https://orcid.org/0000-0003-4430-0569
Figures
Abstract
The Covid-19 pandemic is a textbook case of significant situational stress induced by various disruptions beyond mere health concerns, such as social isolation and financial constraints. For the workforce, it is essential to anticipate how these disruptions may undermine employees’ resilience, to avoid a negative spiral where poor well-being lowers productivity, reduces economic prospects, and continues to increase worker stress. We measure multiple forms of stress and worries as drivers of well-being—health, economic, social, and psychological—encountered by the workforce during the acute period of the Covid-19 pandemic. The study analyzed data from 2,780 employees across five European countries: France, Germany, Italy, Spain, and Sweden. Overall Concern Score: The overall concern score was 56.8% across four domains: health, economic, social, and psychological. Stressors can be synthesized into five typical groups associated with a variety of mediating factors such as institutional trust, lifestyle, and worker education. The implication is that workers’ well-being is heterogeneous and that human resource practices may need at least a segmented approach to well-being if they wish to create an environment of a resilient and productive workforce.
Citation: Bughin J, Cincera M, Reykowska D, Żyszkiewicz M, Ohme R (2025) The great divide between employees: Clustering employee “well-being” during a pandemic. PLoS ONE 20(3): e0294540. https://doi.org/10.1371/journal.pone.0294540
Editor: Marco Cellini,, CNR: Consiglio Nazionale delle Ricerche, ITALY
Received: November 3, 2023; Accepted: February 4, 2025; Published: March 31, 2025
Copyright: © 2025 Bughin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data underlying the results presented in our manuscript come from a confidential survey conducted by NEUROHM. Due to the nature of the data collection and agreements in place, these data are not publicly available. Researchers who wish to access the data can contact the NEURHOM company at anna.szydlo@neurohm.com (Data officer of the company) and/or Rafal Ohme (CEO of the company) at rafal@neurohm.com. The company will facilitate the process according to the established confidentiality agreements.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
According to the Worldometers database, the Covid-19 pandemic had officially affected more than 700 million people worldwide by May 2024, causing around 7 million deaths. This makes it one of the most challenging pandemics to manage since the Spanish flu at the beginning of the 20th century [1].
The impact of such a pandemic is recognized not only on health but also on economic livelihoods. At the height of the pandemic, the WHO, FAO, and ILO issued a joint statement reflecting the negative fallout in other areas, claiming that “tens of millions of people were at risk of falling into extreme poverty and millions of businesses faced an existential threat” [2] demonstrated a significant peak in small businesses, but beyond small entrepreneurs, a large part of the workforce is also employed in large corporations [3,4], and if bankruptcy became more limited for large firms, the workforce during the crisis was not faring well [5]. Stress particularly increased in most exposed professions, such as medical staff, teachers, or security forces, but it spread to the workforce as a whole, with a 70% rise in burn-out during the peak of the pandemic in the US in 2020 [6,7]. Workforce well-being is strongly associated with increased labor productivity [8]. Failure to promptly address well-being may lead to an economic depression, which feeds a negative spiral of ill-being and poor economic prospects, propelling the health crisis into a lasting economic crisis. Many companies have tried implementing preventive health measures against Covid-19, including health protection measures in the workplace or working from home. However, these health measures may lack scope (only around 40% of workers, and especially managers, could work from home; see [9,10] and may miss out on a range of additional concerns, such as job retention and finances, or the ability to protect close ties [11].
This study is grounded in the Conservation of Resources (COR) theory, developed by [12]. COR theory posits that individuals strive to obtain, retain, and protect their resources, which include objects, personal characteristics, conditions, and energies. When these resources are threatened or lost, individuals experience stress. The pandemic has created a significant threat to various resources such as health, economic stability, and social connections, which are critical for employee well-being.
This article is part of a recent stream of wellness science literature investigating the different dimensions of pandemic-related stress beyond pure health worries. [13] examine the combined role of economic concerns, social distancing, and health anxieties in anticipating the possibility of significant mental health developments from the pandemic. [14] consider economic anxiety to be as big a problem as health anxiety in countries such as the USA, the UK, and Israel.
The study aligns with the concept of well-being, rooted in positive psychology, rather than wellness in the context of medicine. Well-being in positive psychology focuses on the holistic experience of life satisfaction and the balance of positive and negative emotions [15]. It encompasses various domains such as emotional, social, and psychological well-being [16]. This framework is appropriate for understanding the diverse impacts of the Covid-19 pandemic on employees, considering their mental and emotional health, social interactions, and economic stability.
The primary research problem addressed in this study is understanding how the multifaceted disruptions caused by the Covid-19 pandemic impact the well-being of employees, and how these impacts can be segmented to inform targeted human resource practices. The objectives of this research are first, to measure the various forms of stress and worries—health, economic, social, and psychological—encountered by the workforce during the acute period of the Covid-19 pandemic. Second, to identify and cluster the typical groups of stressors and mediating factors such as institutional trust, lifestyle, and worker education. Third, to analyze the relationship between these clusters and employee socio-demographic characteristics, protective behaviors, and levels of social trust. Fourth, to examine the differences in well-being concerns across five European countries and how these are influenced by each country’s social and economic fabric. And finally, to utilize online response time as an indicator of attitude strength to adjust survey responses and reinforce the statistical significance of the findings.
To address these objectives several research questions are posed. What are the predominant types of stress and worries affecting employees during the Covid-19 pandemic? How can these stressors be clustered into distinct groups? What socio-demographic characteristics, behaviors, and levels of trust are associated with each cluster of stressors? How do these clusters and their associated characteristics vary across different European countries? And how does response time adjustment affect the reliability of the survey responses regarding employee worries?
The contribution of this research is fivefold. Firstly, to our knowledge, the research is one of the first to study a wide range of concerns perceived by the workforce. The risks assessed include occupational and financial risks (micro-economic risks), risks related to the satisfaction of basic needs (supply chain), violence, and psychological risks (social risks), and the country’s finances (macro-economic risks), in addition to health-related risks. The first finding is that health accounts for just over 40% of all risks expressed by workers.
Secondly, worries can be clustered. Five polarized segments of the working population seem to prevail regarding perceived worries encountered. An important study by [17] also uses clustering techniques to profile employees’ well-being in the UK and France. They also find about five classes of employees. The major difference is that we extend the analysis to five European countries and look at a broader class of worries. Finally, instead of using latent class analysis, we use k-means clustering techniques to allocate workers among clusters.
Thirdly, we analyze and find that risk clusters are linked to employee socio-demographic characteristics, certain types of protective behaviors, and levels of social trust. Notably, a lack of trust in government actions stimulates economic and financial worries, while social trust limits many worries.
Fourth, the analysis is carried out and confirmed for five European countries, also recognizing that countries’ underlying social and economic fabrics influence concerns [18]. In particular, our analysis examines possible differences in social foundations along the lines of the work by [19]. Those country differences should and are confirmed to affect the relative size of clusters but not the clusters themselves by countries.
Finally, as the analysis is based on respondents’ self-reports, we use online response time to adjust survey responses based on the neuroeconomics principle that response time indicates attitude strength [20]. By correcting this response time, we refocus the responses towards a neutral response. This procedure thus reinforces our statistically significant results. This response adjustment procedure has also been used, among others, by [21]) in the context of assessing worries during covid-19, but with a focus on economic worries only.
Materials and methods
This research is part of a vast multinational project on Covid-19, aimed at understanding people’s concerns and behaviors in the face of the pandemic. The focus is on the working population, which typically accounts for over 50% of all citizens and is a crucial driver of economic growth. We concentrated our efforts on the peak of the pandemic in April 2020 in European countries.
Sample scope
Five countries were analyzed: France, Germany, Italy, Spain, and Sweden. These countries, the largest in Europe, represent different socio-economic models [19]. Additionally, these countries had varied policy responses to the Covid-19 crisis, with Sweden opting not to impose containment measures unlike the strict policies in place in Italy, for example. Researchers who wish to access the data can contact the NEURHOM company at at anna.szydlo@neurohm.com (Data officer of the company) and/or Rafal Ohme (CEO of the company) at rafal@neurohm.com. The company will facilitate the process according to the established confidentiality agreements. In addition, since this study involved an anonymous survey, and no personally identifiable information was collected, no ethical approval was required.
Data was collected online, based on representative country samples for age (over 18) and gender, and recruited via a panel agency in April 2020. The total sample comprised over 5,000 responses, or a minimum of 1,000 per country. Focusing on employees alone, the total sample consisted of just over 2,780 employees in five countries (Tables 1 and 2), or 55% of the sample, aligning with employee participation in the 18+ population of these countries.
Sample collection
Respondents were invited by email and informed about the scope of the study. Their task was to assess whether they agreed with the statements displayed on the screen. To avoid coercing responses or receiving answers that did not reflect actual behavior, each question was structured so that answers could be given on a three-point scale (yes, hard to say, no), with the “hard to say” option allowing respondents the flexibility to avoid forced choices.
We checked for the usual response biases, but the uncertainty between what people report and their actual attitudes/behaviors remains a significant caveat. We measure response time and adjust the data in line with [20], who finds a strong correlation between statements and actual behavior in individuals with fast reaction times when expressing their opinions. Similar to [21], the iCode Smart test was used to collect data [22], with response time (RT) collected for each response. Responses with a latency of less than 500 milliseconds (ms) (suspected of being given randomly) or greater than 10,000 ms (suspected of having been given after distraction) were eliminated. This accounted for less than 1% of doubtful responses.
To account for individual differences in reaction speed, we normalized the reaction time data measured in milliseconds. STDRT is the z-score of log(RT), with mean = 0 and standard deviation = 1. We then created a variable, RTC, that takes into account both the explicit response and the reaction time (RT) required to produce the response, i.e., RTC’ - 1/2 = (1 - a) x (Y - N)/4 (0 < RTC’ < 1) where (1 - a) = max(SDRT, 2)/2 and Y-N is the difference between the share of Yes and No reported. Thus, 0 < a < 1 acts as a factor that reduces the difference between responses as a function of response reaction time, which we call the confidence index. When RTC’ converges towards 50%, it implies that everyone’s answer hovers around “Difficult to say,” or that not all answers are credible due to unusual reaction times. The more extreme the value of RTC’, the firmer the position taken by the survey response on the qualification of the statement asked for in the survey. Thus, RTC = 0 corresponds to a solid and dominant NO, and RTC = 1 to an overwhelming YES.
High-level statistics
Worries.
We consider worries as a natural precursor of well-being, as argued by [23]. We examine four types of concern: health (H), economic (E), social (S), and psychological (P). These concerns align with other studies that extend beyond purely health issues related to the pandemic [13,14]. Each worry domain is based on four dimensions, representing different intensities or scopes, e.g., the health concept concerns the self, close ties (children and elderly members), and weak ties; social worries affect family tensions, escalating to domestic violence or divorce. Table 3 shows the RTC value and the confidence index of the responses associated with each risk or concern, ranked from most important to least important, for the entire sample and for the 16 dimensions that comprise H, E, S, and P.
Adding up all the values, the total is 9.1 out of 16, a value of 56.8%. This indicates that most concerns prevail in the employee population during the first wave of the Covid-19 pandemic. Furthermore, H = 61.2%, S = 59.5%, P = 56.0% and E = 50.7%, meaning that each factor is predominantly present in the employee population. Finally, H has the highest value but represents only 40% of the total concerns expressed. The main H concern relates to people at increased risk of death, such as elderly family members, consistent with the findings of other publications, such as [24].
The economic dimensions give rise to different concerns. Supply chain risk (measured by the availability of needed goods) is the least important concern, while the macroeconomic risk of a cash-strapped country is a broader concern. Psychological/mental and social risks are also significant, and domestic violence and divorce rates are risky for the employees sampled.
As our sample selects only employees, we can also compare the magnitude of concern with that of non-salaried employees, e.g., retirees or people of working age who are not working. In this case, the total for retirees is 7.8 (15% lower than for salaried employees), while 8.7 (6% lower than for salaried employees) for the other non-working population. In other words, the salaried population expresses a broader range of concerns than the non-salaried. The weaker expression of worries is explained in particular by economic considerations, with the most significant difference between retirees and salaried employees being by far the risk of job loss [25]. Not surprisingly, retirees are more concerned about their health and less about social risks (as they tend to be more autonomous).
Fig 1 also shows the distribution of risk expression among employees is far from uniform. We calculate that 20% of employees express less than 50% of the type of H, S, P, E risks studied, and 20% of others mention at least 81% during the Covid-19 pandemic. In addition, 18% of employees account for 90% of all concerns or are more concentrated than a typical Pareto distribution.
This type of “power law” is observed in many phenomena. In economics, power laws can be seen in income and wealth, city and company size, stock market returns, or international trade [26]. The power law has also recently been studied as the mechanism underlying the spread of the covid-19 pandemic around the world [27]. The power law has also been studied in the context of psychological disorders, e.g., [28], but this is the first time, to our knowledge, that it is examined in the context of general well-being affected by a significant disturbance such as a pandemic.
In particular, our study focuses on two essential aspects of this apparent power law: how the distribution is determined by the variety and intensity of worries and how perceptions/attitudes correlate with worries.
Contextual factors.
Worries are known to be correlated with sociodemographics and lifestyle, for instance, older people tend to worry more about health issues given the high risk linked to contamination, etc. [29]. We also have collected responses related to various attitudes/behaviors and trusts likely to correlate employees’ expression of risk. As we had a long list of statements, we first applied categorical principal component analysis (CATPCA) to reduce the information. CATPCA was performed using Varimax rotation with Kaiser normalization to maximize the sum variance of the factor coefficients.
Ten factors were derived, representing 69.6% of the total variance. Table 4 presents the ten factors and associated dimensions in the order in which they emerged from the data rotation. Table 5 shows the RTC’ and confidence values, ranked in order of importance of the factors.
Three factors (1, 4, 6) are linked to third-party trust. The first relates to government institutions, the second to healthcare, and the last to how people react to the Covid-19 crisis. Factors 2 and 5 relate to precautionary measures (NPI), with factor 2 encompassing the most critical ones in disease control [30]. Factors 8, 9, and 10 are all linked to the perception of a lasting danger from the virus. Factor 10 is linked to the duration of the crisis, and factors 8 and 9 to vulnerability to the virus and the priority given to health over wealth. Finally, factors 3 and 7 are more closely linked to social support for the individual and the family.
Grouped by themes, Table 4 shows good compliance with the NPI (mean = 62%), even after correcting for probably overestimated answers. The confidence level is the lowest of all themes. This overestimation could be explained by the fact that responses seem to obey the public mandate of quarantine and social distancing measures to limit the spread of pandemics. Nevertheless, as many studies have shown, RTC’ < 100%, i.e., NPI compliance, is incomplete [31]. Third-party trust is relatively well recognized (58%), but people feel very vulnerable (56%). The impact on lifestyle is felt to be more minor (48%). Expectations for the duration of the crisis are that it could be more short-term than long-term, so it seems that most European employees were not necessarily expecting the current second wave.
Trust is significantly higher in healthcare than in government. Nevertheless, public authorities should be trusted to manage the crisis so that citizens adopt the recommended protective actions [18,32]. Finally, vulnerability is more or less perceived by the employee population, and generally speaking, there is a majority to think the crisis will last.
Methodology.
We use a two-step process to analyze the distribution of worries and mediation factors. First, we perform clustering analysis to determine worry segments, then launch logit regression analysis to correlate factors associated with segments.
We prefer a two-step process instead of a complete clustering of all variables in one step, as mediation factors are not at the same level as worry expression. Further, the second step allows us to statistically assess the strengths of mediation factors, while direct mix clustering would prevent this information.
Regarding clustering, we use K-means clustering around the 16 items that present the four areas of concern H, S, P, E, to divide the population into coherent and stable segments. The study by [33] segments risk attitudes via K-means for Bangladesh. However, the study only involves 340 people online and given the country’s current digital development, is not representative of the population. Finally, drivers of group membership factors are not tested, as they are in this study.
The K-means technique minimizes the sum of the squared distances between each possible risk cluster and its centroid and allocates each employee to one cluster. Alternative clustering solutions exist, of which latent class analysis can be possible, as done, e.g., in [17]. The strength of the K-means is that it always ensures that the most similar observations are in the same clusters and that each observation is univocally allocated to one segment versus latent class analysis. Extra analyses with different cluster solutions were carried out, but the 5-cluster solution appears to be the most informative.
Results and discussion
Table 6 shows segment size overall, then by country. We can see that the size of a segment depends on the country. Segment 5 is the largest in Sweden and Germany but the smallest in Italy and Spain. As we shall see later, segment five is made up of the least worried employees, unlike the first segment. The fact that Italy and Spain have such large worried workforces can be explained by the fact that these countries have suffered relatively high contamination, under-capacity in healthcare, and largely imposed containment. This contrasts with Sweden, where no containment was applied, or Germany, where healthcare capacity is relatively high and contamination was less widespread than in southern Europe.
Table 7 shows the mean values per segment for the 16 risks analyzed, while Fig 2 groups the risk profiles according to the four domains H, P, S, E. We confirm that the segments concerned vary in intensity and variety of the risks expressed.
For simplicity, we name the cluster rank according to the decreasing intensity of the level of concern expression - i.e., the greatest for the first segment.
Segment 5 is the only segment whose average RTC’ is < 50% (Fig 3), except that segment 5 perceives social risk similarly to the average of the other segments. Its primary concern is social and is linked to domestic violence and divorce rates.
Table 8 describes compliance with essential non-pharmaceutical interventions (NPIs) as a check to cluster validity. We see that greater use of NPIs is linked to more anxious segments [34–36]. In particular, segment 5 is less compliant with any measure, as it suffers from the lowest level of risk. Segments 1 and 2 prefer to be quarantined or choose to have as few interactions as possible, as they bear the most considerable weight of risk. Segment 2 conforms best to midlife, as it suffers relatively less from loneliness. Segment 3 is relatively unconcerned about its health and more concerned about preserving its job, so it conforms more to social distancing than midlife.
In addition to NPI compliance being a good marker of segment concerns, we now formally test the existence of multivariate segmentation markers using a formal logit statistical model. We include socio-economic factors, as they are well known to have an impact on attitudes and the expression of concerns (e.g., [24,37], and to condition the ability to work remotely [10]).
Each segment logit regression against Segment 1 is presented in Appendix 2. At the same time, Table 9 summarizes the results, showing only those markers that are statistically significant at the 10% statistical threshold and omitting the country effect. For the sake of simplicity, we have also grouped the factors into four broad categories (confidence, NPI compliance, vulnerability and lifestyle). A negative sign means a weaker impact on the probability of belonging to a segment.
Socio-demographic data play a specific role in predicting group membership, but not a major one. Neither family composition nor gender has a statistically significant impact.
A low income (less than 2,000 euros per month) reduces the probability of belonging to segment 5. One of the reasons, already highlighted in the introduction to this study, is that low income is often associated with essential on-site work, exposing people to health risks and vulnerability to the virus. Living in the countryside (in towns with fewer than 100,000 inhabitants) makes an employee less likely to belong to segment 3. The level of education affects the probability of belonging to different segments.
The markers of trust, vulnerability and lifestyle play a significant role in the distribution of employees between the different groups. Perceived vulnerability is an essential discriminating factor in all segments, favoring health risk. Compliance with the NPI (and increased caution) are behaviors that emerge from health risk, but we find that they play a role in addition to vulnerability. One reason for this is that the NPI was imposed as a response to a government mandate and reflects compliance with authoritative measures.
Trust is also essential. Segments 1 and 3 are less inclined to accept their government’s actions to combat the Covid-19 pandemic. Segment 1 is also a segment that trusts its peers more than institutions, for example.
Using the exponential of the point estimates in Table 9, Table 10 calculates the marginal probability impact for the four marker categories. Consider an employee among others with low institutional trust, who also complies with the NPI and feels vulnerable to the virus, who belongs to segment 1. Its counterpart belongs to segment 5. These two segments are also the most and least fragile among employees. An employee unafraid of the virus trusts the healthcare system, respects the NPI while maintaining a healthy lifestyle, and most likely belongs to segment 4.
While segment 5 is less “fragile”, it may represent a risk for other segments, due to low NPI compliance. Similarly, human resources need to adopt a fine-tuned approach to support employees effectively: Segment 4 may agree with the health situation related to Covid, but is more stressed regarding employment. Segment 1 and, to a lesser extent, segment 2 are rather stressed and have physical and mental health issues to resolve if they are to remain productive.
A final remark concerns the differences between countries. Table 11 illustrates the impact of markers on the first segment in the five countries concerned. As with the total, greater compliance with the NPI, greater perception of Covid-19’s vulnerability, strong social fabric or lifestyle maintenance are the norm in all countries.
Spanish employees are more sensitive to trust in people than in government, while Swedish employees are better predicted by their NPI compliance level and social fabric. Regarding social priorities, German employees are more sensitive to health than wealth. Again, these tendencies can be attributed to culture and background, and Sweden has a robust social culture compared to other countries [19] and has not imposed confinement. Swedish employees’ compliance with the NPI is likely to be a more evident discriminatory behavior than in countries where the NPI has been set.
Conclusions
In the context of a major viral pandemic, employee well-being should be a central concern. Employees can face a significant risk of contagion while working on-site and represent one of the primary resources of an economy already strained by the major disruptions induced by the pandemic.
We find that well-being can be severely affected, as employees express a wide variety of concerns beyond physical health, including mental stress and job preservation concerns which, if left unaddressed, can seriously undermine productivity. These concerns may persist even if people telework; other stresses, such as domestic violence, may emerge.
More importantly, we show that workforce fragility is not evenly distributed and exhibits a kind of power curve whereby nearly 20% of employees are responsible for 90% of the variety and intensity of worries. We find that five well-defined segments can be identified using clustering techniques to represent this power law. In addition, we can associate a set of critical markers that predict employee worries. These markers may be particularly useful, for example, for humane enterprises to engage in appropriate selective dialogue with their employees and improve their well-being during this pandemic. This is not just a matter of corporate responsibility but an effective strategy for maintaining high productivity and ensuring business resilience.
This study, while comprehensive, has several limitations. Firstly, the data collection was conducted online and is based on self-reports, which may introduce response biases. Participants might underreport or overreport their stress levels and concerns due to social desirability or other psychological factors. Additionally, the study relies on respondents’ ability to accurately recall and assess their experiences during the pandemic, which may be influenced by memory biases.
Secondly, the study focuses on a specific period during the acute phase of the Covid-19 pandemic. As the situation evolves, the stressors and concerns of employees may change, potentially altering the relevance and accuracy of the identified clusters and their characteristics. Longitudinal studies are necessary to capture these dynamic changes and provide a more comprehensive understanding of employee well-being over time.
Thirdly, the sample is limited to five European countries, which, while diverse, may not fully represent the global workforce’s experiences. Cultural, economic, and social differences across regions could influence the generalizability of the findings. Future research should include a more varied international sample to validate and expand upon these results.
To address these limitations, several remedies can be considered. To mitigate response biases, future studies could incorporate mixed methods, including qualitative interviews and focus groups, to complement quantitative surveys and provide a deeper understanding of employee experiences. Additionally, employing objective measures of stress, such as physiological indicators, could help validate self-reported data.
Longitudinal studies should be conducted to track changes in employee well-being over time, capturing the evolving nature of stressors and concerns as the pandemic progresses or resolves. This approach will provide a more nuanced understanding of how employees adapt and what interventions are most effective at different stages.
Expanding the sample to include a broader range of countries and regions will enhance the generalizability of the findings. Cross-cultural comparisons can reveal how different socio-economic contexts and government responses impact employee well-being, offering insights that are globally applicable.
References
- 1.
Bughin J. Ten moments of truth in the COVID-19 crisis. Policy Punchline. Princeton; 2020.
- 2. Bartik AW, Bertrand M, Cullen Z, Glaeser EL, Luca M, Stanton C. The impact of COVID-19 on small business outcomes and expectations. Proc Natl Acad Sci U S A. 2020;117(30):17656–66. pmid:32651281
- 3. VanderWeele TJ. Challenges estimating total lives lost in COVID-19 decisions: consideration of mortality related to unemployment, social isolation, and depression. JAMA. 2020;324(5):445–6. pmid:32639547
- 4. Carnevale JB, Hatak I. Employee adjustment and well-being in the era of COVID-19: Implications for human resource management. J Bus Res. 2020;116:183–7. pmid:32501303
- 5. Pfefferbaum B, North CS. Mental health and the Covid-19 pandemic. N Engl J Med. 2020;383(6):510–2. pmid:32283003
- 6. Sokal L, Trudel LE, Babb J. Canadian teachers’ attitudes toward change, efficacy, and burnout during the COVID-19 pandemic. Int J Educ Res Open. 2020;1:100016.
- 7. Dinibutun SR. Factors associated with burnout among physicians: an evaluation during a period of COVID-19 pandemic. J Healthc Leadersh. 2020;12:85–94. pmid:32982532
- 8. DiMaria CH, Peroni C, Sarracino F. Happiness matters: productivity gains from subjective well-being. J Happiness Stud. 2019;21(1):139–60.
- 9. Dingel J, Neiman B. How many jobs can be done from home? Covid Economics: Vetted and Real-Time Papers. 2020;1(2020):1–28.
- 10. Sostero M, Milasi S, Hurley J, Fernandez-Macías E, Bisello M. Teleworkability and the COVID-19 crisis: a new digital divide? JRC Working Paper Series on Work, Education and Technology. 2020.
- 11. Vagni M, Maiorano T, Giostra V, Pajardi D. Hardiness, stress and secondary trauma in italian healthcare and emergency workers during the COVID-19 pandemic. Sustainability. 2020;12(14):5592.
- 12. Hobfoll SE. Conservation of resources. A new attempt at conceptualizing stress. Am Psychol. 1989;44(3):513–24. pmid:2648906
- 13. Kämpfen F, Kohler IV, Ciancio A, Bruine de Bruin W, Maurer J, Kohler H-P. Predictors of mental health during the Covid-19 pandemic in the US: Role of economic concerns, health worries and social distancing. PLoS One. 2020;15(11):e0241895. pmid:33175894
- 14. Bareket-Bojmel L, Shahar G, Margalit M. COVID-19-related economic anxiety is as high as health anxiety: findings from the USA, the UK, and Israel. Int J Cogn Ther. 2021;14(3):566–74. pmid:32837674
- 15.
Seligman MEP. Flourish: A visionary new understanding of happiness and well-being. New York: Free Press; 2011.
- 16. Keyes CLM. The mental health continuum: from languishing to flourishing in life. J Health Soc Behav. 2002;43(2):207–22. pmid:12096700
- 17. Harju LK, Rokka J, Lopes MM, Airoldi M, Raïes K. Employee well-being profiles during COVID-19 lockdown: a latent profile analysis of French and UK employees. Front Psychol. 2021;12:645300. pmid:34177699
- 18.
Li Y, Wang W, Wang J. Government intervention behavior and optimization strategy of major epidemic control: based on game theory and practice of China’s H7. 2018.
- 19.
Esping-Andersen G. Social foundations of postindustrial economies. Oxford: Oxford University Press; 1999.
- 20. Fazio R. The role of attitude accessibility in the attitude-to-behavior process. J Consum Res. 1989;16(3):280–8.
- 21.
Bou-Hamad FN, Vaccination and societal behavior under COVID-19. 2022.
- 22.
Ohme R, Matukin M, Wicher P. Merging explicit declarations with implicit response time to better predict behaviour. In: Chkoniya V, Madsen AO, Bukhrashvili P, editors. Anthropological approaches to understanding consumption patterns and consumer behaviour. IGI Global; 2020. p. 427–48.
- 23.
Boehnke et al. Social stress theory application. 2001.
- 24. Dryhurst S, Schneider C, Kerr J, Freeman A, Recchia G, Van Der Bles A, et al. Risk perceptions of COVID-19 around the world. J Risk Res. 2020;24(1):1–13.
- 25. Coibion O, Gorodnichenko Y, Weber M. The cost of the Covid-19 crisis: lockdowns, macroeconomic expectations, and consumer spending. Covid Econ. 2020;20:1–51.
- 26. Gabaix X. Power laws in economics and finance. Annu Rev Econ. 2009;1(1):255–94.
- 27. Komarova NL, Schang LM, Wodarz D. Patterns of the COVID-19 pandemic spread around the world: exponential versus power laws. J R Soc Interface. 2020;17(170):20200518. pmid:32993434
- 28. Katerndahl D. Momentary emotions and coping with chronic illness. Arch Fam Med. 2009;18(6):555–60.
- 29. Fornili M, Petri D, Berrocal C, Fiorentino G, Ricceri F, Macciotta A, et al. Psychological distress in the academic population and its association with socio-demographic and lifestyle characteristics during COVID-19 pandemic lockdown: Results from a large multicenter Italian study. PLoS One. 2021;16(3):e0248370. pmid:33690695
- 30. Bo Y, Guo C, Lin C, Zeng Y, Li HB, Zhang Y, et al. Effectiveness of non-pharmaceutical interventions on COVID-19 transmission in 190 countries from 23 January to 13 April 2020. Int J Infect Dis. 2021;102:247–53. pmid:33129965
- 31. Zickfeld JH, Schubert TW, Herting AK, Grahe J, Faasse K. Correlates of health-protective behavior during the initial days of the COVID-19 outbreak in Norway. Front Psychol. 2020;11:564083. pmid:33123045
- 32. Bughin J, Cincera M, Peters K, Reykowska D, Żyszkiewicz M, Ohme R. Make it or break it: On-time vaccination intent at the time of Covid-19. Vaccine. 2023;41(12):2063–72. pmid:36803893
- 33. Bodrud-Doza M, Shammi M, Bahlman L, Islam ARMT, Rahman MM. Psychosocial and Socio-Economic Crisis in Bangladesh Due to COVID-19 Pandemic. Int J Infect Dis. 2020;8:341. pmid:32676492
- 34. Bughin J, Cincera M, Kiepfer E, Reykowska D, Philippi F, Żyszkiewicz M, et al. Vaccination or NPI? A conjoint analysis of German citizens’ preferences in the context of the COVID-19 pandemic. Eur J Health Econ. 2023;24(1):39–52. pmid:35467175
- 35. Harper C, Satchell L, Fido D, Latzman R. Functional fear predicts public health compliance in the COVID-19 pandemic. Int J Ment Health Addict. 2020;27(1):1–14.
- 36. Lambert H, Gupte J, Fletcher H, Hammond L, Lowe N, Pelling M, et al. COVID-19 as a global challenge: towards an inclusive and sustainable future. Lancet Planet Health. 2020;4(8):e312–4. pmid:32702296
- 37. Papageorge N, Zahn M, Belot M, van den Broek-Altenburg E, Choi S, Jamison J, et al. Socio-demographic factors associated with self-protecting behaviour during the Covid-19 pandemic. Covid Econ. 2020;40:70–91.