersonal relations. Indeed, this pattern of symptomatology is recurring in patients with major depression [46]. Moreover, it is possible to hypothesize that social distancing contributed to the arising of interpersonal relation uncertainties and the presence of negative expectations about interpersonal relationships. In addition, since COVID-19 spreads mainly between people who are in close contact with each other, it could have contributed to raising the perception of the other as a risk for one's health, resulting in hostility or paranoid behaviour. Previous studies, which focused mainly on intimate relationships during the lockdown, indicated that high-stress levels were associated with a decline in intimate relationships [47]. Moreover, the attachment style of partners predicts interpersonal problems and the efficacy of problem-solving strategies [48]. Other studies suggested that working at home can exacerbate familial conflict [49] and that living with others contributed to increased psychological distress [50]. Interestingly, it seems that

quarantine did not have an influence on male students. The results showed no significant differences between male groups. These data are consistent with other studies suggesting that COVID-19 had a more negative impact on females than males [51]. Some evidence suggested that during the lockdown, females reported higher levels of stress [50] and anxiety [19] than males and that they were at more risk of developing depression compared to males [18]. Lastly, our findings did not support our second hypothesis. The results did not show a significant increase in psychological distress in the Third Phase. However, despite the non-significant difference between groups, the males' descriptive statistics show that students reported more anxiety and depressive symptoms, hostility and suspiciousness, and overall psychological distress during the Third Phase. Furthermore, even though the difference was not statistically significant, the effect size ranged from small to medium. The small male sample size may have prevented sufficient power to detect differences between groups. Therefore, the findings observed in this study need to be clarified by increasing our sample size. Moreover, to the best of our knowledge, less is known about the longitudinal changes in males' mental health since most studies focused on the gender differences in relation to mental health during COVID-19. Future studies should investigate variables that could affect the different mental health trajectories across the pandemic between males and females.

#### *5.4. Limitations and Future Directions*

The study has some limits. First, most of the participants were female. For this reason, we decided to run an analysis for males and females separately. However, the tiny male sample size could have increased the occurrence of Type II Errors, reducing the chance of identifying a significant difference that could exist. Future studies should focus on a strategy to bring males closer to the topic of mental health. Second, our sample recruitment was not totally random: students decided spontaneously to participate in the NoiBene program, which could suggest that they have a particular interest in improving their mental health or that they perceive a need for help. This could have contributed both to the high percentage of psychopathological students and to the heterogeneity of our samples. Even if these limits could have contributed to reducing the study's generalizability, the different flow of students that asked to be included in NoiBene at different moments represent an essential indicator of help-seeking behaviour. Lastly, the four groups that we considered in our analysis were composed of different participants, so we did not have the opportunity to examine any changes over time. However, we had the opportunity to compare psychopathological dimensions between groups to understand how different phases of the COVID-19 restrictions impacted the individual's psychological health. It would be interesting in future studies to differentiate between students living in different regions with different risk scenarios, or to take into account risk and protective factors that underly the psychopathology that occurred during the pandemic.

Besides these limitations, the study has different strengths: we had the chance to assess students before and during the pandemic. This allowed us to observe any change that was associated with the quarantine and the following restrictive rules. Moreover, we conducted individual interviews with every student reporting a high level of psychological distress: this had a fundamental impact in terms of helping students to understand the best treatment according to their needs [45]. In addition, self-report questionnaires can yield much valuable and diagnostic information, but they cannot be used to define a diagnosis. Conducting individual interviews allowed us to go beyond this limit and be sure about students' symptomatology.
