*3.4. Moderating Effect of Socioeconomic Status*

As the occupational status was not significantly associated with depression in our previous examination (*r* = −0.04, *p* = 0.32), only education level and household income were included in the moderation analysis. The interaction of socioeconomic variables with precautionary behaviors was first examined and the results showed that education level was not significantly related with precautionary behaviors for predicting older adult's depression (β<sup>1</sup> = 0.12, *t*<sup>510</sup> = 0.78, *p* = 0.44, 95%CI = −0.93 to 2.14; β<sup>2</sup> = 0.04, *t*<sup>510</sup> = 0.23, *p* = 0.82, 95%CI = −1.42 to 1.79). For the household income, a significant moderation effect was identified in the analysis (See Figure 1). Results indicated a significant interaction between household income (average vs. below average) and precautionary behaviors (β<sup>1</sup> = 0.26, *t*<sup>510</sup> = 2.53, *p* = 0.012, 95%CI = 0.31 to 2.44), as well as between household income (above average vs. below average) and precautionary behaviors (β<sup>2</sup> = 0.39, *t*<sup>510</sup> = 3.01, *p* = 0.003, 95%CI = 0.70 to 3.34). The interaction contributed to a significant change in the variance explanation (ΔR2 = 0.02, *p* = 0.007). The total moderation model accounted for 9% of the variance in depression (*p* < 0.001). The descriptive plot of the moderating effects of household income on the relationship between COVID-19 precautionary behavior change and depression level among older adults is presented in Figure 2. For older adults with higher levels of household income, there was only a slight negative association between precautionary behavior change and depression level, whereas for those with average and lower levels of household income, prominent associations between behavior change and depression levels occurred.

**Figure 1.** Moderation effect of household income on behavior-depression association (*n* = 516). RCS = Residualized change score; \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001.

**Figure 2.** Plot of simple slopes showing the association between COVID-19 precautionary behavior change and depression level at different categories of household income.

#### **4. Discussion**

To the best of our knowledge, this is the first online cross-sectional study to explore the characteristics of depression, to examine the association between COVID-19 precautionary behaviors and depression levels, and to identify the role of SES in moderating the behavior– depression association among Chinese older adults during the COVID-19 pandemic. The

findings from the study have fully supported the hypotheses. In particular, during the outbreak of COVID-19, older adults' depression levels differed significantly in a series of characteristics, including marital status, living situation, SES indicators (education level and household income), as well as subjective health status and infected cases of acquaintances. After controlling for the demographic covariates, COVID-19 precautionary behaviors showed a significant inverse association with older adults' depression levels. Of the three SES indicators, only household income significantly moderated the association between COVID-19 precautionary behaviors and depression levels among Chinese older adults.

In terms of the characteristics of depression, as suggested in previous studies, individuals who lack social support from families and friends showed significantly higher levels of depression than those with sufficient social support from families and friends [37–39]. Therefore, it is not surprising that in this study, older adults who have married and lived with their spouse, partners, or children indicated a prominent lower depression level. The findings also revealed that older adults who perceived their health status as poor and had acquaintances being infected showed a significantly higher level of depression. These findings are consistent with previous studies, where older adults with these characteristics may experience greater fear of being infected or dying themselves, leading to higher depression levels [32,37,38]. In line with previous evidence in Chinese adolescents and adults, the findings showed that older adults who had higher education levels and higher household income might be less influenced by the COVID-19 pandemic, indicating a comparative lower depression level [5,22]. The discrepancy with previous evidence occurred in the indicator of occupational status [23,24], where no significant difference was found in the current study. This may be attributed to the reason that the majority of our participants were retired older adults (92.6%).

In terms of the association of individual precautionary behavior towards COVID-19 with depression levels, our findings were consistent with previous studies among Chinese non-infected adolescent and adult populations [5], and with a recent study among Japanese adults with depressive symptoms [17]. Older adults who adopted more precautionary behaviors (e.g., hand washing, facemask wearing, and social distancing) were more likely to have lower depression levels during the COVID-19 epidemic. It is worth noting that the change in COVID-19 precautionary behaviors accounted for 3% of the variance in depression levels, while the SES indicators (education levels and household income) and covariates (infected cases of acquaintances, subjective health status), also played a critical role in predicting older adults' depressive states. These findings emphasize the significance of promoting precautionary behaviors during the COVID-19 pandemic among older adults, as well as the importance of considering the socio-demographic characteristics when designing psychological interventions and making relevant policies to improve mental health outcomes among older adults.

In terms of the moderating effect of SES indicators on the behavior–depression association, household income was found to be a significant moderator. This result agrees with previous studies [25], which indicate that the economic dependency significantly interacted with social activity and depression among older adults (β = −0.16, SE = 0.01, *p* < 0.001) [25]. The findings of the current study support the moderating role of household income in the behavior–depression association, revealing that when we motivate older adults to take COVID-19 precautionary behaviors to reduce their depression levels, we need to especially focus on older adults who are at an economically disadvantaged level. From the government's perspective, the findings indicate the importance and necessity of providing relief funding for low-income households to ease the stress of the pandemic. These findings also bear considerable implications for future preventive measures of epidemics among older adults.

This study has several limitations. First, given the urgency of the research needed on the COVID-19 pandemic and the limited resources available, we have to apply an online cross-sectional approach using snowball sampling, so the participants may vary in relation to the actual patterns of the general elderly population (e.g., in the illiterate or semi-illiterate

samples). Moreover, all the variables were measured by self-reported scales, which might lead to recall bias, self-perception bias, and social desirability effects [22,40]. However, the bias has been found to be lower in anonymous online surveys than in telephone or face-toface paper surveys [15,41]. In spite of online methodologies being an efficient means and cost-effective method to conduct surveys, we adopted several strategies to ensure that the online survey was easy-to-operate. However, many elderly participants were confronted with difficulties in the process of the survey (e.g., operational functionality, submission setting). Further actions are needed to make online surveys more user-friendly for elderly populations. Additionally, the demographic and behavioral factors identified in the present study only explained 15% of the variance of depression levels, so other factors need to be investigated in future studies. In addition, the depression levels did not significantly differ in gender, whereas other studies have found a prominent role for the gender variable in the psychological responses towards the pandemic [21,22]. This point deserves further investigation. Finally, the findings of the present study were obtained from a specific age group within a Chinese context; therefore, it is unclear whether these findings would be generalizable to other age groups and different cultural contexts. Notwithstanding the aforementioned limitations, this study provides invaluable information on the characteristics of depression, and the impact of COVID-19 precautionary behaviors when considering depression levels. The study also provides detail relating to the role of SES indicators in moderating the behavior–depression association among Chinese older adults during the COVID-19 pandemic. The research findings presented here could be used as a meaningful reference, adding knowledge and giving new insights into future research promoting precautionary behaviors and relationships between mental health and older adults during the COVID-19 outbreak and potential future pandemics.
