1. Introduction
At the end of 2019, the COVID-19 pandemic broke out initially in Wuhan, China, and expanded rapidly, creating a massive public health crisis worldwide. During our data collection period in April 2020 the World Health Organization reported over 1 million confirmed cases globally, resulting in a more than tenfold increase in less than a month [
1].
The rapid transmission rate, and the lack of available and effective pharmaceutical treatments and vaccines, required the use of non-pharmaceutical interventions. Examples included social distancing behaviors such as using masks, keeping spatial distance, avoiding crowded places and meeting with friends and family members. These behaviors were believed to be effective in reducing the spread of the disease [
2]. Many national authorities and international health organizations such as the Israeli Ministry of Health, the National Health Commission of the People’s Republic of China and the World Health Organization published guidelines to provide individuals with recommended behaviors. National governments enforced these guidelines based on local situations.
The adoption of preventive behaviors depends on the willingness of the public to engage in radical behavioral changes. Since these behavioral changes involve significant social and economic costs [
3,
4], both governments and individuals have to balance the cost and benefits of modifying their daily activities. Therefore, personal, social, cultural, and developmental factors affect the decision to adopt these behaviors [
5,
6].
Israel and China are at different points on the epidemiological curve. Israel reported its first confirmed imported case on 21 February 2020, about two months later than the suspected cases in Wuhan, China [
7]. On April 1, 2020, China reported 82,631 cumulative confirmed cases, and Israel reported 5129. China had 3321 deaths, 157 times higher than Israel. During this period the two countries differed in their adoption of avoidant behaviors.
From the perspective of a society, behavioral changes as a response to public health recommendations are heavily dependent on the acquisition of new social norms of interaction. Social contact patterns and hygiene behaviors are habitual and are difficult to modify. The COVID-19 pandemic provides us with a natural laboratory for the investigation of rapid and dramatic changes in social behavior [
8].
As a country with a Western culture, Israelis are more likely than the Chinese to express their emotions with close physical contact such as kissing and hugging [
9]. The behaviors involved in social distancing are antithetical to Israeli society. In China, the COVID-19 pandemic coincided with the Chinese spring festival, during which it is customary to participate in family reunions and visit friends and family. Imposing social distancing demands during this period was very challenging for Chinese officials. As to cultural differences [
10], Israeli society is more inclined to individualism, whereby personal desires often outweigh social demands. In contrast, China is a typical collectivistic society characterized by a strong commitment to social norms and strict punishment for deviation from these norms [
11,
12,
13].
Given these similarities and differences, we can obtain useful insights into the factors that affected how people regarded the outbreak, the authorities, and their surroundings, and their subsequent adoption of avoidant behaviors or failure to do so. To accomplish this goal, we utilize the 3Cs model that was created to determine whether or not people would get vaccinated. The 3C’s model was originally used as a framework to analyze the three major components of the vaccine hesitancy in the Strategic Advisory Group of Experts (SAGE) on Immunization, a multidisciplinary working group of scholars and practitioners with the World Health Organization in 2012 [
14]. Since them, this model has been used in various health programs [
15,
16] to understand how individuals’ confidence, complacency, and constraint influence the complex decision-making process on immunization [
17]. We assume that vaccines are a special case of preventive behavior. In the absence of any vaccine for the COVID-19 virus, we should investigate whether the factors involved in the 3Cs model--confidence, complacency and constraint—can be used to determine whether or not people will adopt the preventive behaviors recommended by public health authorities such as social distancing. Furthermore, we explore the ability to generalize the findings to China and Israel, two countries that vary a great deal in the nature of their societies. Doing so will provide evidence to evaluate the extent to which the 3Cs model is universal and invariant according to culture
4. Results
Table 2,
Table 3 and
Table 4 list the variances, covariances, correlations of indicators, and Cronbach’s α for each constructed latent variable. The correlations among the indicators constructed as one latent variable were stronger than the correlations with indicators outside their latent variables. This result supported the evidence that the indicators used to construct the three latent variables did not closely correlate with one another.
The Israeli participants and Chinese participants differed in their levels of the latent variables of complacency. While during the outbreak and in daily life both groups indicated that they thought the virus was serious and they were very susceptible to it, the Israelis thought they were more susceptible to it than the Chinese both in daily life and during the outbreak period. In contrast, the Chinese participants regarded the virus as more serious than the Israelis, both in daily life and during the outbreak period.
For the latent variables of constraints, both the Israeli and Chinese participants thought that engaging in healthy behaviors was beneficial for preventing disease in their daily life. The level of this belief was slightly more intense when compared to the statistics during the outbreak. At the same time, they reported strong assessments about their self-efficacy in being able to engage in healthy behaviors in their daily life or during the outbreak. Nevertheless, the Chinese participants expressed more self-efficacy in this regard than the Israeli participants.
For the latent variable of confidence in social institutions for dealing with the outbreak, both the Israeli and Chinese participants expressed consistent and high levels of confidence in the various institutions. However, the Chinese participants were more confident in the Chinese institutions than the Israeli participants were in the Israeli institutions.
Table 5 presents the descriptive analysis of avoidant behavior.
For avoidant behavior as a dependent variable, the most salient difference between the participants from the two countries was using a face mask when going out. Chinese participants were significantly more likely to adopt this behavior than Israelis (M = 4.69, SD = 0.731 for Chinese participants, M = 2.57, SD = 1.503 for Israeli participants). On average, the Chinese participants were more likely not to share a towel with others and to avoid going to places where it was known that the virus was locally transmitted. However, on average, the Israelis were more likely than the Chinese to avoid public transportation and crowded places.
Multivariate Analysis
Based on the structural equations and statistical assumptions discussed above, we developed the path diagram of the conceptualized 3Cs avoidant behavior model illustrated in
Figure 1.
As
Table 6 indicates, the original proposed model was a good fit with the data for the Israeli and Chinese participants. For the Israeli participants, χ
2(165,
N = 703) = 646 (
p = 0.00), χ
2/df = 3.92, RMSEA = 0.06(0.059–0.070 90% Confidence Interval (CI); GFI = 0.91, SRMR = 0.059. For the Chinese participants, χ
2(166,
N = 0 703) = 629 (
p = 0.00), χ
2/df = 3.79, RMSEA = 0.06(0.058–0.068 90%CI); GFI = 0.91, SRMR = 0.064.
We used a multi-group analysis to compare the fit of the structural equation models for the Israeli and Chinese participants, and to test the invariance of the model. The unconstructed model showed a better fit [χ
2(333,
N = 703) = 1256.452 (
p = 0.00), χ
2/df = 3.773, RMSEA = 0.044(0.042–0.047 90%CI); GFI = 0.912, SRMR = 0.06] than other models with invariance constraints (
Table 6). The estimations of the standardized coefficients for Israeli participants in the multi-group analysis were identical to the values we found in the previous single model. The same outcome was evident for the Chinese participants. We found no difference in the paths. Therefore, we can conclude that the conceptualized model fit the Israeli and Chinese data reasonably well, even though the parameters for the Israeli and Chinese participants were not equal.
5. Discussion
Can the 3Cs model that was developed to determine whether people would get vaccinated to protect their health be applied more widely to determine whether people will engage in behaviors designed to protect them from COVID-19? Furthermore, if we establish that the 3Cs model is useful in this regard, is it applicable to cultures that vary from one another? To answer these questions, we gathered data from Israel and China to investigate the roles of confidence in social institutions, complacency in the face of the threat, and constraints in expectations about the outcomes of engaging in avoidant behavior (see
Figure 2 and
Figure 3).
The results show that constraints are negatively related to the adoption of avoidant behavior, whereas confidence is positively related to engaging in such behavior. However, these relationships are direct in Israel and indirect in China. In Israel, complacency is negatively related to the adoption of avoidant behavior, whereas in China, this relationship is positive. Therefore, we conclude that the 3Cs model works reasonably well in both the Israeli and Chinese samples, and could be applied to a pandemic.
We found support for our hypothesis that constraints are negatively associated with compliance with avoidant behavior. Indeed, the measurements of it in accounting for the direct effects on both the Israeli and Chinese participants’ behavior were statistically close. This robust relationship contributes to the literature on the role of constraints in complying with healthy behaviors and their indicators. Furthermore, the role of expectations about outcomes and efficacy is becoming increasingly recognized as important antecedents to the adoption and maintenance of recommended behaviors.
At the same time, we found that constraints fully mediated the association between confidence and compliance with avoidant behavior in the Israeli model, and partially mediated this association in the Chinese model. This finding demonstrates the influential role of constraints in behavioral changes. Therefore, we conclude that eliminating perceptions about constraints leading to low expectations about outcomes and efficacy is essential for public health programs and communications, especially during the ever-changing, uncertain period of a pandemic.
Another contribution of this research is identifying the central role of confidence in social institutions as contributing to increased adherence to avoidant behavior, both directly and indirectly. In the Chinese model, the association between confidence and engaging in avoidant behavior was mediated by constraints with an effect of β = −0.40, which extended the differential impact of the hypothesis. This result indicates that confidence in social institutions during the COVID-19 pandemic does not affect the adoption of behavioral changes directly. However, it can remove constraints on expectations and, in turn, increase the likelihood that people will adopt avoidant behavior.
Cultural differences may account for the difference in the direct and indirect effects. For example, the Chinese participants reported much more confidence in social institutions than the Israeli participants. While studies have documented that confidence in the authorities and their executive departments is essential for promoting regulations dealing with public health, it alone does not necessarily ensure that people will engage in such behavior. According to our findings, only when people believe that the behaviors are beneficial (outcome expectations) and they are confident that they can engage in them (efficacy expectations) will they actually adopt them. This finding is consistent with previous research on other pandemics such as Influenza A (H1N1) in Hong Kong, In that study, trust in social institutions was not directly associated with avoidant behavior, but was more strongly associated with efficacy expectations. Thus, efforts should be made to promote more confidence in people’s ability to engage in behavioral change by enhancing their understanding of the outcomes of doing so and developing their efficacy to engage in such behavior. This suggestion highlights the communication dynamics between social institutions, especially governments, and the public. Communication strategies during a pandemic should put more emphasis on encouraging people to engage in avoidant behavior by raising their expectations about the beneficial outcomes of doing so and inspiring their efficacy expectations.
Another interesting finding was that the level of confidence in the government, hospitals, and medical workers was rather close in both Israel and China. Given that the levels of trust in these three groups were similar, we can conclude that the public regards them as sharing responsibility during the pandemic. Therefore, if one party in this triumvirate loses the public’s confidence, the other two remaining parties will be adversely influenced by this loss of trust.
The level of complacency in Israel and China had opposite associations with the adoption of avoidant behavior, and the absolute magnitude of the two associations was close. These divergences could be attributed to the difference in perceptions of complacency between these two groups. The Chinese participants felt they were less likely to become infected than the Israelis. However, if they were infected, they expected the impact to be more severe than the Israelis did.
During the data collection period, the outbreak was gradually brought under control in China. Wuhan emerged from lockdown on April 8, 2020, indicating that the threat of infection had declined. This information might have made the Chinese participants feel that it was less likely that they would become infected. Their stronger sense about the severity of the infection might be due to the large number of deaths and infected people who were still in treatment. In contrast, in Israel, new confirmed cases surged at the end of March, but there were only 20 deaths. Given these statistics, the Israelis might have concluded that they were more likely to become infected, but doing so would not necessarily be fatal.
The combination of the Chinese participants’ perceptions about their relative lack of risk of contracting the virus and the high death rate associated with the disease complicated the role of complacency in the adoption of avoidant behaviors. This positive association between complacency and engaging in avoidant behaviors was consistent with a previous study about Influenza A (H1N1) in which the relatively mild effects of the flu provided little motivation for people to change their behavior [
70]. As such, future research should conduct a thorough investigation of the role of complacency in a pandemic.
Another important contribution of this study is our demonstration of the ability to generalize the 3Cs model to contexts other than just becoming vaccinated. Our study is the first to explore the validity of the model in other contexts, particularly with regard to preventive behavior in a pandemic. Despite the fact that the implications of complacency, confidence, and constraints are not identical with regard to being vaccinated and engaging in avoidant behaviors, the model can still provide a mechanism for understanding behavioral change. In accordance with our hypotheses, we indicated that complacency about the possible risks, confidence in social institutions, and constraints regarding expectations are associated with people’s behavior during a pandemic.
Our comparative study also demonstrated the ability to generalize the 3Cs model. We used a conceptual model as a comparative framework to understand how multiple variables function differently in both countries. The Israeli and Chinese participants varied significantly in their levels of complacency, confidence, and constraints, as well as the extent of their engagement in avoidant behavior. For example, both the Israeli and Chinese participants reported significant awareness of the benefits of social distancing. However, the Chinese participants were more optimistic about the situation. They indicated that they felt less susceptible to the disease, had fewer constraints and more confidence than the Israeli participants, but regarded the virus as more serious. Therefore, they were more likely to engage in six out of the nine recommended behaviors. The only exceptions were avoiding shaking hands, crowded, airless places, and public transportation.
Our findings also have several practical implications. First, they suggest that efficient public health campaigns and communications during the outbreak designed to promote the adoption of avoidant behavior must target people’s expectations, particularly with regard to the beneficial outcomes and efficacy of doing so. The goal is to reduce the public’s perceptions about the constraints on their adopting such behavior. Furthermore, these campaigns should promote confidence in social institutions, specifically, the government, hospitals and medical personnel. At the same time, the public’s level of complacency may vary throughout this dynamic situation and deserves a more detailed investigation. Last but not least, the government should work closely with the health care system to build a transparent and responsive communication to the public, to assess individual’s perception, emotional status and increase public support. In the meantime, difficulties in employment, need for daily supplies, epidemic prevention and control supplies, transportation could be the sources of public negative perception and behavior adoption. Policies concerning these issues should be implemented to ease this hardship.
6. Limitations
There are some limitations to this study. First, even though the data came from two countries dealing with COVID-19, the data were cross sectional. Our assessments of perceptions and engagement in avoidant behavior were post hoc and cross-sectional. Thus, we could not evaluate causality properly. The cross-sectional nature of these data limits our ability to draw conclusions about causal directions.
Second, although there is extensive research on whether cognition precedes or follows appraisals, there is no consensus on the issue [
71]. Previous studies showed various possibilities that might contribute to a particular behavior [
72]. In the Israeli model, confidence, complacency, and constraints are antecedents of avoidant behavior, as are complacency and constraints in the Chinese model. Some variables could also influence the behaviors of interest in a serial chain [
73].
At the same time, reverse causality might have affected our interpretation of the perceptions associated with avoidant behavior. Engaging in such behavior might promote positive and negative changes in complacency, confidence, and constraints. Social distancing challenges people’s instincts of connecting with others [
74]. The isolation and loneliness that stems from social distancing might increase psychological and emotional distress. Sudden changes in people’s social interactions may prove burdensome. As a result, behaviors designed to keep people from contracting the virus might have a deleterious effect on other aspects of their physical and mental well-being, including their cognitive perceptions. This possibility suggests that the association between cognitive perceptions and preventive behaviors is more complex and would benefit from longitudinal studies in the future.
In addition, cognitive perceptions and patterns of social contact are culturally based. They are rooted in social norms of behavior, and culturally specific socialization patterns. In this research, we did not collect data about indicators of cultural values. Meanwhile, differences in lifestyle and daily routine could impact individuals’ decision-making processes during the pandemic, and therefore could change the dynamic of virus transmission. These differences roused by geographic, institutional, and social factors could complex the cross-cultural comparison research that has not been included in this study. Doing so would help us determine the interaction between cultural factors and people’s perceptions and behaviors, promoting international comparisons.
Third, the study may suffer from some potential bias. The use of an online survey meant that we could reach only participants who were Internet users. These participants are more likely to receive instant information about the pandemic and may be more avid followers of the news. These factors could bias the results. Furthermore, the data were self-reports, leading to the possibility of social desirability bias resulting from pressure to conform to social expectations. Nevertheless, despite these limitations, our study benefits from the fact that we asked about the actual behaviors people had adopted, not their future plans for doing so. We were also able to validate the possibility of applying the 3Cs model to dealing with the COVID-19 pandemic and provided a cross-national comparison of our hypotheses.