pap 3

To Forgive, or Not to Forgive: How the Sincerity of Apologies Online Affect Forgiveness

Jane Doe

Florida International University

To Forgive, or Not to Forgive: How the Sincerity of Apologies Online Affect Forgiveness

Everybody makes mistakes. From mundane situations such as accidentally grabbing someone else’s lunch to more unique incidents like breaking mom’s favorite vase. While they are not limited to mistakes, an expected apology follows transgressions by recognizing responsibility, consequences, and emotions. Apologies are a two-way street, with the transgressor apologizing and the victim(s) receiving the apology and potentially forgiving the transgressor. However, this act of communication and hopeful mitigation can be driven by numerous emotions that affect its delivery and impression (Hareli & Eisikovits, 2006). Equivalently, the decision on whether that apology should be accepted and forgiven by the recipient is also a product of multiple influences, including type of apology and offense removal (Zeichmeister et al., 2004). A novel factor is the environment, specifically the online environment of the internet and social media platforms. Thus, our study is fixated on analyzing how favorable participants find apologies with varying levels of sincerity (i.e. sincere or insincere) presented on the social network system Twitter, as well as apologies compared to receiving no apology at all. Overall, the study seeks to reveal how the manner in which a transgressor apologizes affects their self-image and perception of the apology.

Even apologies serve a purpose, which is namely to restore relationships between the transgressor and the audience and to repair images of an individual’s self as perceived by others at a social value, otherwise known as efforts to save “face” (Hareli & Eisikovits, 2006; Hu et al., 2019; Matley, 2018). Media figures, such as celebrities and politicians, often engage in apologizing as a strategy to maintain a positive image in the eyes of the public after causing a transgression that defies norms (Hu et al., 2019). Self-presentation, or “face” on social media, has been found to demonstrate a positivity bias, where positive forms of self-presentation defined by norms within an online environment are favorable compared to negative forms (norm defying) (Matley, 2018). Positive self-presentation aims to form a positive impression, such as a photo of a date one planned out for their partner, while negative self-presentation may risk ruining that impression by expressing socially undesirable behavior such as hateful comments. This “saving face” factor encourages apologizing after transgressions to cancel out the offense that damages an individual’s image and reputation, and thus regain that positive self.

Additional components that comprise an apology are the emotions in which the transgressor elicits the apology from. These social emotions are key in the process between relationships, as delivery can affect how the message is received. In a study by Hareli and Eisikovits (2006), participants were asked to envision being insulted and getting their feelings hurt by a friend. When the participants were provided with an apology from this friend later on, each apology included the same phrase but differed in cause of the apology mentioned. The causes were stated to be either of the social emotions: guilt, shame, or pity, as well as combinations of two or all three. The apology was then rated in terms of forgiveness and anger toward that friend. The findings suggest that when guilt and shame were stated as the cause for the apology, ratings of forgiveness were higher. However, expressing pity decreased the participant’s forgiveness and increased anger towards the transgressor. The participants perceived apologies motivated by guilt and shame as sincere, although guilt was considered more sincere than shame when comparing an apology driven by guilt and shame to those driven solely by guilt (Hareli & Eisikovits, 2006). Perception of sincerity within the apology and forgiveness had a strong positive correlation, suggesting that the social emotions (i.e. guilt and shame) may serve as an indicator for perceived sincerity of apologies, and thus impact whether the transgressor will be forgiven or not (Hareli & Eisikovits, 2006). This highlights that the perception of an apology, not just the apology itself, is the true indicator in whether a relationship has successfully been mended or if the transgressor’s public image has been cleansed. Similar, in a study performed by Zechmeister et al. (2004), participants would perceive apologies distinctively by the presence or absence of certain components. Participants considered apologies insincere when the transgressor apologized, but did not remove the offense, and considered apologies sincere when the transgressor apologized, removed offense, and augmented an effort to make amends. Accordingly, the insincere apologies were the least forgiving, and the sincere apologies were more likely to be forgiven (Zechmeister et al., 2004). The exchange of an apology and forgiveness is extremely complex beyond the surface, and it is more difficult considering the two parties involved might not be on the same page.

Further research by Leunissen et al. (2013) observed and found that guilt mediates the perpetrator’s willingness to apologize depending on the intentions behind the transgression. Perpetrators preferred to apologize for an unintentional transgression more so than intentional ones. A disparity existed between perpetrators and victims, where perpetrators are more likely to admit following a transgression if it was unintentional, but victims were more likely to expect an apology when the transgression is intentional. A victim’s need for an apology is mediated, instead, by anger (Leunissen et al., 2013). Additionally, Kirchhoff et al. (2012) conducted research on the relevancy of verbal components within an apology and how it relates to forgiveness. Leunissen et al. (2013) propose that the apology mismatch can function as an index for whether perpetrators are forgiven. The findings suggest that an apology is predicted more so by the perpetrator’s needs and, as such, are forgiven more when they apologize (Leunissen et al., 2013). Thus, the transgressor’s higher need to apologize after unintentional transgressions promotes its probability of being forgiven compared to intentional transgressions.

Considering the elaborate process of apologies known thus far—how function and feelings affect sincerity, and hence, forgiveness—researchers are looking into how much more complex it gets when moved onto an online environment. Social network systems have expanded audiences and communities from local to worldwide, making every post and detail much more profound. This goes back to one’s sense of self and what we want presented online. Matley (2018) investigates this concept by observing the function of speech acts on the platform app Instagram and examines how the hashtag #sorrynotsorry functions as a non-apology marker. Apologies operate as speech acts that serve as justification and accountability for behavior online, all to make users’ social value return to the positive scale (Matley, 2018; Hu et al., 2019). This is especially interesting considering how individuals can choose to present their whole lives for the world to see or keep themselves hidden behind anonymous accounts. This allows for perpetuation of self-conflict between social approval and sharing offensive opinions. Hashtags on social media are used to search for content, but also provides the user with the control to manipulate the interpretation of a post’s content (Matley, 2018). The #sorrynotsorry hashtag (essentially the phrase “Sorry, not sorry”), as studied by Matley (2018), has two components: the “sorry” and the “not sorry”. The “sorry” element functions as a face-mitigation strategy, which attenuates the offenses made within the content of the post in order to reestablish a positive face. The “not sorry” attached along conflicts with the initial “sorry” by expressing disregard for the recipient’s emotions, which threatens positive face and diminishes social approval (Matley, 2018). Moreover, the ironic hashtags attached to posts can be detrimental to social status.

To add onto this research, our present study aims to explore essentially how differing levels of sincerity and the presence of an apology, using hashtags on a social network system attached to an apology post, can affect the favorability of whether the transgressor can be forgiven. Participants were all given a scenario in which Charlie Webb, a user on the social media platform Twitter, apologizes for an incident at the mall. Participants were randomly assigned to different conditions where Charlie offers either a sincere apology, an insincere apology, or no apology. The participants were then asked to rate Charlie’s actions and how favorable they view the apology offered.

In general, it was predicted that participants in the sincere apology condition viewed the apology (and apologizer) more favorably than participants in both the no apology and insincere apology conditions, though participants viewed an insincere apology less favorably than no apology.

More specifically, if participants view a sincere apology (compared to an insincere apology or no apology), then they will more strongly agree that the apology showed an acknowledgement of wrongfulness and an acceptance of responsibility, though participants who viewed an insincere apology will more strongly disagree with these statements compared to those who saw no apology. In addition, if participants viewed a sincere apology (compared to an insincere apology or no apology), then they will more strongly agree that the apology is sincere, with participants ironically finding no apology more sincere than an insincere apology.

Methods Study One

Participants

The following study consisted of 141 randomly assigned participants. Of these participants, 44.7% ( n = 63) were male and 55.3% ( n = 78) were female. Participant age ranged from 17 to 50 years old with the average participant age being M = 24.60 ( SD = 7.92). This sample included 50.4% Hispanic American ( n = 71), 28.4% Caucasian ( n = 40), 10.6% African American ( n = 15), 4.3% Asian American ( n = 6), 1.4% Native Indian ( n = 2), and 5% participants that classified themselves as “Other” ( n = 7). See Table 1.

Table 1

Demographics for Study One .

Materials and Procedure

Before distributing the questionnaires, 141 participants were contacted and asked for informed verbal consent to participate in the research study and advised that there were no risks in participating. If the participant granted oral consent to participate, they were subsequently given one of three documents that contained a screenshot of a Twitter page and four-part questionnaire. Part one of the document included a Twitter page for a user named Charlie Webb. The profile of Charlie’s Twitter page was identical in all documents, including a generic header, profile picture, and “bio” (biography), as well as usual items such as the search bar, trending hashtags, and “follow” recommendations. Most importantly, it consisted of an initial three “tweets” (posts) in which the user Charlie Webb described an incident that occurred at the mall during the COVID-19 pandemic. They said they were not wearing a face mask and defied the norm of social distancing by invading an employee’s space, leading to an argument with the mall employee. However, the following two tweets after the description of the transgression differed between conditions. In the “Sincere” condition, Charlie apologizes for their actions, acknowledges being wrong, accepts responsibility, and accepts their punishment. They follow this statement with the hashtag “#SorrySorrySorry”. In the “Insincere” condition, Charlie apologized similarly to the sincere condition, but ends the tweet with “Ha! #SorryNotSorry”. Lastly, the third condition which is “No” apology states simply suggests leaving the incident in the past and ends with the hashtag “#WhatsDoneIsDone”.

In part two of the study, participants were asked to rate their impression of Charlie on the basis of what they read from the Twitter page provided to them in part one, without looking back to it. The ratings were measured on an interval Likert scale, ranging from 1 (Strongly Disagree) to 6 (Strongly Agree) on statements such as Charlie acknowledging wrongfulness, accepting responsibility, expressing remorse, offering compensation, assuring not to behave in that manner again, whether the apology seemed forced, if the apology seemed sincere, and whether the participant themselves would accept the apology. In part three of the study, participants were offered to rate more statements on the same Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree) as part two judging Charlie’s behavior, such as if it was wrong, understandable, and forgivable, as well as their perceptions of Charlie as being regretful, selfish, moral, and rude. Lastly, participants rated how much they agree with the statement: “I don’t think people should be forced to wear a mask at stores if they don’t want to wear it”. For this study, only two dependent variables were analyzed from the sixteen ratings measured.

In part four of the study, participants were asked to fill out demographic questions, but were advised that it was not necessary to answer if uncomfortable with doing so. The demographic questions included the participants’ providing their gender, age, race/ethnicity, whether English is their first language, and if not, to write in their first language, whether they are a student at Florida International University, and their relationship status. In closing, part five concluded the document with a nominal measurement for attention in which the participants were asked to recall whether the hashtag they saw in the Twitter page previously shown was: “#SorrySorrySorry”, “#SorryNotSorry”, or “#WhatsDoneIsDone”.

After participants completed the questionnaire, they were debriefed regarding the purpose of the study being perception of sincerity within apologies, the manipulation of apologies in all conditions, and the predictions for the study. Analyses were conducted on the attention check of the hashtags and whether participants perceived Charlie’s apology as sincere and if they showed acceptance of responsibility (ratings from part two).

Results Study One

First, a chi square test of independence was conducted using apology condition as our independent variable (sincere, insincere, or no apology). The dependent variable is recall of the hashtag at the end of Charlie’s thread of tweets, in order to determine if participants recalled the condition they were exposed to. The chi square was significant, X2(4) = 121.93, p < .001. Most “Sincere” condition participants recalled #SorrySorrySorry (87.2%); most “Insincere” condition recalled participants #SorryNotSorry (74.5%); and most “No apology” condition participants recalled #WhatsDoneIsDone (68.1%). Cramer’s v was strong. These results serve to confirm that our participants viewed the manipulation as intended. See Table 2.

Table 2

Crosstabs and Chi Square – Study One

Using apology condition (sincere, insincere, or no apology) as the independent variable and participants’ agreement with “Charlie’s apology showed an acceptance of responsibility” as the dependent variable, a one-way ANOVA revealed differences among participants. A significant relationship emerged, F(2, 138) = 8.82, p < .001. A Tukey post hoc test revealed that participants in the sincere condition ( M = 4.79, SD = 0.69) agreed more with the statement than both the participants in the insincere ( M = 4.17, SD = 0.73) and no apology ( M = 4.36, SD = 0.76) condition. However, the insincere and no apology conditions did not differ in agreement with the responsibility statement. See Table 3.

Table 3

Responsibility ANOVA

A second One-Way ANOVA analysis was conducted with apology condition as the independent variable (sincere, insincere, or no apology) and rate of agreement with the statement “Charlie’s apology seemed sincere” as the dependent variable. A significant relationship emerged, F(2, 138) = 11.90, p < .001. A Tukey post hoc test revealed that participants in the sincere condition ( M = 4.34, SD = 0.87) agreed with sincerity statement more than both the insincere ( M = 3.38, SD = 1.13) and no apology condition ( M = 3.87, SD = 0.82), with the no apology condition agreeing with the statement more than the insincere condition. See Table 4.

Table 4

Sincerity ANOVA

Discussion Study One

We predicted that the participants would be aware of the apology manipulation in all three conditions, which was supported by the results in the chi square test of independence. We also predicted that participants in the sincere apology condition would view the apology, and the apologizer, more favorably in terms of agreeing with statements that Charlie accepted responsibility and was sincere, than participants in both the insincere and no apology conditions. However, participants in the insincere condition would view the apology less favorably than those in the no apology condition. The sincerity ANOVA analysis supported the hypothesis, with sincere participants agreeing with Charlie’s apology seeming sincere, more so than the participants in the insincere and no apology condition. On the other hand, while the responsibility ANOVA analysis showed that participants in the sincere condition agreed with the statement that Charlie’s apology accepted responsibility more than participants in both the insincere and no apology conditions, the insincere and no apology conditions viewed the apology equally. While the results showed statistically significant differences, the average rating in the sincere condition leaned more towards agreeing rather than strongly agreeing. Based on the literature, these findings were expected considering how the dependent variables shown, acceptance of responsibility and sincerity, influence the perception of an apology as well as the chances of being forgiven. Perhaps in the responsibility variable, the lack of an apology seemed just as unsympathetic as an insincere attempt of an apology, leading to the results found.

Study Two

Gender, as well as sex, are primarily instilled in society as a dichotomy: man and woman as a gender binary, while sex is distinguished between male and female. A more modern view emphasizes that gender is socially defined and has implicit effects on our social interactions (Benetti-McQuoid & Bursik, 2005). As people engage in social interactions, cognitive processes evaluate behavior by its consistency with one’s organization of knowledge, or schema. These schemas then guide our expectations and judgements. A gender schema serves as a heuristic, or mental shortcut, in which information, involving the self, is processed. Gender serves as part of self-identification, so it is emphasized to adopt a gender role. Gender roles don’t necessarily have to match one’s sex—as people expand their schemas about femininity and masculinity, an individual may decide where their own attitudes are oriented within that spectrum (Bem, 1993, as cited by Benetti-McQuoid & Bursik, 2005). Schematic individuals align themselves with traditional gender roles. Traditional masculinity emphasizes autonomy and assertiveness, while traditional femininity emphasizes being sensitive and communally oriented. This matches with the prioritization of independence and relationships in males and females, respectively (De Boeck et al., 2018; Benetti-McQuoid & Bursik, 2005). Specifically, researchers are now investigating how gender roles affect transgressions, both self-conflict and interpersonal.

Aligning with traditional gender roles serves as a branch of one’s identity, so behaviors that do not fit the expectations for the prescribed gender role result in self-criticism in the form of guilt or shame (De Boeck et al., 2018). Guilt is a by-product of emotion following a wrongdoing where the transgressor feels responsible for having acted that way, which focuses on the behavior. Shame is more concerned with negatively perspective of the self, usually blaming themselves for their internal state, and that self-criticism is reflected onto negative behavior (Benetti-McQuoid & Bursik, 2005; Konstam et al., 2001; De Boeck et al., 2018). Multiple studies observe the link between gender and critical emotions.

In De Boeck et al.’s study (2018), the researchers examined men and women’s anticipated feelings of guilt and shame for wrongdoings by rating their thoughts on hypothetical situations of shoplifting and engaging in violence. They analyzed how social origins, such as assigned values of traditional femininity or masculinity taught in women and men, respectively. Girls are taught values of being empathetic and prioritizing relationships, so girls are expected to comply with their friend’s decisions and make amends to avoid harming relationships. Boys, on the other hand, are primed and encouraged to be independent (De Boeck et al., 2018; Benetti-McQuoid & Bursik, 2005). Delinquent behavior was considered inconsistent to the socially defined feminine schema, and thus, feminine individuals were expected to inhibit delinquency to avoid anticipated negative emotions like guilt or shame. The data collected supported that observation, as female participants experienced a higher threat of shame-guilt for engaging in delinquency than male participants (De Boeck et al., 2018). Benetti-McQuoid and Bursik (2005) reported similar findings, with women reporting greater proneness to guilt and shame and people aligned with feminine gender roles had heightened levels of being prone to guilt and shame. Additionally, gender schematic women favored verbal responses to ameliorate the experience of guilt (i.e. apologies), whereas gender schematic men preferred action-oriented responses.

Guilt and shame can further be observed to mediate the relationship between gender and forgiveness. Konstam et al. (2001) found a positive correlation between forgiveness and guilt-proneness and a negative correlation between forgiveness and shame-proneness. The researchers executed regression analyses comparing the impact of gender and level of guilt- or shame-proneness on forgiveness. Women tended to be more prone to guilt, which predicted forgiveness, while men’s pride in their behaviors (opposite of shame) predicted forgiveness.

While both gender roles seem to clearly influence on forgiveness, there is a significant difference in probability between the two. Recall that guilt, when identified as the motivating cause behind an apology, is a better predictor for sincerity compared to shame, and thus, a better predictor for forgiveness (Hareli & Eisikovits, 2006). If women are more prone to guilt, and men to shame, then women can presumably be perceived as sincerer and thus more likely to be forgiven. Forgiveness often follows apologies, so looking at how gender interacts with apologies can give further information. A common stereotype is that women are more willing to apologize than men. Researchers Schumann and Ross (2010) conducted a study seeking as to why and whether this claim had any truth behind it. Participants were given the task to keep a record of offenses committed (a transgressor perspective) and experienced (victim perspective), which was then submitted onto a daily questionnaire for twelve days. The researchers observed gender as an independent variable and measured the frequency of apologies given and received. The findings revealed that women recorded committing and experiencing more offenses than men. The stereotype mentioned earlier tends to attribute men’s lower offense count on their disposition, or personality (i.e. a fragile ego). The researchers suggest this may be an error and instead offer another explanation with the diary data collected: men may possess a higher threshold for their perception of what is offensive, and thus, a higher threshold for whether a situation calls for an apology (Schumann & Ross, 2010). This proposes that transgressions men commit or experience may not meet or exceed what they consider the “limit” between what is offensive and what is not. The data specifically suggests that men need additional knowledge to decide if something has “crossed the line” (the threshold) in comparison to women.

Schumann and Ross (2010) thus conducted a second study seeking support for the threshold explanation by asking men and women to rate whether an imagined or recalled offense was severe, whether an apology was needed, and whether they’d apologize. Results showed that women perceived both imagined and recalled offenses as more severe, which predicted whether the situation called for an apology, and thus, whether an apology was likely to be given (Schumann & Ross, 2010). However, both theories behind the perception of women apologizing more than men should be kept in mind considering not all people look into research. Stereotypes inform one’s perception of the transgressor, so it is likely that deciding whether to forgive someone can be dependent on internal factors rather than situational. In relation to our study, participants could be considering the willingness to apologize, which may differ by gender, or how severe the transgressor perceives their offensive behavior and how authentic the apology is.

Along with proneness to guilt and shame, the characteristics associated with each gender role coincides with characteristics of specific universal dimensions of social perception. The “warmth” dimension sums up perceptions of people being trustworthy and sincere, qualities that people look for judging new people. The “competent” dimension of the scale accounts for whether that person can pursue their goals by being skillful and capable (Fiske, 2007). This binary dimension of social perception can be perceived ambivalently; one dimension is high, the other is low. Women are typically considered high in warmth yet low in competence, while men are considered high in competence but low in warmth (Fiske, 2007). Traditional gender roles for women emphasize communal (warmth) traits, while traditional gender roles for men emphasize competence. However, judgements of warmth are prioritized over competence. The researchers suggest an evolutionary explanation: determining someone’s intentions and morals (warmth) is more important than whether they can perform those intentions (competence) (Fiske, 2007). It is not black and white, however, as subgroups within each gender can vary in warmth and competence. Nonetheless, people tend to rely on stereotypes and traditional gender roles to determine whether each gender is more warm or competent.

The warmth-competence dimensions are analyzed when judging a person’s disposition and is influenced by gender. Then these social perceptions regulate perception of apologies. Researchers Wei and Ran (2019) studied the impact of an apologizer’s gender on forgiveness in a corporate-consumer context. Chinese participants were enlisted with the task of reading and evaluating an online business article, presumably of a corporate apology. The participants then rated their agreement with statements depicting forgiveness of the company apologizing, perceived competence and warmth of the apologizer, along with other factors such as whether the transgression was related to performance or values. Keep in mind, though the study evaluated Chinese participants, these social dimensions are considered universal. Results showed that male apologizers elicited more consumer forgiveness when the offense was related to performance, while a female apologizer elicited more consumer forgiveness when the offense was related to values. The apologizer’s gender influenced whether they were perceived as warm (feminine) or competent (masculine) and that distinction predicted consumer forgiveness. Perception of warmth or competence mediated whether gender influenced forgiveness.

There is a gap in research on interactions of gender and sincerity effects on forgiveness on social media platforms. To solve this, a replication study with extension was conducted to observe how gender and sincerity can impact participant’s judgements of an apology. Participants viewed a Twitter profile of a woman or a man (both nicknamed Charlie) describing an incident and then apologizing either sincerely or insincerely. Four conditions were now examined: (1) sincere/woman condition, (2) sincere/man condition, (3) insincere/woman condition, and (4) insincere/man condition.

It was predicted that the main effect of sincerity would show participants in the sincere condition judge Charlie’s apology as acknowledging wrongfulness and accepting responsibility more than participants in the insincere condition.

The prediction for the main effect of gender presented that participants would find the woman condition as more sincere and accepting of responsibility than participants in the man condition.

The final prediction analyzes the interaction of a Twitter user’s gender perceived on a profile picture and sincerity of the apology. It is predicted that, in terms of perceived sincerity and acceptance of responsibility, the sincere and woman condition would be favored the most and the insincere and man condition would be favored the least. The sincere man and insincere woman conditions would be between the two.

References

Benetti-McQuoid, J., & Bursik, K. (2005). Individual differences in experiences of and responses to guilt and shame: Examining the lenses of gender and gender role. Sex Roles: A Journal of Research, 53(1-2), 133–142. https://doi.org/10.1007/s11199-005-4287-4