Tempted to join in or not? Moral temptation and self-reported behaviour in bullying situations

Abstract

We investigate the relationship between adolescents’ construction of a transgression relating to a hypothetical temptation and bystander behaviour and bullying (offline and online). A total of 331 Swiss eighth graders completed an electronic questionnaire on bystanding, bullying, moral disengagement, and empathy. Moral functioning was assessed in a hypothetical scenario, using different moral judgements (deontic and self-judgement, judging the transgression; paper-and-pencil measure). Cluster analyses were used to identify patterns of moral functioning. For the open situation (deontic and self-judgement), happy transgressors, happy moralists, ashamed moralists, and indifferent moralists were differentiated, and for the transgression (accomplished deed) moralists and happy opportunists. The analyses yielded significant differences between the different cluster groups. Happy transgressors (open situation) reported higher levels of assisting the bullying than unconcerned moralists. Happy transgressors also reported lower levels of helping than ashamed and happy moralists. Opportunists (accomplished deed) reported higher levels of assisting the bullying, offline bullying, and lower levels of helping the victim. The multivariate GEE analyses showed that happy transgressors reported higher levels of assisting the bully and online bullying than the moralist groups (open situation). The study shows that adolescents who construct a favourable interpretation of yielding to temptation in a hypothetical scenario displayed higher levels of both assisting the bully and online bullying, emphasizing the need for incorporating targeted moral education in bullying prevention.

Statement of contribution

What is already known on this subject?

• Youths’ moral development has been shown to play an important role in explaining involvement in bullying.
• Ascribing positive emotions to a hypothetical moral transgressor is associated with increased levels of bullying.
• Only proactive rule transgressions have been studied, conceptually relating to the role of the ringleader bully.

What this study adds?

• We used a passive moral temptation scenario, conceptually relating to the bystander role. In passive moral temptations, the right way to act is not obvious, and the protagonist has no initial intention to transgress.
• Positive views of yielding to temptation in the scenario were related to supporting the bully and online bullying.
• There is a need for incorporating targeted moral education for bystanders displaying pro-bully behaviour that addresses situations where the boundaries of morality are fuzzy.

Background

Although researchers became aware of the moral dimensions of bullying in the early 2000s (e.g., Hymel, Rocke-Henderson, & Bonanno, 2005; Menesini et al., 2003), the bulk of bullying research as well as prevention and intervention does not systematically include moral development and functioning. Addressing the moral dimensions of bullying presupposes that bullying is seen as relating to the moral domain, that is, (non-)behaviour affecting the rights and welfare of others (e.g., Smetana, 2006). Morality refers to the ‘right and good’, requiring humans to show benevolence and kindness towards others (Gibbs, 2003). The aim of that benevolence is not to harm others’ welfare, but to protect or restore it (Gutzwiller-Helfenfinger, 2018). Against this background, we characterize bullying as an aggressive behaviour that conflicts with individual and social moral standards, because aggressive acts against the victim are intentional and often include humiliating elements (Perren, Gutzwiller-Helfenfinger, Malti, & Hymel, 2012). This applies to both offline (defined as a particular form of repeated aggressive behaviour against a defenceless victim; Olweus, 1993) and online bullying (defined as bullying involving the use of electronic forms of communication; Slonje & Smith, 2008).

Bullying does not unfold in social isolation, nor does it involve only the bully (as the initiator of the aggression) and the victim (as the recipient of the aggression). In the majority of incidents, bystanders are present as witnesses who observe what is happening (Cornell, Shukla, & Konold, 2015; O’Connell, Pepler, & Craig, 1999; Rodkin, Espelage, & Hanish, 2015). Observers can assume various participant roles (Salmivalli, Lagerspetz, Björkqvist, Österman, & Kaukiainen, 1996): Assistants join the bully and actively engage in the aggression; reinforcers support the bully by cheering and laughing, drawing attention to what is happening; defenders try to help the victim and offer support; and outsiders remain passive or unconcerned and sometimes also go away. A growing body of research literature shows that bystander behaviour impacts the occurrence and maintenance of both offline (e.g., Salmivalli, Voeten, & Poskiparta, 2011) and online bullying (e.g., Bastiaensens et al., 2016). Several individual and group characteristics have been identified as factors influencing (offline) bystander behaviour. Among them are empathy (Barchia & Bussey, 2011), group norms (Salmivalli & Voeten, 2004), moral disengagement (Obermann, 2011), basic moral sensitivity (Thornberg & Jungert, 2013), moral emotions (Menesini, Nocentini, & Camodeca, 2013), or attitudes towards bullying (Pozzoli, Gini, & Vieno, 2012). Empathy, moral emotions, moral disengagement, and basic moral sensitivity can be viewed as core aspects (and processes) of moral functioning and are relevant not only for bystander behaviour but also for proactive bullying behaviour (cf. Hymel, Schonert-Reichl, Bonanno, Vaillancourt, & Henderson, 2010).

That moral development and functioning are systematically related to bullying (offline and online) and bystander behaviour in children and adolescents (both cross-sectionally and longitudinally) has been established in numerous studies during the last one-and-a-half decades (see meta-analyses: Gini, Pozzoli, & Hymel, 2014; Killer, Bussey, Hawes, & Hunt, 2019). However, only few studies addressed the interplay between cognitive and emotional (affective) moral processes in the context of a more integrative framework (Conway, Gomez-Garibello, Talwar, & Shariff, 2016; Perren & Gutzwiller-Helfenfinger, 2012). Moreover, most studies used self-report scales to assess moral development and functioning, for example in the context of research on moral disengagement and bullying (Killer et al., 2019). Some studies also used qualitative measures tapping into children’s and adolescents’ conceptions and competencies (Menesini et al., 2003; Perren & Gutzwiller-Helfenfinger, 2012; Perren et al., 2012; Thornberg, 2010, 2015). When qualitative measures included scenarios or vignettes of moral rule transgressions, only proactive transgressions (like, e.g., deliberately and openly pushing another child from the swing to have it to oneself) were thematized (e.g., Conway et al., 2016; Perren & Gutzwiller-Helfenfinger, 2012) neglecting temptation situations. In temptation situations, an opportunity to break a moral rule for personal gain without being observed offers itself. When an individual notices this opportunity, s/he may form the intention to break the rule and then actively translate this intention into behaviour as soon as circumstances are favourable. Thus, the act is the result of premeditation. For example, a child may be in a cloakroom and see that there is some candy in the coat pocket of one of his/her classmates. When the child feels unobserved, s/he steals the candy (Nunner-Winkler & Sodian, 1988).

However, there are temptation situations where an individual does not intend to break a moral rule, and the opportunity to do so imposes itself without the individual’s contribution. For example, a teenager may have a meal in a restaurant and, when paying, get too much return money. The waiter does not realize this, so the teenager is literally thrown into the temptation situation. Hence, if s/he keeps the excess money, it is not the result of an intentional, deliberate act planned in advance. Such so-called passive moral temptations (Heinrichs, Gutzwiller-Helfenfinger, Latzko, Minnameier, & Döring, 2020) have hardly ever been investigated. The present study explores the relationship between moral functioning, specifically participants’ construction of a passive moral temptation situation (judgement and emotion expectation) and bystander behaviour (assisting the bully, ignoring, helping the victim) and bullying (offline and online). Participants’ constructions of the passive moral temptation were assessed in the context of a hypothetical scenario, while a self-report questionnaire was used for both bullying and bystanding.

Moral functioning in a moral temptation situation

Much of the research addressing the moral dimensions of bullying can be grouped into three larger conceptual frameworks. The Moral Disengagement Framework draws on Bandura’s socio-cognitive theory (e.g., Bandura, 2002) and addresses the role of cognitive moral processes in the formation and maintenance of bullying on the individual, group, and contextual level (Gini, Pozzoli, & Bussey, 2015; Menesini, Palladino, & Nocentini, 2015; Obermann, 2011). In this framework, mostly self-report scales are used to assess moral functioning. The Happy Victimizer Framework is based on research within the Happy Victimizer Paradigm (Nunner-Winkler & Sodian, 1988) investigating the development of the relationship between moral judgement and moral emotion attributions (Conway et al., 2016; Kollerová, Janošová, & ŘíČan, 2015; Perren et al., 2012). The measurement approach used consists of open-ended vignettes or scenarios of hypothetical situations, mostly in the context of oral interviews. In the Integrative Framework, moral development and functioning are investigated by assessing multiple indicators (disengagement, values, judgement, emotions, responsibility, etc.) using a combination of self-report scales and scenarios or vignettes (Perren & Gutzwiller-Helfenfinger, 2012; von Grundherr, Geisler, Stoiber, & Schäfer, 2017).

However, studies involving scenarios or vignettes of moral or morally relevant situations have used mostly proactive, obvious transgressions of moral rules like stealing, hitting, excluding someone, etc., to assess children’s (and more recently adolescents’ and young adults’) moral reasoning (e.g., Gasser & Keller, 2009; Nunner-Winkler, 2007; Perren & Gutzwiller-Helfenfinger, 2012). In these situations, the protagonist intentionally harms others to reach his/her goals, suggesting conceptual proximity to the behaviour of ringleader bullies. In case temptation situations were used, they still included a proactive, intentional rule transgression like for example stealing candy from a peer’s cloak when unobserved (Nunner-Winkler & Sodian, 1988). Thus, only very recently have passive moral temptations where a protagonist does not intend to transgress and only realizes that s/he might do so as a result of specific circumstances like receiving too much change money when buying something (Heinrichs, Minnameier, Gutzwiller-Helfenfinger, & Latzko, 2015) been included in moral developmental research. To the best of our knowledge, no research investigating the relationship between moral functioning in the context of a passive moral temptation and bullying has been undertaken so far, marking this as a gap in the research literature.

Why might children’s and adolescents’ moral functioning in a passive moral temptation scenario be relevant to better understand the moral dimensions of bullying (offline and online) and bystanding? The answer to this question lies in the core features of such situations. The first feature refers to intentionality: In a passive moral temptation, a person is ‘thrown’ into a situation where s/he realizes that s/he might transgress a moral rule without ever having planned or intended to (Heinrichs et al., 2020), like for example getting too much change when buying something. As intentionality is one of the touchstones of harmful behaviour and is used to assess the degree of badness of a certain act (Guglielmo & Malle, 2010), and as even young children are aware that intentional harm is worse than unintentional or accidental harm (Nobes, Panagiotaki, & Pawson, 2009), finding oneself in such a situation makes it easier to transgress because one is innocent in the sense of being free from any premeditation. This effect is increased by the high affordance of the situation, a second feature of passive moral temptations: In the case of getting too much change, the protagonist already holds the money in his/her hand or sees it lying on the counter. The change was literally handed over by the shop assistant, indicating that it was now the protagonist’s to keep. Thus, the temptation is forced upon the protagonist. The potential reward of yielding to the temptation lies in the material gain.

A third feature refers to the strength of the moral norm inherent in the situation, that is, whether a negative or a positive duty is involved (e.g., Belliotti, 1981). Getting too much change money, the situation used in our study, differs considerably from proactively stealing money from someone. The former situation is more ambiguous and open, particularly because – unlike the latter – it is not directly related to a negative duty, here ‘thou shalt not steal’. At best, it may be linked to positive duties like ‘you should help someone in need’; and even then, it is not immediately obvious that a shop assistant ending up with a negative balance can be seen as a person in need. As negative duties carry a stronger moral obligation than positive duties (Belliotti, 1981), the relation to a positive, weaker duty increases the ambiguity of the situation. Therefore, the moral content of the situation is not as obvious.

A fourth feature that also contributes to the ambiguity of passive moral temptations refers to responsibility. The protagonist did not create or cause the moral conflict. In our example, it is actually the shop assistant who did not count out the change money correctly and thus caused the moral conflict (keeping the money versus returning it) in the first place. As notions of agency and related responsibility are inherent in basic conceptions of morality (e.g., Bandura, 2002), situations where a protagonist does not proactively cause harm or contribute to creating a conflict are less clear-cut and more open to differential interpretations. Further support for this assumption is offered by research on the so-called action principle (also referred to as omission bias or actor effect) (Hayashi, 2015; Krettenauer & Jia, 2013; Krettenauer, Jia, & Mosleh, 2011) stating that ‘harm caused by action is worse than equivalent harm caused by inaction’. Therefore, actively stealing would be considered worse and thus more blameworthy than not returning change money a waiter mistakenly gave out. Interestingly, passive moral temptations include an additional facet: Harm caused by inaction refers to situations like letting someone drown or not helping someone who took a fall. In such situations, a person’s plight is extant and can be alleviated by action. In passive moral temptations, no one is in need at the outset. Things hang in the balance as long as the protagonist has not yet decided whether or not to act prosocially, that is, return the excess change or not. The plight is only the result of not conforming with the prosocial norm. Thus, keeping the excess change means that money will be missing in the register.

Taken together, these features render passive moral temptation situations ambiguous, equivocal, and more difficult to identify as moral or morally relevant situations relating to moral duties, norms, and values. They are less easy to read in the sense of knowing exactly what would be the morally appropriate course of action or how bad not following the positive obligation would be. Situations like these refer to what we call the ‘fuzzy boundaries of morality’ and promise to help shed additional light on individuals’ inner moral landscapes. The few existing studies (all involving adults) so far have consistently shown that a considerable proportion of participants, while judging the act as wrong, attributed mixed (positive and negative) or positive feelings to protagonists who yielded to the temptation and did not follow the positive moral obligation (Heinrichs et al., 2015, 2020). Also, about half of the participants in those studies judged the transgression (keeping the excess change) as right. This is in contrast with findings from a myriad of studies using proactive transgressions indicating that from about age 9 on, individuals judge moral transgressions as wrong and attribute negative feelings to the rule transgressor (see, e.g., the meta-analysis by Malti & Krettenauer, 2013).

At this point, we can establish a link with bullying: Those children and adolescents who do not start the bullying (as ringleaders) are faced with the decision how to react: assist the bully, support the victim, or remain ‘neutral’ by watching or walking away. Regarding an individual incident, the situation may present itself as follows to them: They did not start the bullying and thus were not proactive agents, nor were they (at least directly) responsible for the onset of the negative acts. They did not intend to harm someone, but were thrown into the situation. At the same time, several behavioural choices offer themselves: join in the bullying, help the victim, or remain passive (e.g., Doramajian & Bukowski, 2015). Helping the victim would include following the weaker positive duty (or prosocial norm) and be indicative of (among other characteristics) empathy, moral courage, and defender self-efficacy (cf. Barchia & Bussey, 2011). Doing nothing helps the individual to stay outside and at least not do anything to actively harm the victim. This passivity might reflect insecurity, the need for self-protection, or a lack of effective, non-aggressive intervention strategies (cf. Thornberg & Jungert, 2013). Like defenders of victims, passive bystanders are low in aggression but are able to prevent victimization towards themselves (Camodeca & Goossens, 2005). Supporting the bully would mean to yield to the temptation in a situation where moral agency and responsibility are weakened because one did not start the bullying. Such a temptation would be attractive for those individuals who are enough motivated to act aggressively towards others but not quite enough to start the bullying. One reward would be a gain in social status. Recent research indicates that supporters of bullies display higher levels of aggression than victims, defenders of the victim, and outsiders (i.e., non-involved, passive bystanders) and share bullies’ high social status (Pouwels, Lansu, & Cillessen, 2016). Accordingly, part of the basic configuration of the situation is similar with that involved in passive moral temptations: the chance of yielding to a temptation (and therefore breaking/not following a moral rule) under favourable circumstances, thereby gaining a (material or immaterial) reward. This conceptual similarity suggests a potential link between individuals’ construction of passive moral temptations and their reactions as witnesses of bullying. Moreover, as aggressive children tend to shift between aggressive roles, that is, sometimes act as bullies, supporters, or reinforcers (Oh & Hazler, 2009), we may also assume a systematic link between their aggressive behaviour and their constructions of passive moral temptations.

Exploring participants’ constructions

When investigating individuals’ moral reasoning, research following or extending the Kohlbergian tradition has often made use of hypothetical scenarios or vignettes followed by a set of (half-)standardized questions requiring participants to decide what the protagonist should do (deontic judgement) and why (justification). Within the Happy Victimizer tradition, scenario measures often introduce a transgression as ‘fait accompli’ and do not leave the situation open initially. Participants are asked to judge whether the transgression is okay/good/right or not okay/bad/wrong, to justify this moral judgement, to indicate how the perpetrator feels (emotion expectation), and to justify this emotion attribution (e.g., Malti & Keller, 2009; Nunner-Winkler & Sodian, 1988). As the transgression has already occurred, a deontic judgement asking what the protagonist should do cannot be made. However, the inclusion of such a deontic judgement makes it possible to assess participants’ initial constructions of a given moral or morally relevant situation. In this way, researchers can gain insight into the features of the situation that are both salient and relevant to participants (cf. Wainryb, Brehl, & Matwin, 2005). Still, asking for a deontic judgement (‘What should X [the protagonist] do’) does not necessarily represent the course of action chosen for oneself. Therefore, it is necessary to include both a deontic judgement and own action decision in addition to a judgement of the hypothetical transgression itself (‘X kept the money. Is it okay or not to keep the money’) to have a fuller representation of participants’ construction of the situation (see also the self-as-perpetrator vs. other-as-perpetrator distinction as made, e.g., in the study by Keller, Lourenco, Malti, & Saalbach, 2003). The varying judgement conditions can be understood as situational variations (cf. Nunner-Winkler, 2013) with consistency between judgements serving as an indicator of a given moral orientation. Accordingly, to have a fuller understanding of adolescents’ moral meaning-making of this passive moral temptation situation, we need to include all three moral judgements (deontic, self, transgression) including justifications, emotion attributions to the perpetrator, and justification of these emotion expectancies.

The present study

The current study aims to investigate the relationship between moral functioning, specifically participants’ construction of a transgression (judgement and emotion expectation [happy vs. unhappy]) and bullying (offline, online) and bystanding (offline) in adolescents. Moral functioning was assessed using three moral judgements (deontic judgement, self-judgement, judging the transgression) in the context of a passive moral temptation scenario. Passive moral temptations mean that the protagonist is ‘thrown’ into a situation, that is, does not proactively intend to transgress. The scenario is about getting too much change money when buying a bicycle light (thus, the question is: keep it or give it back). The fact that no negative duty (like ‘not harming’, implying a strong moral obligation) but a positive duty (like ‘helping’, carrying a weaker moral obligation) is included in this kind of scenarios means that the situation is more open, ambiguous, and thus less morally salient. Therefore, we test the ‘boundaries’ of teenagers’ morality and link it to bullying and bystanding. We expect systematic relationships between the quality of participants’ construction of the transgression and bullying and particularly bystanding (supporting the bully, helping the victim, ignoring). Specifically, we hypothesize that participants who construct a favourable understanding of transgressing in this hypothetical passive moral temptation scenario (high level of transgression-favouring judgements in conjunction with positive emotions) will report increased levels of bullying (offline and online) and of supporting the bully and low levels of supporting the victim. Also, participants who construct an unfavourable understanding of the transgression (low level of transgression-favouring judgements in conjunction with negative emotions) will report low levels of bullying and of supporting the bully and high levels of supporting the victim.

Previous studies involving adolescents have indicated that moral disengagement is strongly associated with bullying behaviour (e.g., Gini et al., 2014); that empathy is associated with both bullying (e.g., Caravita, Di Blasio, & Salmivalli, 2009) and bystander behaviour (e.g., Gini, Albiero, Benelli, & Altoè, 2008); and that sex is also associated with both dimensions (e.g., Gini et al., 2008). Therefore, we included these variables as controls in the multivariate analyses.

Method

Procedure

The present paper includes data from a longitudinal study conducted in Switzerland (netTEEN). Four assessments were carried out between November 2010 and May 2012 with time intervals of 6 months. For the current study, data from the fourth assessment wave are used. As required by Swiss legislation, permission to conduct the study was obtained from the respective school councils. School directors and teachers from the selected schools volunteered, and parents were told about the study and asked to inform teachers if they did not want their children to participate (passive consent). The parents of four adolescents refused their child’s participation at each assessment. Participants were informed about the survey’s procedure and goal and were given the opportunity to refrain from participation with no negative consequences (informed oral consent). Students who did not want to participate were offered another activity during the relevant school period.

An electronic self-report questionnaire was administered in classrooms on netbooks. The moral temptation measure was distributed as a paper-and-pencil questionnaire.

Sample

Three of the 26 Swiss cantons (member states of the Swiss Federal State) were selected for study participation. In each of the cantons, four schools with at least three classrooms were randomly selected. Each school was represented in the present study by three to four classrooms. Only participants from the two cantons from the German-speaking part of Switzerland were given the moral temptation measure (29 classrooms). In total, 462 students participated at the T4-assessment in the German-speaking schools. 51% of participants were female. The fourth assessment was conducted at the end of grade eight; students’ mean age was 14.9 years (SD = 0.51). All students who completed the computer-assisted questionnaire before the scheduled time received the moral temptation measure as paper-and-pencil version. A total of 331 students completed this measure (72% of all eligible students) and are considered as the sample for this study. However, 26 students, despite having valid data for the moral temptation (MT) measure, had either missing or unidentifiable identifications on the paper-and-pencil document (N = 15) or missing scores in all relevant questionnaire scales (N = 11).

The drop-out analysis showed significant differences between those who completed or did not complete the moral temptation measure. Students who did not complete the MT measure reported significantly higher scores of online and offline bullying, assisting the bully, and moral disengagement than students who completed the measure (F = 6.13–11.8; all p < .05). A further drop-out analysis with later cluster memberships (see result section) yielded significant associations between the pattern of missingness and cluster membership. Students who completed the MT measure but failed to indicate valid questionnaire data were more frequently categorized as happy transgressors or happy opportunists than expected (χ² = 12.2–13.7; both p < .05).

Assessment of moral functioning

To establish participants’ moral judgements and emotion expectations, a newly designed passive moral temptation vignette was used. The story is about getting too much change money when buying a bicycle light (thus, the question is: keep it or return it). The shop assistant is the protagonist’s classmate. Both a female and a male version were constructed to match participants’ gender. To have a fuller representation of participants’ construction of the situation, we included both a deontic judgement and own action decision in addition to a judgement of the hypothetical transgression itself (‘X kept the money. Is it okay or not to keep the money?’). The text of the vignette and the respective questions can be found in the Appendix. Basically, getting too much change money is an ordinary, everyday situation and something practically everyone experiences at least once in their lives. In an associated qualitative study involving in-depth interviews, we found that all of six adolescents aged 12–16 had experienced such a situation themselves (Amrein, 2016). Accordingly, the vignette can be assumed to have high ecological validity.

Participants had to make three different moral judgements: (1) a deontic judgement on what the protagonist should do (keep the money, return the money; deontic), justify their judgement, attribute (an) emotion(s) to the protagonist, and justify their emotion expectation(s); (2) indicate what they themselves would do in the given situation (keep the money, return the money; self-judgement), justify their decision, attribute (an) emotion(s) to themselves, and justify their emotion expectation(s); and (3) after being told that the protagonist had transgressed the moral rule (i.e., kept the excess change) to judge the transgression (okay or not okay), justify their judgement, attribute (an) emotion(s) to the protagonist, and justify their emotion expectation(s) (‘classical’ Happy Victimizer condition). The first two moral judgements (1 and 2) are combined for the current paper and are considered as ‘open situation’. For the third judgment, two follow-up questions were asked (in counterbalanced order), again requiring participants to judge the ‘new’ situation and attribute (an) emotion(s): (4) The protagonist did not get a discount although all classmates got one; and (5) the protagonist learns that the shop assistant has to pay any missing sum out of his/her own pocket. All three answers were combined for the current paper and are considered as ‘accomplished deed’ (see also Appendix).

Emotion expectancies consisted of a response scale (happy, proud, indifferent, sad, angry, anxious, ashamed), with students marking the response(s) they deemed appropriate. For the current study, participants’ moral justifications were not included.

Assessment of bystander behaviour

The students were asked about their behaviour in a bullying situation using an adapted version of the participant roles scale (Sutton & Smith, 1999). The 9-item questionnaire consists of three subscales: helping the victim, ignoring, assisting the bully. Possible responses ranged from one (never) to five (always). The helping the victim subscale (pro-victim attitudes) consists of three items (e.g., I try to help the victim) and has a high internal consistency (Cronbach α = .80). The assisting the bully subscale consists of three items (e.g., I find it funny and I laugh with them) with a high internal consistency (α = .82). The ignoring bullying (passive) subscale consists of three items (e.g., I pretend that I did not see anything) with a high internal consistency (α = .81).

Assessment of bullying behaviour

Offline bullying was assessed using an adapted version of a validated offline bullying scale (Alsaker, 2003). This scale was introduced by a definition of bullying (based on Olweus, 1993) and consisted of six items encompassing a set of different aggressive behaviours (e.g., laughing at people, insulting, excluding, or hitting someone). At each assessment, participants were asked how often they had performed these behaviours during the past 4 months. Participants rated each item from one (never) to five (almost daily). The mean score of the six items was computed to obtain a single score of offline bullying (α = .78). Higher scores indicate more offline bullying.

Online bullying was assessed using a scale developed by Sticca, Ruggieri, Alsaker, and Perren (2013). The scale encompassed a total of six items tapping different forms of online bullying (e.g., sending mean or threatening messages to single individuals, groups, or publishing on the Internet). At each assessment, participants were asked how often they had performed these behaviours during the past 4 months. Possible responses ranged from one (never) to five (almost daily). The mean score of the six items was computed to obtain a single score of online bullying (α = .95). Higher scores indicate more online bullying.

Assessment of moral disengagement and empathy

Moral disengagement was assessed using the Moral Justifications for Bullying (MOJUS) scale (Perren, Rumetsch, Malti, & Gutzwiller-Helfenfinger, 2014). Participants were given two written hypothetical bullying scenarios describing an adolescent excluding or humiliating a peer, respectively. In total, participants were given 11 statements assessing moral disengagement (e.g., ‘This schoolmate deserved it’). Participants were asked whether they agreed with the statements. Responses ranged from 1 (not true) to 4 (true). Scores from the moral disengagement items were averaged to obtain a single moral disengagement score (α = .91). Higher scores indicate higher levels of moral disengagement.

Empathic concern was assessed using four items (e.g., ‘When I see another child who is being picked on, I feel sorry for him or her’) from Malti et al. 2010, (adapted from Zhou, Valiente, & Eisenberg, 2003). Participants were asked whether they agreed with the statements. Responses ranged from 1 (not true) to 4 (true). Scores were averaged to obtain a single score for empathic concern (α = .92). Higher scores indicate higher levels of empathic concern.

Results

Moral functioning: Descriptive results and response patterns

Results of the cluster analyses

Table 1 shows the frequencies of participants’ answers on moral judgement and emotion expectancies. For the subsequent analyses, we counted the frequency of moral transgressions (sum of ‘keep the money’, range 0–2 for the open situation; and sum of ‘ok’, range 0–3 for the accomplished deed situation), the frequency of happy emotions (happy or proud) and the frequency of unhappy emotions (sad, angry, anxious, ashamed). Following a person-centred approach, we conducted a two-step cluster analysis. First, we conducted a cluster analysis for the open situation (Q1–Q2), in which participants could decide on the behaviour. Second, we performed a cluster analysis for the second situation (accomplished deed, Q3–Q5) in which participants had to evaluate the protagonist’s moral transgression in various conditions.

Table 1. Frequencies for answers to Q1–Q5 in the passive moral temptation vignette

	Transgression (keeping the money)			Moral emotions
	Yes (keep)/ok	No (keep)/not ok	Both	Happy	Proud	Indifferent	Sad	Anxious	Angry	Ashamed
Open situation
Q1 Deontic	68	259	4	38.7%	46.5%	16.3%	1.8%	1.2%	1.2%	8.2%
Q2 Self-judgement	79	238	4	38.0%	37.4%	20.8%	1.5%	1.5%	1.2%	10.6%
Accomplished deed
Q3 Transgression	78	249	2	20.0%	8.9%	21.0%	3.3%	12.5%	0.6%	41.3%
Q4 Transgression – Condition 1	103	209	6	13.8%	7.9%	20.7%	12.6%	5.9%	6.2%	45.6%
Q5 Transgression – Condition 2	90	227	1	15.4%	7.0%	22.1%	10.7%	6.6%	9.5%	42.8%

Note

• Emotions attributed most frequently are indicated in bold script.

For the open situation, we conducted a two-step cluster analysis with the number of the clusters to be determined automatically. Log likelihood was used as distance measure. All three variables (number of transgressions, happy emotions, unhappy emotions) were entered as continuous measures. Schwarz’s Bayesian criterion was used (BIC) to establish the number of clusters. The analysis yielded a four-cluster solution. The average Silhouette measure yielded a good cluster quality regarding cohesion and separation (S = 0.7). Cluster centroids of this four-cluster solution can be seen in Table 2. Based on the cluster centroids, we labelled the clusters ‘ashamed moralists’, ‘happy moralists’, ‘indifferent moralists’, and ‘happy transgressors’. All ‘moralist’ groups show a low level of transgression (they would return the money), but ashamed moralists report unhappy emotions (mostly ‘ashamed’), happy moralists report happy emotions, and indifferent moralists very low levels of positive and negative emotions. Happy transgressors report a high level of moral transgression (they would keep the money) and would feel rather happy about it.

Table 2. Moral temptation answers: Final cluster centroids for both situations

	Moral transgression		Happy emotions		Unhappy emotions		N (%)
	Mean	SD	Mean	SD	Mean	SD
Open situation
Co1 (Ashamed moralist)	0.22	0.51	0.52	0.68	1.56	0.50	50 (15.1)
Co2 (Happy moralist)	0.03	0.17	2.00	0.00	0.00	0.00	166 (50.2)
Co3 (Indifferent moralist)	0.19	0.40	0.63	0.49	0.00	0.00	52 (15.7)
Co4 (Happy transgressor)	2.00	0.00	1.22	0.89	0.08	0.27	63 (19)
Accomplished deed
Ca1 (Moralist)	0.17	0.38	0.13	0.37	2.54	0.60	196 (59.2)
Ca2 (Happy Opportunist)	1.85	1.10	1.40	1.16	0.50	0.66	135 (40.8)

Note

• Mean scores are based on the frequency of reported behaviour and emotion categories.

For the accomplished deed situation, a second two-step cluster analysis was computed. The analysis yielded a two-cluster solution with a good cluster quality (S = 0.6). Cluster centroids of this two-cluster solution can be seen in Table 2. Based on the cluster centroids, we labelled the clusters ‘moralists’ and ‘happy opportunists’. Moralists show low levels of transgression (they think it is not ok to keep the money), low levels of happy emotions, and high levels of unhappy emotions (mostly ashamed). Happy opportunists show high levels of transgression and report happy emotions. Additional analyses on the role of sex for moral functioning can be found in Appendix S1.

Overlap between moral functioning patterns

In a next step, we cross-tabulated the cluster memberships. The chi-square tests yielded a significant overlap between both clusters (Chi² = 139.4, df = 3, p<.001). Table 3 shows the observed and expected counts and the standardized residuals. Students with the ‘happy opportunist’ cluster membership in the accomplished deed situation were more frequently moral transgressors or indifferent moralists in the open situation and less frequently than expected ashamed moralists or happy moralists in the open situation.

Table 3. Results of the cross-tabulation between cluster membership in the open situation and the accomplished deed situation

MT Cluster the open situation	MT cluster the accomplished deed situation		Total
	Ca1 (Moralist)	Ca2 (Happy Opportunist)
Co1 (Ashamed moralist)
Count	39.0	11.0	50
Expected count	29.6	20.4
Stand. residual	1.7	−2.1
Co2 (Happy moralist)
Count	138.0	28.0	166
Expected count	98.3	67.7
Stand. residual	4.0	−4.8
Co3 (Indifferent moralist)
Count	15.0	37.0	52
Expected count	30.8	21.2
Stand. residual	−2.8	3.4
Co4 (Happy transgressor)
Count	4.0	59.0	63
Expected count	37.3	25.7
Stand.residual	−5.5	6.6
Count	196	135	331

Associations between moral functioning and bullying/bystander behaviour

Bivariate analyses

First, we computed Pearson correlations between all study variables. For this analysis, moral functioning variables were included as linear variables (see Table 4). Frequency of moral transgression for both the open and the accomplished deed situation was related significantly positively with Assisting the bully, Offline and Online bullying, and Moral Disengagement, and significantly negatively with Helping the victim, Sex, and Empathy. Happy emotions in the open situation were significantly negatively related with Assisting the bully, while happy emotions in the accomplished deed situation were positively related with Assisting the bully, Offline and Online bullying, and Moral Disengagement, and significantly negatively with Helping the victim, Sex, and Empathy. Unhappy emotions in the open situation were significantly positively correlated with Sex and Empathy, and unhappy emotions in the accomplished deed situation were significantly positively related with Helping the victim, Sex, and Empathy, and significantly negatively with Assisting the bully, Online and Offline bullying, and Moral Disengagement.

Table 4. Bivariate associations between moral functioning and bullying/bystander behaviour

	M	SD	Transgress q12	Happy q12	Unhappy q12	Transgress q345	Happy q345	Unhappy q345	Assisting bully	Ignoring	Helping victim	Offline bullying	Online bullying	Sex (female)	Empathy
Transgress q12	0.46	0.80
Happy q12	1.41	0.81	−.16**
Unhappy q12	0.25	0.60	−.08	−.54**
Transgress q345	0.86	1.12	.73**	−.08	−.16**
Happy q345	0.65	1.01	.50**	.11*	−.10	.59**
Unhappy q345	1.70	1.18	−.57**	.14**	.20**	−.65**	−.59**
Assisting bully	1.78	0.87	.30**	−.13*	.02	.34**	.25**	−.34**
Ignoring	2.34	0.97	−.04	−.09	.04	.05	−.02	−.04	.34**
Helping victim	2.59	0.96	−.26**	−.01	.10	−.31**	−.31**	.34**	−.20**	−.19**
Offline bullying	1.23	0.51	.27**	.01	−.07	.27**	.29**	−.25**	.45**	.11**	−.16**
Online bullying	1.20	0.44	.22**	−.05	−.05	.18**	.15**	−.12*	.30**	.12**	−.07*	.69**
Sex (female)	0.51	0.50	−.27**	.04	.16**	−.36**	−.35**	.41**	−.19**	.00	.26**	−.14**	−.08*
Empathy	2.93	0.84	−.38**	−.05	.19**	−.43**	−.39**	.44**	−.30**	−.14**	.47**	−.20**	−.052	.34**
Moral diseng.	1.78	0.71	.24**	−.11	−.11	.26**	.23**	−.32**	.50**	.22**	−.29**	.40**	.27**	−.22**	−.30**

Note

• * p < .05 ** p < .001

Univariate analyses

Next, we conducted several univariate analyses predicting offline bullying, online bullying, and bystander behaviour. Cluster memberships of the open situation and the accomplished deed situation were used as categorical independent variables. To control for the nestedness of the data (students in classrooms), a GEE model (generalized estimating equations) in SPSS was used. The working correlation matrix was set to ‘exchangeable’ to allow for correlations within the same classroom. Pairwise comparisons (with Bonferroni correction for multiple comparisons) were used to establish differences between the different groups.

Cluster memberships of the open situation

The GEE analysis yielded significant overall effects for Assisting the bully, Wald Chi² (df = 3) = 21.57, p < .001; Helping the victim, Wald Chi² = 18.81, p < .001; and Offline bullying, Wald Chi² = 8.29, p = .040. The effect for Online bullying was marginally significant, Wald Chi² = 7.47, p = .058. No significant overall effect emerged for Ignoring: Wald Chi² = 2.26, p = .521.

Pairwise comparisons (see Table 5) indicate that happy transgressors report higher levels of Assisting the bullying than happy moralists. Happy moralists furthermore report lower levels of Assisting than unconcerned moralists. Regarding Helping the victim, the analysis indicated that happy transgressors report lower levels of helping than ashamed and happy moralists. No significant pairwise differences emerged regarding Offline and Online bullying.

Table 5. Differences in bystander and bullying behaviour by cluster memberships

	Assisting the bully		Ignoring		Helping the victim		Offline bullying		Online bullying
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Open situation
Co1 (Ashamed moralist)	1.85	0.88	2.44	1.01	2.84^a	0.86	1.14	0.26	1.39	0.61
Co2 (Happy moralist)	1.54^a,b	0.60	2.29	0.90	2.70^b	0.93	1.14	0.24	1.37	0.61
Co3 (Indifferent moralist)	1.91^a	0.84	2.48	1.08	2.50	0.91	1.28	0.37	1.57	0.87
Co4 (Happy transgressor)	2.33^b	1.23	2.29	1.04	2.12^a,b	1.04	1.53	1.05	1.84	1.24
Accomplished deed
Ca1 (Moralist)	1.59^d	0.66	2.30	0.88	2.82^d	0.85	1.13^d	0.22	1.41	0.64
Ca2 (Happy Opportunist)	2.10^d	1.06	2.41	1.10	2.20^d	1.01	1.38^d	0.75	1.59	1.02

Note

• Means with the same superscripts are significantly different from each other (within a column).

Cluster memberships of the accomplished deed situation

The GEE analysis yielded significant overall effects for Assisting the bully, Wald Chi² = 22.73, p < .001; Helping the victim, Wald Chi² = 24.15, p < .001; and Offline Bullying, Wald Chi² = 6.15, p = .013. The effect for Online Bullying was marginally significant, Wald Chi² = 7.47, p = .058. No significant overall effect emerged for Online Bullying, Wald Chi² = 2.06, p = .151; and Ignoring, Wald Chi² = 0.76, p = .384.

Happy opportunist report higher levels of assisting the bullying and offline bullying and lower levels of helping the victim than moralists.

Multivariate analyses

To compare the relative strengths of different aspects of participants’ moral functioning, a general linear model was computed. Cluster memberships of the open situation and the accomplished deed situation were used as categorical independent variables. The ‘moral transgressor’ or ‘happy opportunist’ cluster was set as reference group, respectively. Sex, moral disengagement, and empathy were used as covariates. Bystander behaviour (helping the victim, ignoring, assisting the bully) and bullying behaviour (offline and online) were used as (separate) dependent variables. To control for the nestedness of the data (students in classrooms), a GEE model (see above) was used.

The GEE analysis for ‘Assisting the bully’ yielded significant multivariate effects for Cluster membership in the open situation (Wald Chi² = 8.885, p = .031), but not for Cluster membership in the accomplished deed situation. The parameter estimates are shown in Table 6. Happy transgressors indicated a higher frequency of assisting the bully than happy moralists and indifferent moralists. The difference between happy transgressors and ashamed moralists was only marginally significant. In addition, empathy and moral disengagement were significant predictors. The higher the level of empathy and the lower the level of moral disengagement, the lower the frequency of assisting the bully.

Table 6. Results of the GEE analysis predicting Bystander Behaviour

Dependent variable	Assisting the bully				Ignoring				Helping the victim
	B	SE	Wald Chi^2	p	B	SE	Wald Chi^2	p	B	SE	Wald Chi^2	p
Intercept	2.32	.33	49.0	.000	2.34	.30	62.2	.000	1.33	.29	21.6	.000
MT_Co1 (Ashamed moralist)^a	−0.36	.21	3.0	.084	0.43	.28	2.4	.120	−0.13	.20	0.4	.529
MT_Co2 (Happy moralist)^a	−0.62	.23	7.2	.007	0.26	.21	1.5	.217	−0.15	.18	0.7	.405
MT_Co3 (Indifferent moralist)^a	−0.50	.19	6.7	.010	0.30	.23	1.6	.203	−0.07	.23	0.1	.776
MT_Ca1 (Moralist)^b	−0.03	.10	0.1	.721	0.03	.17	0.0	.885	0.23	.12	3.7	.054
Sex (being female)	−0.04	.08	0.3	.620	−0.11	.12	0.8	.378	0.26	.10	7.7	.005
Empathy	−0.22	.07	9.9	.002	−0.18	.07	6.5	.011	0.43	.08	28.6	.000
Moral disengagement	0.35	.09	17.0	.000	0.17	.10	2.9	.089	−0.09	.07	1.5	.219

Note

• ^a Reference group: MT_Co4 (happy transgressor).
• ^b Reference group: MT_Ca2 (happy opportunist).

The GEE analysis for ‘Ignoring bullying’ yielded no significant effects for Cluster membership (see Table 6). Only empathy emerged as a significant predictor: the higher the level of empathy the lower the frequency of ignoring bullying.

The GEE analysis for ‘Helping the victim’ yielded a marginally significant effect for Cluster membership in the accomplished deed situation, but not in the open situation. The parameter estimates are shown in Table 6. Happy opportunists indicated a tendentially lower frequency of helping the victim. In addition, sex and empathy were significant predictors. Being a girl and a higher level of empathy predicted higher scores of helping the victim.

The GEE analysis for ‘Offline bullying’ yielded no significant effects for Cluster membership. The parameter estimates are shown in Table 7. Only moral disengagement emerged as a significant predictor. The higher the level of moral disengagement, the higher the frequency of offline bullying.

Table 7. Results of the GEE analysis predicting bullying behaviour

	Offline bullying				Online bullying
	B	SE	Wald Chi^2	p	B	SE	Wald Chi^2	p
Intercept	1.20	.18	45.0	.000	1.01	.13	58.7	.000
MT_Co1 (Ashamed moralist)^a	−0.27	.17	2.6	.105	−0.28	.11	6.3	.012
MT_Co2 (Happy moralist)^a	−0.26	.16	2.8	.093	−0.27	.11	6.1	.014
MT_Co3 (Indifferent moralist)^a	−0.27	.19	2.1	.144	−0.23	.13	3.2	.075
MT_Ca1 (Moralist)^b	−0.07	.06	1.4	.239	0.08	.04	3.2	.074
Sex (being female)	0.03	.05	0.4	.553	0.07	.04	2.9	.090
Empathy	−0.02	.05	0.2	.665	0.01	.04	0.1	.738
Moral disengagement	0.19	.08	6.4	.011	0.16	.06	8.6	.003

Note

• ^a Reference group: MT_Co4 (happy transgressor).
• ^b Reference group: MT_Ca2 (happy opportunist).

The GEE analysis for ‘Online Bullying’ yielded a marginally significant multivariate effect for Cluster membership in the open situation (Wald Chi² = 6.77, p = .079), and for Cluster membership the accomplished deed situation (Table 7). Moral transgressors (happy transgressors and happy opportunists) indicated a higher frequency of online bullying than ashamed moralists and happy moralists. The difference between happy transgressors and indifferent moralists was only marginally significant. Happy opportunists in the accomplished deed situation also showed marginally higher levels of online bullying than moralists. In addition, moral disengagement was a significant predictor. The higher the level moral disengagement, the higher the frequency of online bullying (Table 7).

Discussion

We explored the relationship between adolescents’ moral functioning and both bullying (offline, online) and bystanding. Moral functioning was assessed by eliciting different moral judgements (deontic judgement, self-judgement, judging the transgression) and related emotion expectancies in the context of a passive moral temptation scenario (getting too much change money). Passive moral temptations are situations where the protagonist does not proactively intend to transgress but where the temptation is forced on him/her. Such situations are more ambiguous and less morally salient than proactive transgressions and have not yet been related to bullying and bystanding. We expected systematic relationships between the quality of participants’ construction of the transgression (judgements and emotion expectancies) and bullying and bystanding (supporting the bully, helping the victim, ignoring). Our findings show that such a systematic relationship does indeed exist: Adolescents who constructed a favourable interpretation of yielding to a passive moral temptation in a hypothetical scenario displayed higher levels of both assisting the bully and online bullying, confirming earlier research including proactive rule transgressions (e.g., Perren & Gutzwiller-Helfenfinger, 2012). Also, the higher adolescents’ favourable view of the transgression (i.e., yielding to the temptation) was, the higher was their level of moral disengagement and the lower their level of empathy.

Moral functioning in a moral temptation situation

Following a person-centred approach, participants’ constructions were determined for both the open situation (deontic and self-judgement) and the accomplished deed by performing cluster analyses including the number of transgressions (i.e., judgements in favour of keeping the money), happy emotions, and unhappy emotions, respectively. We found four clusters for the open situation: ashamed moralists (low level of transgression and unhappy emotions [mostly ashamed]); happy moralists (low level of transgression and happy emotions); indifferent moralists (low level of transgression and very low levels of positive and negative emotions); and happy transgressors (high level of moral transgression and happy emotions). For the accomplished deed, we found two clusters: moralists (low levels of transgression and unhappy emotions); and happy opportunists (high levels of transgression and happy emotions).

The cluster analyses yielded a more complex pattern for the open situation than for the accomplished deed. In the open situation, the three moralist groups differed regarding their emotion expectancies (ashamed, happy, indifferent), whereas only one moralist group was identified for the accomplished deed. This indicates that indeed two different situational contexts were being judged (cf. Nunner-Winkler, 2013) with the open situation allowing for a broader array of emotion expectancies to go with morally appropriate deontic and self-judgements (i.e., return the money). It seems that not only positive emotions (corresponding with moral pride) were expected for morally appropriate judgements, but also negative (especially shame) and ‘neutral’ emotions, suggesting that the inclusion of participants’ initial constructions of the situation offers additional insights into their inner moral landscape (cf. Wainryb et al., 2005).

While extending Happy Victimizer research, the clusters correspond well with some of the patterns found in moral (developmental) research (e.g., Oser & Reichenbach, 2005). These patterns include the Happy Victimizer (in favour of transgressing the rule and positive emotions) corresponding with the happy transgressor (open situation) and the happy opportunist (accomplished deed); the Unhappy Victimizer (in favour of transgressing the rule and negative emotions) corresponding with none of our patterns; the Happy Moralist (in favour of following the rule and positive emotions) corresponding with the happy moralist (open situation); and the Unhappy Moralist (in favour of following the rule and negative emotions), corresponding with the ashamed moralist (open situation) and the moralist (accomplished deed) situation. Whereas we did not find any cluster corresponding with the Unhappy Victimizer pattern, we identified an additional cluster, showing that there were shades and gradations regarding participants’ construction of the open situation: About a sixth of participants decided in favour of the moral rule but did not expect the protagonist (and the self) to feel particularly good or bad about it (indifferent moralist). We can only speculate at this point whether indifferent moralists in our sample see moral behaviour as a matter of course not deserving special attention or self-praise, or whether they do not frame the decision to return the money as a moral issue at all. This question needs to be pursued in further research.

Taken together, our findings confirm earlier research using passive moral temptations including adult participants. That research indicated that a far higher percentage of participants than would be expected (according to the Happy Victimizer Paradigm) showed transgression-friendly constructions (Heinrichs et al., 2015, 2020). Moreover, we found a high degree of overlap between the happy transgressor and the happy opportunist clusters: Nearly all happy transgressors (open situation) fell into the happy opportunist cluster (accomplished deed), suggesting a high degree of consistency in transgression-friendly constructions across the two situational contexts in our adolescent sample.

Moral functioning and bullying/bystander behaviour

We hypothesized that participants who constructed a favourable understanding of transgressing in this passive moral temptation scenario (high level of transgression-favouring judgements along with positive emotions) would report increased levels of bullying (offline and online) and of supporting the bully, and low levels of helping the victim. Complementarily, we expected that participants who constructed an unfavourable understanding of the transgression (low level of transgression-favouring judgements along with negative emotions) would report low levels of bullying and of supporting the bully and high levels of supporting the victim. These hypotheses were partly confirmed. First, in the open situation, moral functioning predicted supporting the bully, helping the victim, and online bullying: Happy transgressors reported higher levels of antisocial behaviour than happy moralists (assisting and online bullying), ashamed moralists (assisting only tend; online bullying), and indifferent moralists (assisting, online bullying only tend.). Also, they reported lower levels of helping the victim than ashamed and happy moralists. However, no systematic relationship was found for offline bullying or ignoring. Second, in the accomplished deed situation, moral functioning predicted online bullying and helping, but only marginally significantly: Happy opportunists reported higher levels of online bullying and lower levels of helping than moralists. No association was found for assisting, offline bullying, or ignoring.

Regarding bystanding, results for the open situation offer a clear picture: Transgression-friendly constructions of the passive moral temptation were significantly positively related to assisting the bully and negatively to helping the victim, and not to ignoring. Participants who judged that the protagonist should keep the money and feel happy about it and that they themselves would keep the money and feel happy about it reported higher levels of assisting the bully and lower levels of helping the victim. This suggests that the basic configuration of both the passive moral temptation situation and bullying situations might share essential similarities: In both cases, the rule transgression is neither intended nor planned: It offers itself due to specific circumstances and represents a temptation that cannot be withstood. This goes together with a weakened notion of agency (and the related feeling of responsibility) because someone else’s actions brought about the situation: The shop assistant gave too much change money; the bully started the bullying. Both (own) intentionality and agency are core features to consider when judging the harmfulness and therefore moral relevance of acts (Bandura, 2002; Guglielmo & Malle, 2010). If both can be shown as not contingent on one’s own role in the context of a moral or morally relevant situation, then one can decide to yield to the temptation. In the passive moral temptation situation, the moral norm involved refers to a positive duty which by nature carries a weaker moral obligation than a negative duty (Belliotti, 1981). Against this background, we may ask ourselves whether this does not also hold for bullying from the perspective of a bystander: As the bullying is neither intended nor initiated, the question might not be as much whether to bully or not, but rather whether to help the victim or not. Of those peers who decide not to help the victim, some may see a welcome opportunity to join in the aggression while still being able to reason that they did not plan nor start it. At this moment, motives like gaining (or not losing) social status, pressure from aggressive peers, or friendship loyalty towards the bully might become more powerful and inform behaviour (cf. Forsberg, Thornberg, & Samuelsson, 2014).

With respect to aggression, transgression-friendly constructions in the open situation were related to online bullying only. This raises the question whether online bullying also shares some specific features with passive moral temptations and with bystanding. It is possible that some forms of online bullying like forwarding a nasty picture one received or joining in when someone is being insulted on a social media platform or app also resemble passive moral temptation situations: Again, someone else had started the aggression, and one had not accessed the platform or app with the intention of bullying someone in the first place. Thus, the boundaries between online bullying and bystanding might be blurred in situations of this kind. This assumption is supported by a study by Barlińska, Szuster, and Winiewski (2013). Their findings showed that earlier experiences of cyberperpetration increased the likelihood of negative bystander behaviour (assisting, reinforcing) in an adolescent sample, suggesting an overlap of aggressive behaviours in online bullying. The authors state that ‘a bystander audience in cyberspace can play an active role by participating in the victimization. Although they may not have created a text or image, individuals are complicit in spreading it to ever-widening audiences’ (p. 38). However, we can only speculate whether this blurring effect might have occurred in our study. We assessed online bullying using items that described behaviours without reference to any specific context and, moreover, did not assess online bystanding. The relationship between moral functioning in the context of passive moral temptations and online bystander behaviour needs to be investigated in subsequent research to gain a fuller picture of the resonance such temptations find in participants displaying various forms of online bystanding.

Empathy and moral disengagement

As in previous studies, empathy in our study was significantly associated with bystander behaviour: Higher levels of empathy predicted lower levels of assisting the bullying (cf. Gini et al., 2008) and ignoring (cf. Nickerson, Mele, & Princiotta, 2008) and higher levels of helping the victim (cf. Caravita et al., 2009). However, empathy did not predict offline or online bullying, although we found a significant negative bivariate relationship with offline bullying. This partly contradicts research showing systematic relationships between empathy and online (Kowalski, Giumetti, Schroeder, & Lattanner, 2014) and offline bullying (Zych, Ttofi, & Farrington, 2016). The picture becomes clearer when distinguishing between cognitive empathy (understanding someone else’s emotions) and affective empathy (sharing someone else’s emotions; cf. Eisenberg, 2000). Whereas offline bullying was found to be associated with affective empathy only (Jolliffe & Farrington, 2011), online bullying was negatively associated with both forms of empathy (Ang & Goh, 2010). The measure used in the present study consisted of four items assessing empathic concern, which is more closely related to sympathy. Sympathetic responses can be based either on empathic sadness or cognitive empathy (Eisenberg, 2000) and thus may include affective, cognitive, or both elements. It is possible that our empathy measure led to differential interpretations by participants. Indeed, using the same measure in an earlier study, a negative bivariate relationship between empathic concern and cyberbullying was found, although the former was not a significant (concurrent and longitudinal) predictor of cyberbullying (Sticca et al., 2013). Additional research including empathic concern (as a subdimension of empathy, cf. Eisenberg, 2000) as well as further subdimensions of empathy is needed to shed more light on our findings.

Confirming earlier research, moral disengagement predicted active antisocial behaviour, that is offline bullying (cf. Gini et al., 2014), online bullying (cf. Chen, Ho, & Lwin, 2017) and assisting the bully (cf. Thornberg & Jungert, 2013). Furthermore, we found positive bivariate associations with transgression-friendly judgements in both the open and the accomplished deed situation, as well as positive associations with happy emotions and negative associations with unhappy emotions in the accomplished deed situation. It seems that moral disengagement might also play a role in adolescents’ constructions of a passive moral temptation situation, both on the level of judgement and of emotion expectancies (accomplished deed only). That emotion expectancies were related to moral disengagement in the accomplished deed situation, only, implies that there is an actual need to do so when a transgression has already happened and must be ‘explained away’. This is exactly what constitutes the necessity to use moral disengagement strategies: to neutralize feelings of guilt (cf. Bandura, 2002).

Our findings have several developmental implications. They suggest that holding a favourable view of resisting to a passive moral temptation, in conjunction with the use of moral disengagement strategies and a lack of empathy, may put supporters of bullies at risk of maintaining or even intensifying their pro-bullying behaviours. This intensification in turn may render proactive bullying behaviour more attractive and may motivate supporters to start acting as ringleader bullies. As bullying (if not intervened against) offers many rewards to bullies, like for example enhancing their social status (cf. Pouwels et al., 2016), these observed rewards may make it even easier for supporters to downplay the harm they cause. As a consequence, switching between aggressive roles may become normal and easy, because the other aggressive youth can be expected to support the bullying (cf. Oh & Hazler, 2009).

In the context of online bullying, where boundaries between bullying and bystanding are less clear (cf. Barlińska et al., 2013), this risk of maintaining pro-bullying behaviours and extending them to proactive bullying may even be higher, particularly in situations where youth receive nasty pictures, texts, or videos exposing another peer. One reason might lie in the high affordance of such situations: The material is already located on the device without the recipient’s active contribution, and ready to be forwarded. A few clicks, and it is done. Thus, the potentially ‘toxic’ combination of holding a favourable view towards yielding to a passive moral temptation, combined with the use of moral disengagement strategies and a lack of empathy, may prove especially detrimental in the cybercontext. However, as we did not assess online bystanding, this interpretation remains speculative at present.

Strengths, limitations, and directions for future research

This exploratory study combines both strengths and weaknesses. A major strength lies in the assessment of moral functioning in the context of a ‘non-bullying’ scenario. This adds to the validity of the measurement, as there is no conceptual overlap between the scenario and the bullying and bystanding scales. Accordingly, we assume that the moment of temptation might prove one core moral element explaining pro-bullying bystander behaviour. Second, using different forms of moral judgement may have lent additional strength to our measurement approach, as it enabled participants to respond to various facets of the initial situation. This gave them the opportunity to deal with complexity in a situation which was rather open, ambiguous, and far from clear with respect to morality. A third strength lies in the ecological validity of the scenario used, as getting too much change is a common, everyday experience shared by many people. Thus, despite using a hypothetical scenario, we may still assume that the situation was familiar enough for participants to render it credible and relevant.

With respect to limitations, a first drawback is that we included only students as informants and based the bullying, bystanding, and empathy measures on self-report scales. Also, we used only one passive moral temptation scenario and thus restricted our measurement to a money-related situation where the protagonist was not in the presence of peers. Besides the need for replication of our findings, future research involving passive moral temptation situations will have to include additional contexts like for example cheating at school, as well as variations regarding the peer context. Also, including intensity ratings for moral emotion expectations might shed further light on the way passive moral temptations – as relating to positive duties (prosocial norms) – are interpreted. Previous research indicated that the intensity or magnitude of expected negative emotions was weaker in the context of failing to act prosocially than in the context of active antisocial behaviour (Krettenauer & Jia, 2013). Moreover, there is also a need to include online bystander behaviour in order to explore potential differences or similarities with offline bystanding. A further restriction is that – as in all aggression and bullying research – the use of self-report scales on bullying and bystanding may have led some participants to give socially desirable answers despite our anonymous assessment procedure. Finally, only part of eligible participants had time to complete the temptation measure. As our drop-out analyses showed, there was a significant relationship between missingness in the moral temptation measure and aggressive behaviour on the one hand and between missingness in the aggression scales and membership in the happy transgressor or happy opportunist clusters on the other hand. Thus, our findings can be generalized only very cautiously.

Despite these limitations, our exploratory study makes an important contribution regarding the moral dimensions of bullying and particularly bystanding. We were able to offer first indications of a highly meaningful, differential relationship between moral functioning in a passive moral temptation situation and self-reported pro-bully offline bystanding (assisting and reinforcing). This opens new avenues for bullying prevention and intervention. Our findings suggest that assistants and reinforcers of offline bullying might be in need of and particularly benefit from modules targeting the ability to identify the moral norms and related obligations (and responsibility) in less clear-cut, more ambiguous passive temptation situations, and to consider them for subsequent judgements and behaviour. In line with the four component model of moral action (e.g., Rest, Narvaez, Bebeau, & Thoma, 2009), we assume that this ability, often termed moral sensitivity, forms one important basis to bridge the gap between moral knowledge and moral action. In the context of peer aggression, basic moral sensitivity in bullying (as assessed using a self-report scale tapping students’ attitudes on bullying) has previously been shown to be related negatively to pro-bully behaviour and positively to outsider and defender behaviour (Thornberg & Jungert, 2013). Accordingly, what bystanders displaying pro-bully behaviour may need with respect to the promotion of their moral development might refer less to education in the area of clear-cut, proactive rule transgressions, but in areas where the boundaries of morality are fuzzy.

Acknowledgements

This study was supported by a grant from the Swiss National Science Foundation (SNF No. 100014_130193/1) to the second author. The authors would like to acknowledge the collaboration of all students involved in data collection. Furthermore, we would like to thank all study participants and teachers for the participation over the whole duration of the study. Open access funding enabled and organized by Projekt DEAL.

Conflicts of interest

All authors declare no conflict of interest.

Author contribution

Eveline Gutzwiller-Helfenfinger, Ph.D. (Conceptualization; Data curation; Formal analysis; Methodology; Resources; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing) Sonja Perren (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Project administration; Resources; Validation; Visualization; Writing – original draft; Writing – review & editing).

Appendix

Moral temptation vignette (female version)

Jana is a girl your age. She urgently needs a new light for her bike, so she goes to a bike shop nearby. The shop is owned by the parents of a classmate of Jana’s. The classmate sometimes helps out at the shop. She sells Jana a bike light. When Jana wants to pocket the change, she realizes that her classmate gave her too much money back in change (10 Euros)

Part 1: Open situation

Q1. Deontic Judgement

What should Jana do? (return the money/keep the money)

Why should she do this?

How does Jana feel? (happy/proud/indifferent/sad/angry/anxious/ashamed)

Why does she feel this way?

Q2. Self as Protagonist

What would you do if you were Jana? (return the money/keep the money)

Why would you do this)

How would you feel? (happy/proud/indifferent/sad/angry/anxious/ashamed)

Why would you feel this way?

Part 2: accomplished deed situation

[The story continues:] Jana pockets the money.

Q3. Transgression

Is it okay to pocket the money? (yes/no)

Why is it (not) okay to pocket the money?

How does Jana feel after she pocketed the money? (happy/proud/indifferent/sad/angry/anxious/ashamed)

Why does she feel this way?

[The story continues:] Jana did not get any discount. She knows that all her classmates get a discount in this shop.

Q4. Transgression

Is it now okay to pocket the money? (yes/no)

Why is it (not) okay to pocket the money?)

Now that Jana knows this: How does she feel after she pocketed the money? (happy/proud/indifferent/sad/angry/anxious/ashamed)

Why does she feel this way?

[The story continues:] The classmate tells Jana: ‘If there is money missing in the till, I have to pay it out of my own pocket’.

Q5. Transgression

Is it now okay to pocket the money? (yes/no)

Why is it (not) okay to pocket the money?)

Now that Jana knows this: How does she feel after she pocketed the money? (happy/proud/indifferent/sad/angry/anxious/ashamed)

Why does she feel this way?

Open Research

Data availability statement

Data available upon request from the second author.