Imagine a doctor, meticulously cleansing their hands before tending to patients in a hospital. This rigorous handwashing is a fundamental component of safety protocols adhered to by healthcare professionals. Yet, when we extend this scenario to include actions such as dedicating additional personal time to a thorough review of medical records, it unveils a more profound motivation that transcends mere compliance.

This paper closely examines compliance with COVID-19 preventive regulations, such as handwashing and social distancing, as an illustrative case study to distinguish compliance from prosociality. Our investigation delves into how different facets of empathy relate to behaviors benefiting others (prosociality) and/or oneself (compliance). By unraveling the driving factors that lead individuals to invest extra time, effort, and care for the benefit of others, we aim to gain a deeper understanding of the motivational forces at play behind these behaviors.

Empathy during a pandemic

During the pandemic outbreak, citizens worldwide were asked to comply with preventive behaviors aimed at mitigating the spread of the COVID-19 virus, such as keeping physical distance and covering their faces with masks. These behaviors came at considerable personal cost (e.g., reducing social contact, altering one’s appearance and breathing; see Witteveen & Velthorst, 2020) but have a clear prosocial outcome of protecting others from potential infection (e.g., Brooks & Butler, 2021). To encourage these behaviors, decision-makers were seeking interventions and nudges aimed toward increasing prosociality. According to the Empathy-Altruism Hypothesis (Batson et al., 2015), when individuals feel compassion for the suffering of others, they are more likely to engage in prosocial actions. Leveraging the idea that empathy is a driving force to engage in prosocial behavior (Batson, 2019), several studies provided preliminary evidence that empathy can motivate adherence to preventive behavioral measures. For example, in one study, Karnaze et al. (2022) reported that people with more concern for others were more likely to keep social distance and wear masks in public. In another study, Pfattheicher et al. (2020) demonstrated that experimentally inducing empathy for those most vulnerable to the virus promoted motivation to adhere to physical distancing and covering faces with masks. Other studies, however, did not find a consistent relation between empathy and adherence (e.g., Favero & Pedersen, 2020). These conflicting findings cast doubt on whether interventions aiming to increase empathy can effectively promote public health behaviors to mitigate future viruses. To clarify this ambiguity, the current research adopted a more nuanced measurement of different facets of empathy and tested their relations to several different COVID-related behaviors.

Different facets of empathy and their relation with COVID-related behaviors

Empathy is a complex and multifaceted concept, with different definitions and measures in the literature (e.g., Batson, 2009; Hall & Schwartz, 2018; Israelashvili & Karniol, 2018; Reniers et al., 2011; Weisz & Cikara, 2021). In light of these diverse perspectives, we have adopted a recent and comprehensive definition of empathy, considering it as an affective response to perceiving and understanding another’s emotional state (Cuff et al., 2016). Acknowledging the importance of diverse empathic reactions, our research focused on two key empathic reactions - empathic concern and personal distress (Davis, 1983). Empathic concern refers to feelings of compassion and care for the suffering of others, while Personal distress, refers to the discomfort one can experience in reaction to another’s misfortune. Empathy-invoking situations can elicit empathic concern, personal distress, or both reactions (Davis, 1987). Yet, existing research on the relationship between empathy and adherence to preventive behaviors measured empathic concern alone while leaving the other facet of empathy-related response of personal distress overlooked.

This gap is problematic, as the Empathy-Altruism Hypothesis (Batson et al., 1987; Batson et al., 2015), suggests that when individuals feel personal distress, they are generally less likely to engage in prosocial actions. Indeed, previous lab studies have documented that empathic concern is associated with increased altruistic motivation to help others, whereas personal distress predicts reduced altruistic motivation to help others (Batson, 2019; Batson et al., 1987; Batson & Oleson, 1991). However, Batson’s nuanced perspective (for review see Batson 2019) suggests that personal distress might still be positively associated with prosocial outcomes when helping the other person is beneficial for the self. Individuals with high personal distress may be more inclined to engage in actions that alleviate their own discomfort, such as maintaining personal hygiene or keeping physical distance, rather than behaviors solely for the benefit of others, such as providing physical or financial assistance for others. This distinction highlights the complexity of empathic reactions and their varying implications for adherence and prosocial behaviors, depending on the motivations (concern, distress) involved.

The present research

The goal of the current study was to analyze different facets of empathy (concern, distress) and examine their respective relationships with COVID-19-related behaviors. Data were collected as part of an international collaborative survey addressing various topics related to COVID-19 beliefs and behaviors (see Van Bavel et al., 2022). Here, our focus is on the association between COVID-related behaviors assessed in the survey and specific questions related to empathy, included exclusively in the Israeli sample to address the current research question. The choice of an Israeli sample for our study stems from the unique context of the COVID-19 pandemic in Israel, characterized by strict lockdowns involving school closures, business shutdowns, and movement restrictions that began one month before we conducted the study. Specifically, the study was concluded in late April 2020, a few weeks after people started to die from the virus in Israel and approximately eight months before the national availability of mass vaccinations began in December 2020. This factor created a highly charged environment where empathy could play a pivotal role and made Israel an ideal and accessible setting. Our aim was to investigate how individuals’ emotional responses in this emotionally charged context related to their compliance with pandemic measures and their willingness to assist those affected by the pandemic and the associated restrictions.

To capture a prosocial act during the pandemic, the study included a donation behavior. Donating money to charity organizations during COVID-19 is an altruistic act aimed at alleviating the burden of the pandemic. Specifically, COVID-19 altruism was measured through charity generosity (Sjåstad, 2019). To gauge affective empathy facets, participants were asked to report their feelings toward a person who tested positive for the COVID-19 virus, using a list of empathy-related emotion words previously employed to measure Empathic Concern and Personal Distress (see Israelashvili et al., 2020b).

Based on the Empathy-Altruism hypothesis (Batson et al., 1987; Batson et al., 2015), we hypothesized (Hypothesis 1) that providing financial help to individuals affected by the pandemic is a behavior primarily benefiting the other person. Consequently, it will be positively associated with empathic concern and negatively associated with personal distress. Additionally, we reasoned that behaviors of maintaining physical hygiene and physical distancing are actions benefiting both oneself and others. Therefore, we expected (Hypothesis 2) that they would be positively associated with both empathic concern and personal distress.

Method

Participants

Participants were 1,263 Israeli citizens (Mage = 41, SDage = 15; 51% female, 49% male; 78% Jewish, 12% Arab; 10% Other), consisting of a nationally representative sample of gender, age, and Jewish–Arab groups in the population. Half (52%) were fully or partially employed. A sensitivity analysis conducted in G-power suggested that with the standard criterion of α = 0.05, the regression analysis with two predictors (Personal distress and Empathic concern) had a power of 0.80 to detect a small effect (f2 = 0.008). The Ethics Committee of the Psychology Department at the Hebrew University approved the study, and informed consent was obtained from all participants.

Measures

Generosity

The measure of generosity was adapted from Sjåstad (2019). Participants were asked to imagine that they were provided with $100, and how much (if any) they would keep themselves (%), how much they would give to the Red Cross charity in their country (%), and how much they would give to a global Red Cross charity (%), adding up to a total of 100%. Given the negligible percentage of money donated in Israel to the global Red Cross charity (Median = 0%; Mean = 10%; i.e., floor effect) and the biased distribution of this data (Zkurtosis = 41.45), our analysis focuses on the percentage of donation to the Red Cross charity in their country (i.e., Israel) as the measure of generosity. We used a hypothetical scenario to measure generosity, based on preceding research showing that real and hypothetical economic decisions are positively correlated (e.g., Ben-Ner et al., 2008; Johnson & Bickel, 2002; see also Camerer & Mobbs, 2017). In the current study, the average amount given to the national Red Cross charity (“Magen David Adom”) was 39.5$ (SD = 31.7).

Physical distancing

The physical distancing measure was adapted from Van Bavel et al. (2022) and originally included five items (e.g., “During the days of the coronavirus (COVID-19) pandemic, I have been staying at home as much as practically possible”; 0 = strongly disagree, 10 = strongly agree; for full list of items see Supplementary Materials). Follow-up analyses indicated that a reliable scale could be created by dropping one item (i.e., during the pandemic, I have been visiting friends, family, or colleagues outside my home; Cronbach’s α increased from 0.45 [before] to 0.74 [after]). The same unreliable item was also dropped for all other samples who participated in the international survey; see Van Bavel et al., 2022). Thus, the composite measure of physical distancing was calculated based on the four items (M = 8.49, SD = 1.70, Cronbach’s α = 0.74).

Physical hygiene

A physical hygiene scale included five items (e.g., “During the days of the coronavirus (COVID-19) pandemic, I have been washing my hands longer than usual”; 0 = strongly disagree, 10 = strongly agree; for a complete list of items see Supplementary Materials). Averaging the responses to all items provided the measure of physical hygiene (M = 7.89, SD = 1.78, Cronbach’s α = 0.78).

Empathic concern and personal distress

We measured two facets of affective empathy: Empathic Concern and Personal Distress. Specifically, participants were asked: “When you think about an individual who has just tested positive for COVID-19, how does it make you feel?” Participants responded by indicating the intensity of several empathy-related emotions (sympathetic, compassionate, moved, uneasy, upset, overwhelmed; 0 = not at all, 6 = very much). These emotion words have previously been used to measure Empathic Concern and Personal Distress (see Batson et al., 1987; Israelashvili et al., 2020b). Confirmatory factor analysis yielded the predicted two-factor model, reflecting Empathic Concern (sympathetic, compassionate, moved) and Personal Distress (uneasy, upset, overwhelmed), [χ 2 = 19.082 (p < .001; df = 8); RMSEA = 0.033 (90% CI: 0.014,0.053); SRMR = 0.015; TLI = 0.994; CFI = 0.997]. Accordingly, the index of Empathic Concern was calculated by averaging the ratings of feeling sympathetic, compassionate, and moved (M = 4.21, SD = 1.43, Cronbach’s α = 0.88), and the index of Personal Distress was calculated by averaging the ratings of feeling uneasy, upset, and overwhelmed (M = 2.70, SD = 1.69, Cronbach’s α = 0.87). The correlation between these two facets of empathy was positive (rSpearman =0.49, p < .001).

Procedure

All three dependent measures (Generosity, Physical distancing, and Physical hygiene) were administered first, in randomized order, as part of a larger international collaboration run during the early months of COVID-19. This survey also included other measures less relevant to the current question (e.g., relating to morality, national identification, etc.; for full list of measures, see Azevedo et al., 2023). Each participating collaborator was then able to add individual questions of interest, and so the empathy measure was added at the end of the Israeli survey.

Data analytic strategy

Statistical analyses were conducted using jamovi version 2.3. We assessed the structural validity of the two-factor model of empathy using chi-square (χ2), Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR), Tucker-Lewis Index (TLI), and Comparative Fit Index (CFI). Concurrently, we evaluated all single-factor scales through Reliability Analysis (Cronbach’s alpha). The two facets of empathy exhibited non-normal distributions (Shapiro-Wilk values > 0.9, p < .001), necessitating the use of a non-parametric test (i.e., Spearman correlation) to calculate their relationship.

To explore the connections between individuals’ levels of empathic concern (EC) and personal distress (PD) with COVID-related behaviors, we employed a multiple regression analysis. In each regression analysis for the three dependent variables (DVs)—Generosity, Physical distancing, and Physical hygiene—both Personal Distress and Empathic Concern were entered as predictors simultaneously. This simultaneous entry allowed us to discern the unique contribution of each empathic reaction while controlling for the other, providing a comprehensive understanding of how different facets of empathy relate to each outcome variable. Prior to analysis, we followed García and colleagues’ (2015) recommendation, standardizing all variables to reduce multicollinearity (VIFs before > 10; VIFs after < 2). The alpha level set for determining significance was 0.05, and 95% confidence intervals and standardized beta coefficients are presented in the multiple linear regression analyses. All three dependent measures (Generosity, Physical distancing, and Physical hygiene) deviated from normal distribution (Shapiro-Wilk values > 0.5, p < .001). To enhance robustness and address violations of normality assumptions, we utilized a bootstrap technique with 5,000 samples (Efron & Tibshirani, 1994) to assess the significance of all effects.

Results

COVID-related behaviors

To test whether individuals’ levels of empathic concern and personal distress were associated with their COVID-related behaviors, we conducted a multiple regression analysis for each of the three DVs (Generosity, Physical distancing, Physical hygiene). All three models were significant and explained 3–9% of the variance in the COVID-related behaviors: Generosity, F(2, 1259) = 22.5, p < .001; Physical distancing, F(2, 1245) = 25.8, p < .001; Physical hygiene, F(2, 1259) = 62.8, p < .001 (see Table 1 for all statisticsFootnote 1).

Table 1 Standardized weights of emotional concern and personal distress in accounting for individual differences in COVID-related behaviors, using generosity, physical distancing, and physical hygiene in separate linear regression analyses (N = 1262)
Full size table

The results of the regression analyses indicate that EC is positively associated with COVID-related behaviors. EC was positively related to endorsing prosocial/preventive behavior for all measures. The relation between PD and COVID-related behavior depends on the specific behavior. Figure 1 illustrates these effects.

Fig. 1
figure 1

Standardized coefficients of empathic concern and personal distress in accounting for individual differences in generosity, physical distancing, and physical hygiene behaviors. Error bars represent SEs

Full size image

Discussion

The outbreak of the COVID-19 pandemic called for urgent attempts to translate psychological science into real-life impact (van Bavel et al., 2020). To address this challenge, several studies suggested a link between eliciting empathy and adherence to health recommendations. However, the accumulating evidence shows this relation is inconsistent, ranging from moderate (e.g., Gamble et al., 2023) to weak (e.g., Pfattheicher et al., 2020; Karnaze et al., 2022) to non-significant (e.g., Favero & Pedersen, 2020). In the present study, we adopted a more nuanced approach that considers different facets of empathy (empathic concern, personal distress; Davis, 1983) and examines their relations to preventive measures and prosocial underpinnings of adherence behaviors. The crux of our findings is that people who experience empathic concern maintain greater physical distance and physical hygiene, and act more generously. We further found that individuals who experience personal distress are less likely to act generously, albeit more likely to maintain personal hygiene and physical distance. We now turn to discuss these findings in detail.

The findings that people with higher empathic concern were more likely to adhere to preventive measures are consistent with similar results obtained in several other studies (e.g., Gamble et al., 2023; Karnaze et al., 2022; Petrocchi et al., 2021; Pfattheicher et al., 2020) and together indicate their robustness. In an extension of existing research, the current study is the first to show that people who experience self-focused empathic reactions, characterized by personal distress, also engaged in more compliance. Thus, adherence to preventive measures can be driven by two motivations: protecting others and protecting oneself from getting infected. On the practical level, our findings that heterogenous empathic reactions (concern, distress) are associated with compliance suggest fewer barriers to using empathy as a path to encourage behavioral compliance. Any messaging or intervention that engages empathic concern or personal distress (cf., Favero & Pedersen, 2020) will effectively promote adherence to preventive behaviors.

The second contribution of the current study is the findings that different facets of empathy– empathic concern vs. personal distress were differentially associated with generosity, or motivation to act prosocialy. Specifically, as predicted by the Empathy-Altruism hypothesis (Batson et al., 2015) we found that people who experienced empathic concern acted more generously, while people who experienced personal distress showed less generosity. These patterns are consistent with previous low-powered studies with university students (e.g., Batson et al., 1987; Batson & Oleson, 1991, 2015), which found that empathic concern is associated with increased prosocial behavior, whereas personal distress was associated with reduced help to others (see also Eisenberg et al., 2006; Trommsdorff et al., 2007; Zaki, 2014; Batson 2019). In an extension of previous findings and perspectives, the present study demonstrates that these patterns remain stable even in a global pandemic and across more demographically diverse populations.

Interestingly, we found that individuals who experienced more personal distress were less likely to act generously, albeit more likely to maintain personal hygiene and physical distance. The negative effect of personal distress on generosity but the positive effect on personal hygiene and distancing seems related to the question of who benefits from each action. People who adhere to health recommendations help themselves, and others stay healthy. In contrast, people who donate money to others prioritize the financial interest of others over saving the same money for themselves. Taken together, the finding that people who experience personal distress adhere to preventive measures while showing less generosity suggests that compliance per se is not prosocial behavior. This aligns with the literature on the Empathy-Altruism Hypothesis (for a review, see Batson, 2019), which posits that personal distress is positively associated with prosocial outcomes only when helping the other person is beneficial for the self (e.g., washing hands to protect oneself from contracting the virus). However, interventions that elicit high levels of personal distress might also prompt people to act selfishly, such as engaging in hoarding behaviors during the pandemic, as the primary motivation of feeling personal distress is to benefit oneself, not another person. On a practical level, our findings highlight that any intervention or nudge aiming to increase prosociality should focus on enhancing empathic concern while minimizing personal distress.

Limitations

There are several limitations to note about this study. In general, self-reports may not accurately reflect actual behaviors (e.g., due to social desirability; Sassenrath, 2020). However, in the COVID-19 pandemic, a recent study demonstrated that self-reported adherence to social distancing was positively associated with real-world behavioral measures of reduced traveling (Gollwitzer et al., 2020). Similarly, several studies found that hypothetical and actual economic decisions are positively correlated (e.g., Ben-Ner et al., 2008; Johnson & Bickel, 2002; Camerer & Mobbs, 2017). While this suggests that the self-report measures utilized in the current study likely cohere with actual behaviors, future research that directly measures these outcome behaviors will increase confidence in the patterns we observed. For example, utilizing wearable devices equipped with sensors to monitor physical movement, proximity to others, and quality of interactions in real-time could provide an objective assessment of compliance and interaction quality. An additional limitation of our current research is its reliance on cross-sectional data. While previous studies have begun to establish specific causal relationships of empathic concern with prosocial or adherence behaviors (e.g., Batson et al., 2015; Pfattheicher et al., 2020), future research employing experimental designs could extend our findings by establishing causal links between concern and distress and both prosocial behavior and public health adherence. However, we acknowledge the challenges associated with replicating the unique environment of the COVID-19 pandemic for future experimental studies. Furthermore, the identification of small effect sizes, though common in this field (e.g., Karnaze et al., 2022; Pfattheicher et al., 2020), emphasizes the need for a careful interpretation of our findings. Additionally, both sampling firms that helped collect the participants used a quota sample, a non-probability sampling method. There is no guarantee that our sample is representative of the general population in terms of other features beyond gender, age, and ethnical groups in the population. Finally, it is recognized that a multitude of factors are associated with adherence to public health measures. These factors include personal health beliefs (Romano et al., 2022), conspiracy theory beliefs (Van Oost et al., 2022), and psychological reactance (Díaz & Cova, 2022). Demographic elements, such as gender (Hajek & König, 2022; Romano et al., 2022), in addition to interpersonal and institutional trust (Yuan et al., 2022), political ideology (Becher et al., 2021), social dominance orientation (Choma et al., 2021), and autonomous motivation (Morbée et al., 2021), have also been found to correlate with public health measures. Future interventions aimed at mitigating the spread of viruses will benefit from investigating the relative importance of each factor before deciding on which individual and situational factors their intervention should target.

Conclusion

To summarize, exposure to others’ distress may simultaneously involve heterogeneous empathic reactions, eliciting feelings of empathic concern and personal distress. Their positive relations to compliance suggest fewer barriers to using empathy to encourage adherence to health recommendations. Their opposing relations to prosociality suggest that increased empathy may not always translate into individuals acting more prosocially. In practical terms, our findings underscore the importance of recognizing the nuanced nature of empathic reactions. For managers and leaders, fostering a workplace environment that encourages empathic concern while mitigating personal distress can inspire individuals to not only meet established standards but to surpass them.