Group-based shame, guilt, and regret across cultures.

To date, there has been no systematic examination of cross-cultural differences in group-based shame, guilt, and regret following wrongdoing. Using a community sample (N = 1358), we examined people's reported experiences of shame, guilt, and regret following transgressions by themselves and by different identity groups (i.e., family, community, country) in Burkina Faso, Costa Rica, Indonesia, Japan, Jordan, the Netherlands, Poland, and the United States. We assessed whether any variation in this regard can be explained by the relative endorsement of individualistic or collectivistic values at the individual level and at the country level. Our findings suggest that people's reported experience of these emotions mostly depends on the transgression level. We also observe some variation across individuals and countries, which can be partially explained by the endorsement of collectivistic and individualistic values. The results highlight the importance of taking into account individual and cultural values when studying group-based emotions, as well as the identity groups involved in the transgression.

INTRODUCTION

Do I feel any guilt about colonialism? Not really. For one thing, it happened in the past and I didn't have anything to do with it.

–Robert James, Quora, 12 March 2017

Many Sinhalese feel extremely ashamed about it. That is one reason why they don't want to recall it—a collective shame.

–Jayadeva Uyangoda, The Hindu, 20 July 2003

Do British people feel guilty for what their country has done to India? The answer to this seemingly straightforward “yes” or “no” question on Quora garnered 75 responses as of November 2020. James’ response (quoted above) is emblematic of many of these: why should people today feel guilty for something that happened in the past? Notably, these reactions seem to reflect a broader public sentiment within the United Kingdom regarding its colonial past. In 2014, a YouGov study found that of the 1741 British participants sampled, 19% felt ashamed when thinking about the British Empire compared to 59% who felt proud (Dahlgreen, 2014). This sentiment contrasts with the shared feeling of shame—described by Jayadeva Uyangoda in an interview—about the 1983 anti‐Tamil pogroms and riots among the Sinhalese in Sri Lanka. In the same article, a fellow academic describes the event as, “a shock for the Sinhalese… There has been guilt about it” (Sambandan, 2003).

Although the two examples present opposing emotional responses to past transgressions, the possibility of such feelings following wrongdoing committed by one's group (mostly country)—including group‐based guilt, but also shame and regret—has become an increasingly studied topic over the past two decades (e.g., Branscombe & Doosje, 2004; Leach et al., 2013; Lickel et al., 2011). Building on social identity and self‐categorization theory (Tajfel & Turner, 1979), researchers have proposed “that when people identify with their group, they will appraise social objects or events in terms of their implications for the group” (Smith & Mackie, 2016, p. 174). More specifically, it has been posited that people are able to experience an emotional response to wrongdoing committed by their group (even when they were not personally involved), when they identify with it and appraise its actions as wrong or immoral as per their group norms (e.g., Doosje et al., 1998; Rees et al., 2013).

These specific emotions may be triggered by different concerns, such as a concern for the ingroup's reputation or moral failure (shame), a sense of shared responsibility (guilt and arguably regret), or by an empathetic concern for how the actions affect others or the outgroup (regret) (e.g., Branscombe et al., 2016; Gausel et al., 2012; Imhoff et al., 2012; Komiya et al., 2011; Lickel et al., 2011). What these emotions share, however, is their cooperative or “affiliative” social function (e.g., A. H. Fischer & Manstead, 2016; Mesquita et al., 2017), as they may trigger reconciliatory actions that can help repair and improve social relationships (e.g., Branscombe et al., 2016; Halperin, 2015; Imhoff et al., 2012). In other words, they share the potential to contribute positively to broader reconciliatory processes, hence garnering academic interest.

Despite studies lending empirical support that people can experience group‐based shame and guilt, however, findings on the prevalence of these emotions are often inconsistent or—in the case of group‐based regret—limited. Take for example two separate studies that looked at emotional responses among British students towards their country's colonial past. In one study by Allpress et al. (2010), students reported moderate feelings of shame and guilt for the British's aggressive response to the Mau Mau revolution in Kenya. Yet, in a study by Morton and Postmes (2011), the students sampled disagreed with any feelings of guilt for the British involvement in the slave trade. Of the two contradictory outcomes, Leach et al. (2013) note that Morton and Postmes’ findings are more in line with other studies on feelings towards European colonization—which mostly report either low levels or no self‐critical sentiment (such as group‐based shame and guilt). Following a review of 35 studies, they conclude that “explicit and strong self‐criticism for past generations’ genocide, or other mass violence, is a rarity” (Leach et al., 2013, p. 47).

One potential concern with this conclusion, however, is that despite the range of countries covered in these studies, most (86%) use Western samples—including people from the US, Australia, or various European countries. In addition, 60% of these Western samples are student‐based. As in other studies, the predominance of such “Western, Educated, Industrialized, Rich and Democratic” or “WEIRD” samples has implications for the generalizability of the reported findings (see Henrich et al., 2010 for a discussion), as the results do not reflect the diversity of the general population at the nation level—or at the global level. This raises the question whether the absence of a broader cross‐cultural perspective has led to a more restricted view of the prevalence and degree to which people report group‐based shame, guilt, and regret.

The lack of a cross‐cultural perspective is surprising as there are theoretical grounds to assume that cultural differences—particularly the extent to which people see themselves as more independent or interdependent from their group—may affect how they experience these group‐based emotions. These differences are considered part of the individualism or collectivism cultural syndromes, or “schema of shared norms, beliefs and practices.” The former represents a schema in which people value independence and autonomy, and prioritize personal goals, while interdependence, and group norms and goals such as harmonious relationships and mutual obligations are more valued within the collectivism schema (Oyserman, Coon, et al., 2002; Triandis, 1996).

Although cross‐cultural research in this regard is limited, researchers have suggested that these individualistic and collectivistic values may affect how people respond to transgressions by the ingroup. For example, in a study comparing Chinese and American emotional responses to situations concerning either one's self or another, Stipek (1998) found that both Chinese and American participants reported higher levels of shame and guilt for personal transgressions than when a family member did something. The difference between the two levels was less pronounced for the Chinese participants, however, as they reported higher levels of guilt and shame for something done by a family member than the American participants. Stipek posited that this may be because of greater emphasis on interdependence in China, as “Chinese boundaries between self and other are weaker” (Stipek, 1998, p. 625). Furthermore, various studies that examine cultural variation in personal shame and guilt suggest that individualistic and collectivistic values also affect the salience of these emotions, in part because the degree of interdependence may be more or less salient but also because these emotions trigger responses that are in line with salient cultural objectives such as maintaining harmonious relations (e.g., Fontaine et al., 2006; Sheikh, 2014; Young et al., 2021). It is plausible that these effects extend to their group‐based counterparts.

So far, however, previous research has not systematically examined cross‐cultural differences in shame, guilt, or regret following country or other group level transgressions. The primary objective of this article, therefore, is to examine whether these three group‐based emotions differ across cultures, and whether variation in this regard can be explained by the relative endorsement of individualistic or collectivistic values at both the individual and the cultural level. Given the paucity of cross‐cultural theorizing on group‐based emotions, we draw on theorizing and research from a social identity perspective to develop our expectations on how individual variation in the endorsement of individualistic and collectivistic values may affect these three group‐based emotions. More specifically, we build on the assumption that the endorsement of these values reflects people's relationship with and orientation towards groups, which allows us to draw parallels with the theorizing on ingroup identification and group‐based emotions. At the cultural level, we rely on theorizing and research on how cultural mandates may shape how people “do” emotions.

Additionally, given that this is the first study (to the best of our knowledge) to systematically examine any cultural variation in group‐based emotions, we felt that there are insufficient empirical grounds to make any predictions yet on how the relative endorsement of individualistic and collectivistic values may differentially impact group‐based shame, guilt, and regret. Hence, despite us seeing shame, guilt and regret as distinct, our expectations concerning these emotions are based on their shared characteristics as self‐critical emotions with the potential for prompting reconciliatory action.

Individual level variation in group‐based shame, guilt, and regret

Starting with how individual variation in the endorsement of collectivistic values may affect group‐based shame, guilt, and regret, the literature suggests that there may be two possible—albeit contradictory—trajectories. On the one hand, people who more strongly endorse collectivistic values may be more likely to experience group‐based emotions. On the other hand, the opposite may be true for critical group‐based emotions such as guilt (e.g., Doosje et al., 1998; Roccas et al., 2006), and arguably shame and regret. This paradoxical set of outcomes is predicated on the effect that group identification has been theorized and found to have on the experience of such emotions. Even though this theorizing and research is based on more specific group identities, we expect people who strongly endorse collectivistic values to exhibit similar tendencies to those exhibited by people who identify strongly with their group because of the relative importance of, and their shared orientation towards, the group.

The core premise underlying much of the current thinking on group‐based emotions is that people can experience these when they identify with and see themselves as part of a particular social group (as also postulated in social identity theory and self‐categorization theory) (e.g., Branscombe et al., 2016; Mackie & Smith, 2014). Intergroup emotions theory (Mackie & Smith, 2014, 2015) builds on this premise by positing that people's emotional and behavioral reactions to intergroup situations are informed by their group's concerns or goals, rather than by their own personal concerns. They appraise situations in which their social or group identity is made salient in terms of what these situations might mean for their group, and any emotional response is the result of group‐based triggers, which in turn can have group‐based consequences. This process is only assumed possible if people consider themselves a member of their group, which is why group membership is considered a prerequisite for group‐based emotions. Group identification, or the degree to which people categorize themselves as members and how integral this group membership is to their identity, is seen as a moderator that affects both the likelihood and intensity of group‐based emotions (e.g., Iyer & Leach, 2009; Mackie & Smith, 2015).

Studies on group identification and group‐based guilt and shame, however, present a mixed, even paradoxical, picture (e.g., Iyer & Leach, 2009; Klein et al., 2011; Roccas et al., 2006). On the one hand, there is empirical evidence that suggests that people who are strongly group‐oriented are more likely to report group‐based emotions. For example, Johns et al. (2005) found that American students, who identified more strongly with being American, reported more shame for events that described severe and harmful prejudice shown towards people of Middle Eastern descent by other Americans. Building on this, we would assume that people who adhere more to collectivistic values are more likely to report these group‐based emotions, because group membership is more likely to be an integral part of their identity due to a heightened sense of interdependence.

On the other hand, researchers have suggested that this positive relationship is more likely for positive or less‐aversive group‐based emotions—such as pride and hope—than for negative group‐based emotions—such as guilt, shame, and regret (e.g., Halperin, 2015; Iyer & Leach, 2009; Smith et al., 2007). In the case of negative or critical group‐based emotions, evidence suggests that their aversive nature may actually trigger the reverse, and that people who are more strongly group‐oriented may be less likely to experience such critical emotions because they are motivated to maintain a more positive group image and to protect their group identity (e.g., Doosje & Branscombe, 2003; Doosje et al., 1998). For example, Doosje et al. (2006) found that among Dutch students presented with an account of the Dutch colonization of Indonesia that incorporated both favorable and unfavorable aspects, the students who identified more strongly with the Netherlands were more likely to display defensive behavior and less likely to report guilt compared to those who identified less. It is possible that for people who strongly endorse collectivistic values, a scenario in which their group has done something wrong may potentially trigger similar defensive responses, so as to avoid disrupting their positive group image or to maintain ingroup harmony (e.g., Leung & Bond, 1984). People may, for example, justify or deny the group's actions and, therefore, express less group‐based shame, guilt, and regret.

While there are grounds to assume that a stronger endorsement of collectivistic values may affect group‐based emotions in either direction, it is less clear how the endorsement of individualistic values may affect group‐based emotions. At most, we assume that people who adhere more strongly to individualistic values are less group‐oriented and more self‐focused. Hence, they may be more likely to appraise situations in terms of the implications they have for them personally, rather than how they may affect their group. Building on this assumption, we would anticipate that people who endorse more individualistic values are less likely to report group‐based shame, guilt, and regret.

Cultural level variation in group‐based shame, guilt, and regret

We also examine how the relative endorsement of individualistic and collectivistic values within a cultural setting may affect group‐based shame, guilt, and regret. Here too, there are grounds to assume that a stronger cultural endorsement of collectivistic values is likely to affect the experience of these group‐based emotions. Although the dynamics underlying the emotional process at the cultural level is different, our expectations are similar to those at the individual level, namely that a more collectivistic cultural context may either be conducive and therefore “encourage” group‐based shame, guilt, and regret or—in contrast—hinder or “suppress” these emotions.

Here, we build upon previous work showing that how people “do” emotions is informed by cultural mandates or “cultural norms [and] ideals … for how to be a good person, how to interact, how to build good relationships, or … how to feel” (Mesquita et al., 2017, p. 97). Researchers have shown that how people interpret a situation and their subsequent emotional response “is dependent in part on whether it is culturally acceptable and normative to do so, and thus on whether it is functional within the particular sociocultural context” (Mesquita et al., 2017, p. 97). Thus, these different cultural mandates can affect the salience or threshold of emotional reactions by either encouraging or upregulating certain emotions or by suppressing or down‐regulating them. For instance, previous studies have found cultural variation in the frequency and intensity of individual‐level emotions, including—for example—a study in which Japanese participants reported experiencing negative engaging emotions such as guilt and shame more intensely than Americans (e.g., Boiger et al., 2014; Kitayama et al., 2006; Mesquita et al., 2017).

The underlying premise for the effect of cultural mandates on the salience or threshold of emotional reactions is that emotions are promoted when their social functions (either more “affiliative” or “distancing”) are in line with or fit with the cultural mandate, or suppressed if their social functions clash with the cultural mandate (e.g., A. H. Fischer & Manstead, 2016; Mesquita et al., 2017). Within more collectivistic cultural contexts, emotions that have a more affiliative social function—that is, help “establish and maintain cooperative and harmonious relations” (A. H. Fischer & Manstead, 2016, p. 3)—are likely to be encouraged, as the group's concerns (such as ensuring social harmony and fulfilling mutual obligations) take precedence over one's own. Going back to the previous example, it is likely that shame and guilt are encouraged within Japan (and therefore more prominent), because these emotions are in line with the cultural emphasis on maintaining social harmony as they can motivate actions that help repair social relationships (Kitayama et al., 2006).

Overall, the literature on how the cultural context may affect personal (i.e., following personal actions) shame, guilt, and regret suggests that these emotions are likely to be encouraged within collectivistic settings because their potential for socially oriented consequences (such as improved social harmony) is most salient. For example, Rozin (2003) observed that Indian participants were more likely to agree with the statement “shame is more similar to happiness than to anger” than American participants, because they considered both shame and happiness as being socially constructive rather than disruptive. In the case of regret, researchers have suggested that the relative importance of mutual obligations and social harmony may instill greater regret for actions that hamper social relations. For example, Komiya et al. (2011) found that Japanese participants were more likely to regret situations in which they had hurt or annoyed friends or family than their American counterparts. The literature also suggests that guilt may be more desirable in collectivistic cultural contexts due to their reparative action tendencies (such as apologies), as these can help restore social relationships (e.g., Baumeister et al., 1994; Silfver‐Kuhalampi et al., 2013).

Much of this evidence supports the notion that shame, guilt, and regret following personal transgressions are encouraged in collectivistic settings, as they can be conducive to restoring one's personal relationship with one's group. It is possible that this pattern also holds for group‐based shame, guilt, and regret following group level transgressions. A. H. Fischer and Manstead (2016) theorize that the collectively experienced threat to the group's social identity may motivate these emotions to drive pro‐social behaviors such as reparations or collective action and thereby strengthen social bonds. This would suggest that group‐based shame, guilt, and regret are encouraged and more likely within more collectivistic cultural contexts.

It is also possible, however, that this experienced threat and the aversive nature of these emotions may have the opposite effect. Instead of triggering reparative actions that improve ties with the outgroup, the salience of maintaining social harmony within collectivistic settings may potentially trigger defensive responses that emphasize group loyalty to ensure greater ingroup cohesion (e.g., Branscombe et al., 1999; Ellemers et al., 2002; Roccas et al., 2006). Hence, it is plausible that group‐based shame, guilt, and regret following group transgressions may be suppressed within collectivistic cultural contexts to avoid disturbing the unity and identity of the ingroup (Oyserman, Coon, et al., 2002) rather than restoring the ingroup's relationship with the outgroup.

Thus, there are grounds to assume that a stronger cultural endorsement of collectivistic values may make group‐based shame, guilt, or regret either more or less likely. It is possible, however, that a stronger cultural endorsement of individualistic values inhibits group‐based shame, guilt, or regret, also because independence and autonomy are culturally mandated in such contexts. This may have the effect that people's emotional experiences are more likely to be driven by personal concerns, motivations, and goals and therefore less conducive to group‐based emotions. There is also evidence that emotions with a “distancing” function (such as anger) are more salient in more individualistic cultural contexts because these facilitate differentiation and competition, rather than emotions with an “affiliative” function (e.g., De Leersnyder et al., 2015). This would suggest that group‐based shame, guilt and regret are less likely in these contexts.

The present study

To summarize, the goal of this study is to examine whether group‐based shame, guilt, and regret following country‐level wrongdoing varies as a function of individual and cultural endorsement of individualistic or collectivistic values. Drawing on the literature, we could predict either positive or negative relationships between the relative endorsement of collectivistic values at the individual and cultural level and the extent to which people may report these group‐based emotions.

At the individual level, we examine the following competing expectations:

1a. People who more strongly endorse collectivistic values are more likely to report group‐based shame, guilt, and regret; or,

1b. People who more strongly endorse collectivistic values are less likely to report group‐based shame, guilt, and regret.

Similarly, at the cultural level, we examine the following competing hypotheses:

2a. Group‐based shame, guilt, and regret are more likely within cultural contexts in which collectivistic values are more strongly endorsed; or,

2b. Group‐based shame, guilt, and regret are less likely within cultural contexts in which collectivistic values are more strongly endorsed.

In contrast, our expectations regarding a stronger endorsement of individualistic values are comparatively straightforward. We expect:

3a. People who more strongly endorse individualistic values are less likely to report group‐based shame, guilt, and regret.

3b. Group‐based shame, guilt, and regret are less likely within cultural contexts in which individualistic values are more strongly endorsed.

To assess these expectations, we conducted a study across eight countries from different continents: Burkina Faso, Costa Rica, Indonesia, Japan, Jordan, the Netherlands, Poland, and the United States. Given the relative focus on Western countries in previous work on group‐based emotions, we chose these countries to ensure greater geographic diversity, and because the relative salience of individualistic or collectivistic values across these countries was presumed to vary. In addition to capturing a diverse and unique country‐based sample, we also relied on community rather than student samples to ensure broader within‐country diversity.

Given that our primary scholarly interest is in people's emotional responses following wrongdoing committed by their country, but also that much of the work on group‐based emotions is on the country level, our main focus is on the reported experience of group‐based shame, guilt, and regret following country level transgressions. We wanted to take into account, however, the possibility that these group‐based emotions may vary depending on how relevant a group might be for someone. For example, emotional responses following actions by one's country may be less strong than following transgressions by one's family, because the situation may be less (directly) relevant, and of lesser concern for the person (e.g., Lickel et al., 2005). For comparative purposes, we therefore also measured people's responses to transgressions by their family and community (i.e., neighborhood, village, or town). In addition, we assessed their emotional responses following personal transgressions. As much of the theorizing has been on the person level, this provided us with a theoretical anchor against which we can examine potential similarities and differences in group level and person level responses, especially as personal and group‐based shame, guilt, and regret share many of their emotion‐specific antecedents. The former, however, are triggered by appraisals at the person level—how does this situation concern me—while group‐based emotions follow group level appraisals—how does this situation concern my (in)group (e.g., Iyer & Leach, 2009).

METHOD

Participant samples

A total of 1371 people from Burkina Faso, Costa Rica, Indonesia, Japan, Jordan, Poland, the Netherlands, and the United States participated. In Burkina Faso, all participants were from the Mossi ethnic group, while in Indonesia and the United States, all participants came from Java and Alabama respectively. Our final sample included 1358 people, as we excluded 13 participants who were either not from the country in question, or had spent most of their life elsewhere. To ensure that we obtained a cross‐section of the population that was comparable across countries, we applied a stratified sampling approach. A sampling matrix was developed to identify an ideal sample—balanced in terms of gender, age categories (18–34, 35–64, and 65 or older), locality (urban/rural) as well as education levels (no or little education up to university level). To ensure a directly comparable categorization for education, we based the final categorization on the International Standard Classification of Education (ISCED) (UNESCO Institute for Statistics, 2012). A breakdown of the final sample per country, as well as overall demographics can be found in Table 1.

TABLE 1

Sample demographics

		Age						Education
Country	N	M	Range	Male	Female	Rural	Urban	Low	Medium	High
Burkina Faso	150	45.1	19–73	77	73	75	75	90	21	37
Costa Rica	153	44.7	18–87	73	80	65	88	37	60	56
Indonesia	181	44.0	18–85	81	100	95	86	53	71	57
Japan	214	48.1	18–88	107	107	97	114	78	47	89
Jordan	183	42.7	18–90	77	106	72	110	36	23	121
Netherlands	160	44.4	18–89	63	97	69	91	35	59	65
Poland	163	43.8	18–81	76	87	74	89	20	56	86
United States	154	42.4	18–84	56	96	42	112	49	9	90

Procedure and materials

Participants were either sent an online questionnaire or approached in person by interviewers with the same cultural background, in which case they were asked to fill out a pen‐and‐paper version of the questionnaire. The default varied across countries depending on the levels of online accessibility and computer literacy. For example, low internet connectivity and poor computer literacy in Burkina Faso meant that all data were collected in person. In contrast, in Japan, all data were collected online due to high internet connectivity and a preference for online participation. While participants in Japan were identified through a survey company panel, we adopted non‐probabilistic sampling techniques to identify participants elsewhere. This included snowball sampling using the networks of local researchers as a starting point, but also convenience sampling among local networks such as church groups, or community homes, or by approaching people in communal places, including for example university campus or bus/train stations. To distribute the survey online, researchers shared the link within specific WhatsApp groups, or by emailing specific contacts.

We developed the questionnaire in English, before formally translating it into Arabic, Dutch, French, Indonesian, Japanese, Polish, and Spanish. We then verified independent back‐translations through discussions with local experts to confirm the accuracy and validity of both the key terms and measures applied, including for example the individualism‐collectivism scale. Although most participants self‐completed the questionnaire—where necessary, researchers assisted by reading out the questions and items, and filling in the verbal responses. An overarching ethical approval for this study was obtained from the Tilburg School of Humanities and Digital Sciences Research Ethics and Data Management Committee. Additionally, where required, country‐specific ethical approvals were also obtained from local ethical committees. In addition to obtaining informed consent from all participants, all data collection and processing was conducted in accordance with the ethical codes of conduct of the European Charter for Researchers and the Netherlands Code of Conduct for Scientific Practice (2012).

Measures

To assess people's emotional reactions following wrongdoing, we asked participants to report how intensely they would experience shame, guilt, and regret in a situation in which they themselves, their family, their community (specified as their neighborhood, village, or town) or their country had harmed a group of people, using a seven‐point scale (0 = not at all to 6 = very strongly).1 Although previous studies have applied composite scales to measure group‐based shame and guilt (e.g., Doosje et al., 1998; Lickel et al., 2005), we opted for a single emotion item because we were concerned with cross‐cultural and cross‐level comparability. Using multiple emotion terms to capture a single sentiment (e.g., embarrassment and shame for “shame”) becomes problematic because emotion terms often do not translate directly and can cluster differently across cultures (e.g., Breugelmans & Poortinga, 2006; Edelstein & Shaver, 2007). In addition, while multi‐componential approaches such as the GRID model have been validated cross‐culturally with regard to individual emotions, given the paucity of cross‐cultural research on the antecedents and appraisals concerning group‐based emotions, we were concerned about the cross‐cultural validity and feasibility of such an approach (Fontaine et al., 2013). Using a more elaborate measure may be more robust, but it would require rigorous validation, especially given that the appraisal process for group‐based emotions is likely to be more complex (Gausel & Brown, 2012), and equally prone to linguistic complexities. In addition, the use of multiple items would—by lengthening the questionnaire—also add an additional burden on our participants. To determine what the best approach might be, we had extensive discussions with local experts and contacts about various “shame,” “guilt,” and “regret” terms. We also checked the emotion terms in a pilot (N = 345) that we conducted across the same eight countries through an online survey (with the exception of Burkina Faso, where the data were collected through a pen‐and‐paper survey). From these discussions and the pilot, we concluded that a single emotion term was preferable over a more elaborate approach, especially considering that we were working with community samples with different levels of education and literacy rates, and with often little experience in completing questionnaires. The final selection of these terms in each language were discussed at length to ensure they captured similar emotional responses. The correlations between these measures varied across countries and across the different transgression levels (ranging between r = 0.33 and r = 0.88 for the person level, between r = 0.21 and r = 0.77 for family, r = 0.17 and r = 0.86 for community, and r = 0.18 and r = 0.84 for country, ps < .001).2 In addition to shame, guilt, and regret, we included four other emotions—fear, anger, sadness, and humiliation—in the scale. This array of alternative plausible emotions was added to reduce potential response bias among participants and has therefore not been included in these analyses.3

Comparability was also a key factor in our decision to present participants with a broad and abstract description of “harm” as a form of wrongdoing—further defined as either treating others unjustly or hurting them physically—rather than with a real‐life situation. Considering the nature of the study, it was crucial for us to present participants with a situation that was comprehensible, as well as comparable across all countries and all transgression levels (country, community, family, personal). This enabled us to circumvent the issue of identifying real life situations that were directly comparable in terms of severity, timing, impact, and other contextual factors, which was problematic considering the diversity of the countries involved. We assumed that abstractification would facilitate a more consistent interpretation of the scenario across participants, and minimize the possible influence of conflicting socio‐political narratives.

To overcome the debatable validity (see Oyserman, Coon, et al., 2002; Taras et al., 2014; Vignoles et al., 2016 for a discussion), but also complexity and length of existing individualism‐collectivism scales, we applied a consensus approach to construct an individual level scale that integrated core items from seven different measures including those developed by Oyserman and Lauffer (2002), R. Fischer et al. (2009), Vignoles et al. (2016), and various measures quoted in Taras et al. (2014). We limited the scale to 12 items—six per subscale—to avoid overburdening our participants. Considering their different ages, educational backgrounds, and levels of experience with survey scales, it was crucial to ensure we had a short, comprehensible and yet valid scale to assess people's relative endorsement of individualistic and collectivistic values.

Drawing from a pool of 223 items across the seven measures, we focused solely on items measuring independence, duty and harmony—identified by Oyserman, Coon, et al. (2002) as core individualism‐collectivism framework elements. The final item selection was based on how well each item captured the construct's meaning—in English, but also when translated in the other languages—and how frequently it was included across scales. If an item existed across multiple scales, we assumed greater agreement and considered it a more representative measure. Our individualism subscale included, for example, items that emphasized independence and autonomy such as—“It is important for you to act as an independent person” and “You do your own thing regardless of what others think.” The collectivism subscale included items that emphasized harmony—for example, “It is important for you to maintain harmony within your group(s)”—and duty—“You always put your group's goals first, even if it means giving up your personal goals.” To minimize introspection (Vignoles et al., 2016), we altered items slightly (where necessary) to ensure that “you” was used as the identifier for each statement. Respondents were asked to rate each statement on a seven‐point scale (1 = strongly disagree to 7 = strongly agree).

The scales were tested in the aforementioned pilot (N = 345) that we conducted across the same eight countries. A factor analysis yielded a three‐factor solution, but the results indicated that the reliability of the scales varied, scoring better in some countries than in others. We therefore discussed items that performed less well within each country with local experts, and where necessary, edited the wording to ensure greater clarity across the various languages.

To examine the dimensions of the final individualism and collectivism scales, we conducted multigroup exploratory factor analyses with varimax rotation. These analyses suggested that a two‐factor solution was the most optimal, with the collectivism items loading on one factor and the individualism items loading on the other factor. To check the similarity of the factor solutions across countries, we ran a Procrustes rotation (using the average correlation matrix of the items across all countries as a reference). This analysis revealed that the individualism dimension showed good factorial agreement across countries (average Tucker's phi is 0.94), but that the factorial agreement for the collectivism dimension was low for Costa Rica and Burkina Faso (Tucker's phi 0.76 and 0.84, respectively). Inspection of the item loadings revealed that one of the collectivism items (“You will stay with a group even if you are not happy with them”) had a negative loading in Burkina Faso and a very low (0.18) loading in Costa Rica. We therefore removed this item and reran the analyses. This resulted in a better solution for Burkina Faso (phi = 0.96) and so we decided to use this shortened scale, be it that for Costa Rica Tuckers phi is still low (0.71). Table 2 presents the alphas, mean scores, and standard deviations on the individualism and collectivism measures for the eight countries. The final scale can be found in the appendix.

TABLE 2

Mean scores, standard deviations and Cronbach's alphas for the individualism‐collectivism scales

	Collectivism			Individualism
	M	SD	α	M	SD	α
Burkina Faso	4.89	1.03	0.69	3.93	1.09	0.63
Costa Rica	4.87	0.92	0.53	5.06	0.96	0.58
Indonesia	5.57	1.03	0.79	5.01	1.03	0.72
Japan	4.18	0.85	0.73	4.17	0.83	0.70
Jordan	5.10	1.22	0.68	4.45	1.35	0.68
Netherlands	4.31	0.98	0.71	4.54	0.93	0.68
Poland	4.13	1.24	0.80	4.32	1.13	0.75
US	4.36	1.01	0.68	5.32	0.90	0.71

Note. Within‐country comparisons of the collectivism and individualism scales were significantly different at 0.05 or beyond for all countries, except for Japan (p = .91) and Poland (p = .21).

To be able to control for the possibility that people's responses to ingroup transgressions may vary as a function of the extent to which they identify with a particular group, we drew on Postmes et al.’s (2013) single‐item measure of social identification by asking participants to indicate to what extent they identify with their family, their community, and their country, using a seven‐point scale (0 = not at all to 6 = very strongly).

RESULTS

The data collected in this study have a multilevel structure: the emotions for transgressions at the country, community, family, and person level (Level 1) are nested within individuals (Level 2), who are nested within countries (Level 3). Although the number of countries is low for a three‐level analysis, we did estimate three‐level models (whereby effects at the country‐level were modeled as fixed) to obtain an understanding of the patterns at the both the individual and country‐level.

Descriptive statistics

To obtain basic multilevel summary statistics such as the mean and the variances at each level of analysis, we estimated so‐called null models whereby no predictors were entered. The results of these analyses can be found in Table 3. As the variance estimates show, most of the variance for guilt, regret, and shame is at the transgression level (Level 1) and at the individual level (Level 2). The variance estimates also show that the means for the countries (Level 3) vary, but that the within‐country variance is higher than the between‐country variance (which is relatively low, particularly for shame).

TABLE 3

Variance estimates for feelings of shame, guilt, and regret

		Variance
	M	Level 1 (Transgression)	Level 2 (Person)	Level 3 (Country)
Shame	4.05	1.76	1.52	0.12
Guilt	3.70	2.52	1.33	0.23
Regret	3.56	2.55	1.51	0.22

In a set of follow‐up analyses, we estimated the overall and country‐level means for feelings of shame, guilt, and regret for harm inflicted by oneself, one's family, one's community, and one's country. These results can be found in Table 4 and are illustrated in Figure 1. We compared these estimates within each country, using chi‐squared tests of fixed effects.

TABLE 4

Country‐level estimated means for shame, guilt, and regret across different transgression levels

		Personal	Family	Community	Country
Shame	Burkina Faso	3.70a	3.87a	2.96b	2.65c
	Costa Rica	4.17ac	4.26a	3.26b	3.90c
	Indonesia	4.64a	4.92b	4.41a	4.54a
	Japan	4.18a	4.39b	3.62c	4.12a
	Jordan	4.29a	4.64b	4.06a	4.18a
	Netherlands	4.77a	4.59a	2.98b	3.23b
	Poland	4.43a	4.34a	3.72b	3.85b
	US	4.82a	4.44b	3.88c	3.57d
Guilt	Burkina Faso	4.32a	3.87b	2.95c	2.68d
	Costa Rica	4.63a	3.32b	2.22c	2.53d
	Indonesia	5.12a	4.83b	3.87c	3.83c
	Japan	4.80a	4.50b	3.26c	3.58d
	Jordan	5.19a	4.67b	3.95c	3.79c
	Netherlands	5.09a	3.42b	2.14c	2.16c
	Poland	4.99a	3.31b	2.31c	2.42c
	US	5.14a	3.66b	2.90c	2.85c
Regret	Burkina Faso	4.48a	3.63b	2.82c	2.42d
	Costa Rica	4.79a	3.08b	1.99c	2.26d
	Indonesia	5.09a	4.60b	3.91c	3.78c
	Japan	4.73a	4.26b	3.04c	3.25d
	Jordan	4.70a	3.73b	2.96c	2.83c
	Netherlands	5.01a	2.72b	1.79c	1.72c
	Poland	4.04ab	4.26b	4.06a	4.10ab
	US	4.74a	3.41b	2.77c	2.77c

Note. Means within the same row not sharing a superscript were significantly different at 0.05 or beyond.

FIGURE 1

Country‐level estimated means for feelings of shame, guilt, and regret across different transgression levels (0 = not at all to 6 = very strongly

As can be seen in Table 4, the pattern of results for feelings of shame differed across the eight countries in our sample. For example, the differences between personal and group levels of shame were least pronounced in Japan, Indonesia, and Jordan. In these countries, the mean levels of shame for harm inflicted by oneself, one's country, and (in Indonesia and Jordan) one's community did not differ. In the United States, however, the mean levels of shame were higher for harm inflicted by oneself than for harm committed by one's family, community, and country. In Burkina Faso, Poland, Costa Rica, and the Netherlands, participants reported similar levels of shame for harm inflicted by oneself and one's family, but lower levels of shame for transgressions by the country and community.

The pattern of results for feelings of regret and guilt was more consistent across the different countries, as mean levels of guilt and regret were generally lower for harm inflicted by one's country than for harm inflicted by oneself or by one's family. In Costa Rica and Japan, mean levels of guilt and regret were higher for harm inflicted by the country than for that inflicted by the community, whereas this was the reverse in Burkina Faso. In the United States, Jordan, Indonesia, Poland, and the Netherlands, no differences were found in the extent to which participants reported feelings of guilt or regret for harm inflicted by their country or community. Only in Poland were no differences found in the mean levels of regret for harm inflicted by oneself and by one's family, country, or community.

Individual variation in group‐based shame, guilt, and regret

We first examined whether experiences of personal and group‐based shame, guilt, and regret vary as a function of the extent to which people endorse more individualistic or collectivistic values. For each of these emotions, we conducted separate analyses. In each of these analyses, we included the person, family, community, and country as dummy‐coded predictors (uncentered) at Level 1, whereby we dropped the intercept. At Level 2, we added the individualism and collectivism scores. These analyses are sometimes also referred to as “slopes as outcomes” analyses. We also added identification with family, community, and country as control variables at these respective levels. As these models did not include the country means for these measures, we centered the predictors at Level 2 around the group mean to remove any between‐country variation from these measures (e.g., Brincks, 2012). We also included the random error terms in the models.

As can be seen from Table 5, people's individualism scores were not related to the feelings of shame, guilt, or regret they reported for harm inflicted by their country, their community, their family, or themselves (βs ≤ 0.08, ps ≥ .11). For collectivism, however, we found numerous significant relationships. Participants who were higher on this measure reported more guilt, shame and regret following wrongdoing by their country, but also following wrongdoing by their community, their family, and by themselves.4 Follow‐up analyses revealed that the coefficients for harm inflicted by the country, community, or family did not differ from the coefficients following personal wrongdoing (x 2s < 2.87, ps > .08). Participants’ identification with their country, community, or family was positively related to the extent to which they experienced shame, guilt, and regret at these levels (βs > 0.12, ps < .001).

TABLE 5

Individual level relationships between endorsement of individualistic and collectivistic values and feelings of shame, guilt, and regret at different transgression levels

		Collectivism			Individualism
		β	SE	95% CI	β	SE	95% CI
Shame	Person	.22***	0.048	[0.12, 0.31]	0.02	0.048	[−0.07, 0.12]
	Family	.25***	0.051	[0.15, 0.35]	−0.02	0.050	[−0.12, 0.08]
	Community	.23***	0.053	[0.13, 0.34]	−0.04	0.051	[−0.14, 0.07]
	Country	.14**	0.042	[0.05, 0.22]	−0.02	0.040	[−0.10, 0.05]
Guilt	Person	.21***	0.044	[0.12, 0.29]	0.01	0.044	[−0.08, 0.09]
	Family	.17**	0.054	[0.06, 0.28]	−0.02	0.053	[−0.12, 0.09]
	Community	.12*	0.057	[0.01, 0.23]	0.03	0.054	[−0.08, 0.14]
	Country	.10*	0.042	[0.02, 0.18]	0.01	0.040	[−0.07, 0.09]
Regret	Person	.19***	0.048	[0.10, 0.29]	0.08	0.047	[−0.02, 0.17]
	Family	.12*	0.058	[0.01, 0.24]	0.00	0.056	[−0.11, 0.11]
	Community	.12*	0.059	[0.00, 0.23]	0.03	0.057	[−0.08, 0.14]
	Country	.09*	0.044	[0.00, 0.17]	0.06	0.042	[−0.02, 0.14]

Note: In these analyses, we controlled for participants’ identification with their family, community, and country at these respective levels (βs > 0.14, ps < .001 for shame, βs > 0.20, ps < .001 for guilt, and βs > 0.12, ps < .001 for regret).

p < .05, **p < .01, ***p < .001.

Cultural variation in group‐based shame, guilt, and regret

In a second set of analyses, we examined relationships between cultural level variation in the individualism and collectivism measures, and feelings of shame, guilt, and regret at the various transgression levels. We did this by adding the country means of the individualism and collectivism measures at Level 3 (grand‐mean centered) in the models that we estimated for each of these emotions. In each of these analyses, the predictors at Level 2 were centered on the grand‐mean as well since this allowed us to control for Level 2 mean scores (Brincks, 2012). Given the low number of countries, we modeled the Level 3 coefficients as fixed. The results of these analyses are displayed in Table 6.

TABLE 6

Cultural level relationships between endorsement of individualistic and collectivistic values and feelings of shame, guilt, and regret at different transgression levels

		Collectivism			Individualism
		γ	SE	95% CI	γ	SE	95% CI
Shame	Person	−0.33†	0.151	[−0.62, −0.03]	0.35†	0.161	[0.04, 0.67]
	Family	−0.22†	0.125	[−0.47, 0.02]	0.10	0.132	[−0.16, 0.36]
	Community	−0.04	0.129	[−0.29, 0.21]	0.00	0.136	[−0.27, 0.27]
	Country	−0.02	0.100	[−0.21, 0.18]	0.00	0.107	[−0.21, 0.21]
Guilt	Person	−0.25	0.167	[−0.58, 0.08]	0.17	0.177	[−0.18, 0.52]
	Family	0.05	0.131	[−0.21, 0.31]	−.62***	0.139	[−0.89, −0.35]
	Community	.27*	0.133	[0.01, 0.53]	−.51***	0.140	[−0.79, −0.24]
	Country	0.13	0.100	[−0.07, 0.32]	−.56***	0.108	[−0.77, −0.35]
Regret	Person	0.04	0.304	[−0.56, 0.63]	0.09	0.319	[−0.54, 0.71]
	Family	−.34*	0.139	[−0.61, −0.07]	−.84***	0.148	[−1.13, −0.55]
	Community	−.32*	0.140	[−0.60, −0.05]	−.91***	0.148	[−1.20, −0.62]
	Country	−.52***	0.105	[−0.73, −0.32]	−.85***	0.113	[−1.08, −0.63]

p < .05, **p < .01, ***p < .001.

The analyses show that—even though we found no relationships with the individual mean scores on individualism—there are numerous significant relationships between the country‐level means for this measure and the group‐based emotions that participants reported. When the country mean on individualism was higher, participants reported less guilt and regret for harm inflicted by their family, community, and country, and these coefficients were also significantly different from the coefficients for harm inflicted by themselves (for regret: x 2s > 7.74, ps < .01, for guilt: x 2s > 11.01, ps < .01). For shame, however, no such relationships were found (at the .05 level). For the country means on collectivism, the pattern is also different from the models in which we focused on individual mean scores. Here, almost no relationships (at the .05 level) were found for guilt and shame (with the exception of guilt following harm inflicted by the community), and negative relationships were found with the extent to which participants reported feelings of regret for wrongdoing by their country, community, and family (γ = −0.52, p < .001, 95% CI [−0.73, −0.32], γ = −0.32, p < .05, 95% CI [−60, −0.05], γ = ‐0.34, p < .05, 95% CI [−0.61, −0.07], respectively).

DISCUSSION

In this article, we set out to understand whether the experience of group‐based shame, guilt, and regret following wrongdoing differs across cultures and whether any variation in this regard can be explained by the extent to which people endorse more collectivistic or individualistic values, or live in a cultural setting where such values are more salient. Overall, our findings suggest that the reported experience of these group‐based emotions does vary to some extent across individuals and countries, which can partially be explained by the endorsement of collectivistic as well as individualistic values.

Although there has been some debate about the extent to which self‐critical emotions such as shame and guilt occur (e.g., Leach et al., 2013), we found that people do report shame and, to a lesser extent, guilt and regret following wrongdoings by their group. This does, however, vary depending on which group committed the transgression as well as individual and cultural level differences. It is important to note that most of the variation observed can be attributed to the different transgression levels (i.e., country, community, family, and personal), particularly for guilt and regret. Nonetheless, we do observe some variation in the experience of these three group‐based emotions across countries. In Indonesia, Japan, and Jordan, for example, people reported similar levels of shame for personal transgressions and transgressions by the group. In the other countries, shame for harm committed by the country or community was less than personal shame. The pattern for group‐based guilt and regret across countries was more consistent, as participants across the eight countries were less likely to report group‐based than personal guilt. This was also mostly true for regret, except in the case of Poland where people reported similar levels of regret for both personal and country transgressions. Overall, these findings suggest that there may be more variation in people's emotional responses to wrongdoing than previous research suggests, and that group‐based shame in particular may be more prevalent than previously assumed.

Our results indicate that part of this variation can be explained by the extent to which people themselves endorse collectivistic values, and the extent to which such values, but also individualistic values, are endorsed at the cultural level. Interestingly, the pattern of results at the two levels—individual and cultural—are distinct, and can to a certain extent be described as inverse. While the results are partially in line with our expectations, they also suggest that the interplay between these cultural values and the experience of these (group‐based) emotions is more nuanced than assumed. Crucially, our results highlight the importance of distinguishing between which values (more collectivistic or individualistic) are endorsed, at what level (individual or cultural) they are endorsed, but also between the different emotions in question.

At both the individual and cultural level, we expected that collectivistic values could either be positively or negatively related to group‐based shame, guilt, and regret. At the individual level, we find support for the former of these two competing expectations: despite their aversive nature, people who endorse collectivistic values are more likely to report these group‐based emotions. At first glance, this suggests that—as assumed—people who strongly endorse collectivistic values do exhibit similar tendencies to those exhibited by people who identify more strongly with their group, especially given that we also found positive relationships between our identification measure and all three group‐based emotions. Nevertheless, we also found positive relationships with shame, guilt and regret following personal transgressions. This suggests that people's endorsement of collectivistic values positively relates to these emotions regardless of whether they are triggered by individual or group actions. So while people who value strong and cohesive social bonds may be more likely to report group‐based shame, guilt, and regret following transgressions committed by related others because they are more group‐oriented, it is also possible that we observe this positive relationship because they value the shared potential of these three emotions to help restore and repair social relationships (e.g., A. H. Fischer & Manstead, 2016; Young et al., 2021).

We had expected that the shared potential of these three emotions to motivate reparative actions may result in group‐based shame, guilt, and regret being encouraged at the cultural level because they are in line with the cultural mandate on maintaining and strengthening social relationships. At this level, however, we find no support for this assumption. Instead, we find some support for the second of our two competing expectations—namely that group‐based regret is suppressed within collectivistic cultural contexts. We find almost no relationships, however, with either group‐based shame or guilt.

We can only speculate as to why we observe this difference between the levels and the emotions. One possible explanation is that the nature of these group‐based emotions as either more self/ingroup focused (shame and guilt) or more other/outgroup focused (regret) may be amplified at the cultural level and hence influence which emotions are discouraged within certain cultural contexts. Although all three emotions may trigger reparative actions, in the case of shame and guilt, these actions may be more geared towards alleviating ingroup distress or improving ingroup reputation (e.g., Brown et al., 2008; Iyer et al., 2003) rather than improving relations with the outgroup. In contrast, group‐based regret may be more similar to empathy in that it focuses more on the negative outcome for the other (Imhoff et al., 2012). Even though our scenario did not use a specific “outgroup,” it did include an abstract “others” as a victim group, while the transgressors were presented as “your” family, community, or country. Hence, it is likely that this was perceived as an ingroup violation against an unknown outgroup. The pattern of results suggests that if the focus of a group‐based emotion, like group‐based regret, is towards the outgroup, it may be suppressed within more collectivistic settings.

When it comes to the endorsement of individualistic values, we expected that people who more strongly adhere to these values would report less group‐based shame, guilt, and regret. This expectation did not hold. Given that we did find positive relationships with ingroup identification, these findings suggest that the personal endorsement of individualistic values may not affect how people appraise situations concerning the various groups with which they identify. We did, however, find partial support for our expectation that these group‐based emotions are less likely within more individualistic cultural contexts. This finding potentially provides further support for the argument that the relative self or other focus of these emotional responses may be amplified at the cultural level, with these group‐based emotions being suppressed to reinforce the cultural mandate. Thus, given that people within more individualistic cultural contexts are encouraged to see themselves as independent, they may also be less likely to appraise wrongdoing from a group perspective and more likely to emphasize individual accountability (e.g., Hornsey et al., 2006; Jetten et al., 2002). This may also explain why we see negative relationships with group‐based guilt and regret specifically, as they—unlike shame—are likely to be triggered by a sense of shared responsibility (e.g., Branscombe et al., 2016; Komiya et al., 2011). Here too, we can only speculate about possible explanations as we did not assess the effect of perceived collective responsibility or other possible antecedents such as concern for the ingroup's reputation or moral failure. Future research should try to unpack these relationships further to examine how these cultural values affect the appraisal process of these individual group‐based emotions.

It is also difficult to draw any strong conclusions at the cultural level. Even though we included a uniquely diverse set of countries, the small number has implications for the generalizability of the findings and conclusions drawn. To increase the robustness of these findings, future research should consider examining these relationships across a larger but equally diverse sample. Simultaneously, given the diversity of the sample, we needed to be mindful that our measures allowed for cross‐sample comparison. Using an abstract scenario of harm facilitated cross‐country as well as cross‐level comparability, and allowed us to circumvent any issues that could arise from using actual situations such as conflicting narratives. It would have been helpful, however, if we had asked participants to note down any real‐life situations that may have come to mind, so we could have monitored any possible confounding effects of severity, timing and direct impact (e.g., Gausel & Brown, 2012; Iyer & Leach, 2009). Additionally, given that we used community samples which included participants with no or limited experience with surveys, we applied an abridged version of an individualism and collectivism scale as well as single item emotion intensity measures. Although this has allowed us to present concrete evidence of cross‐cultural differences in group‐based shame, guilt, and regret, more can be done to unpack the underlying emotion‐specific processes.

As Wierzbicka notes, “emotion words” may differ to some degree in meaning across languages as they “reflect, and pass on, certain cultural models” (Wierzbicka, 1999, p. 32). Future studies could hence benefit from applying a multi‐componential approach to help overcome possible concerns that emotion terms are not always directly translatable (see also Boiger et al., 2018). Given that we found that the correlations between shame, guilt, and regret varied substantially across countries and across transgression levels, it is likely that these different correlations reflect underlying linguistic differences between the emotion terms used. That said, they are also likely to reflect an interplay between the context (i.e., the different transgression levels) and the emotional experience. Given our approach, it is not possible to disaggregate these effects, but a multi‐componential approach could enable researchers to be more mindful of, and explicitly assess, context‐dependent (e.g., group‐specific) variability, also in terms of the relationships between the emotions. Additionally, it may also clarify further how cultural dimensions of group orientation affect how people differentially appraise situations that trigger group‐based shame, guilt, and regret. This can be facilitated by a more thorough assessment of how measures of individualism and collectivism, for example, map onto and reflect dimensions of ingroup identification as identified by other researchers (e.g., Leach et al., 2008; Roccas & Schwartz, 2008). Understanding these nuances opens up new avenues for research, including a closer examination of how cross‐cultural differences can help explain when group‐based shame, guilt, and regret may increase support for various reparative actions in the wake of wrongdoings.

FUNDING STATEMENT

This study was funded by a European Research Council Consolidator Grant awarded to Juliette Schaafsma (reference: 682077‐APOLOGY) under the Horizon 2020 research and innovation program.

CONFLICT OF INTEREST

The authors declared no potential conflicts of interest.

ETHICS STATEMENT

Overarching ethical approval for this study was obtained from the Tilburg School of Humanities and Digital Sciences Research Ethics and Data Management Committee. Country‐specific ethical approvals were also obtained from the Scientific Ethics Committee at the University of Costa Rica, the Research Ethics Committee at Atma Jaya Catholic University, the Ethical Committee for Experimental Research Involving Human Subjects at the Graduate School of Humanities at Kobe University, the Research Ethics Committee at the Department of Psychology at the University of Warsaw and the University of Alabama at Birmingham Institutional Review Board. All data collection and processing were conducted in accordance with the ethical codes of conduct of the European Charter for Researchers, the Netherlands Code of Conduct for Scientific Practice (2012), the Ethical Principles and Code of Conduct of the American Psychological Association (APA, 2012), and the Code of Ethics of the American Anthropological Association (AAA, 1998). All participants in this study gave consent before participating.

Supporting information

Click here for additional data file.

Supporting information

Click here for additional data file.