This study started with a discussion in a focus group in Kuwait. The ten people who participated in the focus group opined that assessing threat and coping strategies is essential to manage the COVID-19 situation. Along with this insight, the group also suggested that different cultural systems and relevant mindsets will differ in adhering to government recommendations. This motivated us to study the research question across different countries with polarized mindsets and different cultural systems. Thus, the study expanded to other countries with different cultures. Due to resource constraints, the sampling was limited to countries that the authors have firsthand experience in explaining the similarities and differences.

A global survey-deploying firm collected the data for this study using online platforms. The firm recruited respondents from the United States, Kuwait, and South Korea in May 2020. The firm sampled respondents using an age, gender, ethnicity, and geographic region–based strata and quota matching process. Participation in the survey was free and voluntary; the respondents filled in electronic informed consent that was shown on the first page of the survey. The firm protects the confidentiality of anonymous respondents.

Data was collected using a survey instrument, as shown in Multimedia Appendix 1 Table A1. The questions asked participants about the cause and current state of the COVID-19 situation, their opinion on the government’s role during the COVID-19 pandemic, the use of health information sources and social media for COVID-19–related information, and PMT measures for the COVID-19 pandemic adapted from previously validated scales [36-39]. The survey items included simple information-seeking questions and several existing validated scales from prior studies [35,40-47].

The survey instrument was pilot-tested using a sample of 48 respondents, leading to minor refinements to a few items. A total of 482 participants took the survey. Because of missing responses to the items, 64 observations were excluded, resulting in a sample size of 418. Responses were coded, validated, and analyzed using Stata version 16 (StataCorp).

Table 1 shows the descriptive statistics and pairwise correlations among the key variables used in this study. Out of 418 participants, 299 (58.7%) were female. The sample’s largest age group was 18-27 years (n=192, 37.3%). This group was followed by the 28-37 years age group (n=150, 29.1%), 38-47 years age group (n=70, 13.6%), 48-57 years age group (n=43, 8.4%), and 58 years or older group (n=60, 11.7%). Multimedia Appendix 1 Figure A1 shows the countrywise comparison of the respondents’ age. In terms of income level, 102 (20.5%) participants make less than US $30,000 annually; 92 (18.5%) make US $30,000-US $50,000; 102 (20.5%) make US $50,000-US $80,000; 65 (13.1%) make US $80,000-US $100,000; 57 (11.5%) make US $100,000-US $150,000; and 79 (15.9%) make more than US $150,000. The household income distribution as varied by country is shown in Multimedia Appendix 1 Figure A2. A detailed distribution of several demographic controls used in the models is available in Multimedia Appendix 1 Table A2.

The main dependent variable in this study is adherence. As shown in Multimedia Appendix 1 Table A1, adherence was measured using three questions of whether they would comply with the social distancing measures. The items’ internal consistencies were tested using Cronbach alpha (0.81), and the standardized score was generated for the adherence variable. Table 1 shows that, on average, adherence is 4.18 out of 5, showing that most people are adhering to social distancing recommendations. The mean adherence level in Kuwait was the highest (mean 4.53, SD 0.81), followed by the United States (mean 4.14, SD 0.92) and South Korea (mean 3.53, SD 1.18). Multimedia Appendix 1 Figure A3 displays the mean adherence across countries.

Three main independent variables were of interest in this study to examine adherence. First, the independent variable threat appraisal consists of both the severity and vulnerability of the situation [46]. Severity is the perceived degree of harm from engaging in unhealthy behavior, the extent to which one will experience or die from contracting COVID-19 upon not following social distancing recommendations. Vulnerability is the perceived probability of threat occurrence, the extent to which one will contract COVID-19 upon not following social distancing recommendations. Both variables were operationalized using three questions adopted from previous studies to assess the severity and vulnerability of the COVID-19 pandemic (see Multimedia Appendix 1 Table A1). The items’ internal consistency was tested using Cronbach alpha (0.63 and 0.76, respectively), and the standardized score was generated for both severity and vulnerability variables.

The second main independent variable was coping appraisal, which consists of both self-efficacy and response efficacy of the situation [46]. Self-efficacy is the perceived belief that one can successfully maintain a safe distance from others when in contact or stay home to avoid all contacts. Response efficacy is the perceived efficacy of adherence that adopting social distancing will be effective in reducing the threat of COVID-19. Both variables were operationalized using three questions adopted from previous studies to assess the self-efficacy and response efficacy of adherence to the COVID-19 pandemic regulations (see Multimedia Appendix 1 Table A1). The items’ internal consistency was tested using Cronbach alpha (0.6 and 0.8, respectively), and the standardized score was generated for both self-efficacy and response efficacy variables.

Lastly, the knowledge variable was coded to reflect the respondents’ overall knowledge of COVID-19, as displayed in Multimedia Appendix 1 Table A1. The mean for knowledge was 15.04, with a minimum of 4 and a maximum of 24. The mean COVID-19 knowledge was highest in Kuwait (mean 16.32, SD 3.05), followed by the United States (mean 15.04, SD 3.62) and South Korea (mean 11.85, SD 3.74). Multimedia Appendix 1 Figure A4 displays the mean knowledge across countries.

To examine threat appraisal and coping appraisal, this study focuses on COVID-19 information sources used and on COVID-19 social media use. COVID-19 information sources was operationalized as the total number of sources used to attain COVID-19 health information. Table 1 displays the mean for the whole sample as 2.26, showing that individuals use two information sources on average to attain COVID-19 information. The mean was highest in Kuwait (mean 2.50, SD 1.13), followed by the United States (mean 2.31, SD 1.34), and South Korea (mean 1.68, SD 1.11). COVID-19 social media was also operationalized as the total number of social media platforms used to attain COVID-19 health information. Table 1 displays the mean for the whole sample as 1.16, showing that individuals use one social media platform to attain COVID-19 information. The mean was highest in Kuwait (mean 2.10, SD 1.38), followed by South Korea (mean 1.48, SD 1.36) and the United States (mean 1.29, SD 1.41). COVID-19 social media was further examined by categorizing the platforms to social network platforms (Facebook and LinkedIn), media sharing platforms (Instagram, Snapchat, TikTok, and YouTube), and texting and microblogging platforms (Twitter and Whatsapp). Details of the analysis are shown in Multimedia Appendix 1 Table A5.

In addition to these key variables of interest, several control variables such as age, gender, and household income were included to account for counterfactual explanations relevant to our models (see Multimedia Appendix 1 Table A1 for details).

Following PMT, the empirical model specifies how individuals express their opinion toward adherence of social distancing guidance by the government through threat appraisal, coping appraisal, and knowledge of COVID-19. Furthermore, threat appraisal and coping appraisal were specified through COVID-19 sources of information (COVID-19 information sources and COVID-19 social media). A set of control variables to enhance our empirical model’s robustness included demographics characteristics of the survey participants, such as gender, age, and household income. The formal specification of the general model is as follows:

Threat appraisal model: Threat appraisal_i = β₀ + β₁ × COVID-19 information sources_i + β₂ × COVID-19 social media_i + β₃ × Country + Controls_i + ɛi   (1)

Coping appraisal model: Coping appraisal_i = β₀ + β₁ × COVID-19 information sources_i + β₂ × COVID-19 social media_i + β₃ × Country + Controls_i + ɛi   (2)

Adherence model: Adherence_i = β₀ + β₁ × Threat appraisal_hat_i + β₂ × Coping appraisal_hat_i + β₃ × Knowledge_i + β₄ × Country + Controls_i + ɛi   (3)

Where controls includes gender, age groups, and household income. The country dummy was included in the full sample model but was removed for subsample analyses.

Seemingly unrelated regression (SUR) was used to estimate the β coefficients of the key parameters and employ robust standard errors to test the models. are the disturbances associated with each observation. SUR was used to estimate to what extent our set of key variables influence adherence. The adherence model (Equation 3) uses the predicted values of threat appraisal (threat appraisal_hat) and coping appraisal (coping appraisal_hat) from the first stage models (Equations 1 and 2).

In general, this study found that coping appraisal, threat appraisal, and knowledge positively influence adherence. Furthermore, using various COVID-19 information sources influences coping appraisal, and using social media for COVID-19 information influences threat appraisal. Tables 6 and 7 summarize the findings from this study (Multimedia Appendix 1 Table A6 displays the summary of findings on the specific PMT constructs and the social media platform constructs). In this section, we elaborate on the findings to provide useful policy and managerial insights.

First, the study found that using more information sources positively influences the coping appraisal associated with the COVID-19 pandemic situation. Undoubtedly, this finding’s importance relates to the increasing use and trade-off of information sources by the public. As much as people like to use many information sources, whether it is beneficial or not remains a question. Furthermore, citizens may be swayed by popular information sources such as the internet, TV, and newspapers. This finding highlights the benefits of using multiple information sources during the COVID-19 pandemic. This could be due to the high amount of misinformation and that the pandemic is ever-evolving. Therefore, attaining multiple information sources allows individuals to obtain more accurate information that helps with their coping appraisal. This finding is in line with a recent COVID-19 study that displayed an association between more sources of information and higher self-confidence to cope with COVID-19 in health care workers, yet the direction of the association was not confirmed in their study [22]. Thus, this study sheds light on this literature and displays the importance of multiple sources of information to cope with the COVID-19 pandemic [48]. The study also found that information sources are more influential on the coping appraisal in South Korea than Kuwait, displaying that individuals in South Korea are more rationally using information sources to cope than individuals in Kuwait.

Second, this study found that using social media to get COVID-19 information positively influences threat appraisal. This is an important finding to validate the increasing trustworthiness of social media on people’s decisions and the role of social media in this pandemic. Social media is one of the main channels used to provide updated COVID-19 information [49]. This study’s findings are consistent with previous studies [50] that showed that the frequency of social media was associated with high odds of anxiety [51]. This association might be because, throughout the COVID-19 pandemic, many false reports and misinformation bombarded social media, which resulted in a lot of confusion and anxiety [50]. Furthermore, many individuals use social media to express their feelings of worry, anxiety, nervousness, and fear, which is contagious in social networks [52]. As we stated earlier, the social media role in people’s minds during the pandemic is not free from harmful aspects such as increased anxiety, sleep disturbance, suicidal thoughts[19], misinformation, and distress [20,21]. Thus, it is interesting to note the positive aspects such as social media’s role in adherence decisions. As for country comparisons, we found that social media is more influential on threat appraisal in the United States than in South Korea and Kuwait. A possible reason for this finding is that social media information in the United States might be more fearful or driven with more riddles (ie, whether to trust Twitter postings or not) compared to South Korea and Kuwait.

Third, threat appraisal positively influences social distancing adherence. One of the primary emotional responses during a pandemic is fear [53]. This fear is a defensive system to combat ecological threats [54]. A meta-analysis found that targeting fears can be useful in situations, such that appealing to fear leads people to change their behavior if they feel capable of dealing with the threat but leads to defensive reactions when they feel helpless to act [55]. This study sheds light on this stream of research by displaying that fear or threat appraisal was useful in the COVID-19 pandemic since it changes an individual’s behavior and influences their social distancing adherence.

Based on PMT, a fear appeal is a cognitive assessment that prepares individuals against the severity of a threatening event. People consider the pros and cons, and the probability of the event occurring to develop a response. The recommended response in the COVID-19 context is adherence. The cognitive assessment enhances the fear appeal and subsequent motivation to protect oneself—without which the recommended action would not be effective across citizens. Besides, there is a more substantial threat appraisal for social distance adherence in the United States than in South Korea and Kuwait. This follows through with the social media country comparison finding; in addition, the finding shows that threat appraisal in the United States is more rational than that of South Korea and Kuwait.

Fourth, coping appraisal positively influences social distancing adherence, and this effect is more influential in the United States than Kuwait, again displaying that individuals in the United States are more pragmatic in the coping and adherence process. One of the main factors that have been stressed upon during the COVID-19 pandemic is how to cope with the current situation [15]. Studies have shown that problem-focused coping is associated with better adherence to health-related behaviors and a higher sense of control [56,57]. The findings support these perspectives and suggest, as per PMT, that the preventive actions will be preferred in a high threat situation when both the self-efficacy and the efficacy of the adherence plans are high. This finding agrees with this research stream and displays the importance of coping appraisal during the COVID-19 pandemic to adhere to social distancing recommendations.

These findings partially explain the variations observed in the adherence outcomes, which can be explained due to the variations observed in the threat and coping appraisals. Finally, as an additional outcome, we found that knowledge positively influences social distancing adherence in the whole sample, the United States, and Kuwait. There were no comparative difference in results across countries. This sheds light on the importance of correct knowledge during this pandemic to be able to adhere to social distancing recommendations. Association between adherence and better knowledge of the disease is consistent with recent findings and provides evidence for using proper interventions of proper communication related to the disease on improving adherence [56,58].

A set of implications and recommendations for public health officials and policy makers can be drawn from this study. This study points to the gap in the responsible behavior of individuals regarding adherence. The current government efforts to mitigate and the expectation that everything should be back to normal have differing consequences. Existing work suggests that policy recommendations’ efficacy and outcomes depend on the individuals’ beliefs and subsequent actions [59,60]. The first step in this process is the firm belief of whether the recommended action will mitigate the threat or manage the fearful situation.

Thus, the key to minimizing rejection and maximizing acceptance of recommendation is reliant on the messages and information provided to citizens. Often, the media outlets and information sources are left to craft their messages independently or freely without the policy makers’ recommended guidelines. There needs to be policy guidelines with careful attention about what messages to give and how to give them, so that it does not instill too much fear nor allow citizens to be too careless (ie, messages and information need to have a balance to involve appropriate threat and coping appraisals in citizens’ minds).

This study examines factors that influence citizens adhering to social distancing at a point in time. However, the citizen might go back and forth in the adherence process, and the threat appraisal, coping appraisal, knowledge level, and sources of information may change over time. This is a limitation of this study, as the data set used is a cross-sectional survey. Future studies could examine how a citizen’s coping appraisal, threat appraisal, and adherence changes over time.

In addition, using the random sampling process in the United States, Kuwait, and South Korea samples may include fewer familiar respondents to the study context. In particular, the questionnaire was only online. Therefore, respondents were all users of the internet. The study does not examine noninternet users, which could have differential impacts. Thus, the generalization of the sample to a uniform national culture characteristic is a limitation of this study. Future studies could conduct online and offline surveys, and examine the difference in threat and coping appraisal in terms of adherence behavior.

The findings should be taken with care due to the sample being representative of three countries; thus, future studies could expand to more countries to examine even more cultural differences. Another limitation is the small sample size for South Korea, limiting the generalizability of the results.

As the COVID-19 global pandemic continues to grow and governmental restrictions are ongoing, it is critical to understand people’s frustration to reduce panic and promote social distancing to control the pandemic. This study points to a threat and coping appraisal mechanism that may clarify the adherence variations. This study also highlights that social media has an impact on the threat appraisal of the COVID-19 pandemic. Furthermore, the intensity of information sources used to attain COVID-19 information impacts the coping appraisal for the pandemic.