A Theory of the Self for the Sociology of Morality

ASRXXX10.1177/0003122411433762S 433762tets and CarterAmerican Sociological Review 2011

American Sociological Review 77(1) 120–140 A Theory of the Self for © American Sociological Association 2012 DOI: 10.1177/0003122411433762 the Sociology of Morality http://asr.sagepub.com

Jan E. Stetsa and Michael J. Carterb

Abstract Sociology has seen a renewed interest in the study of morality. However, a theory of the self that explains individual variation in moral behavior and emotions is noticeably absent. In this study, we use identity theory to explain this variability. According to identity theory, actors are self-regulating entities whose goal is to verify their identities. An individual’s moral identity—wherever it falls on the moral–immoral continuum—guides behavior, and people experience negative emotions when identity verification does not ensue. Furthermore, the identity verification process occurs within situations that have cultural expectations—that is, framing rules and feeling rules—regarding how individuals should act and feel. These cultural expectations also influence the degree to which people behave morally. We test these assumptions on a sample of more than 350 university students. We investigate whether the moral identity and framing situations in moral terms influences behavior and feelings. Findings reveal that the identity process and framing of situations as moral are significantly associated with moral action and moral emotions of guilt and shame.

Keywords emotions, identity, morality, self

The sociology of morality is experiencing a Sociological theory and research is needed to resurgence in the discipline (Abend 2010). understand this variability. For this, we need a This is timely given the culture of unchecked theory of the self. We rely on identity theory in consumption and greed that contributed to the sociology (Stryker [1980] 2002; Stryker and 2008 downturn of the U.S. economy. Sociolo- Burke 2000) to explain the internal operations gists are re-examining issues discussed in ear- of the self. We apply these internal operations lier work, such as the relationship between the to reveal individuals’ variability as moral moral order and market society (Fourcade and actors, and we test this theory on a sample of Healy 2007), class and morality (Sayer 2010), college students. and moral order and community (Vaisey 2007). We focus on the identity verification pro- While morality helps maintain the social order, cess in identity theory (Burke and Stets 2009). we must be careful not to reify the impact of social institutions on moral behavior and dis- aUniversity of California, Riverside count the self as an agent of moral action. If we bCalifornia State University, Northridge recognize social actors as moral actors, then at issue is why some individuals behave morally Corresponding Author: Jan E. Stets, Department of Sociology, University and others less so, and why some feel bad of California, Riverside, CA 92508 for immoral actions while others do not. E-mail: [email protected] Stets and Carter 121

Identity theory assumes that humans actively impossible without a moral system, as indi- engage in goal-directed action as they interact viduals will act solely in their own interest, with their environment. People continuously disregarding the collective interest. Self- evaluate their actions in relation to their inter- serving action leads to anomic conditions and nal identity standard. Their identity standard is pathological for society; happiness results defines who they are along various dimen- from social systems with strong moral orders. sions such as being (more or less) moral. Durkheim’s theorizing revealed much about Identity verification occurs when individuals’ macro-level moral processes. However, at the perceptions of who they are in situations cor- micro level, he did not provide insight into respond to their identity standard. Non-verifi- why some people behave morally while oth- cation between their identity standard and ers do not. Not everyone behaves morally, so who they are in situations generates negative what explains this variation? The answer may emotions. Additionally, the identity verifica- reside in how individuals see themselves in tion process occurs within settings that are moral terms and behave accordingly. framed as (more or less) relevant to one’s Like Durkheim, Goffman (1967) empha- identity. Identity theorists have not thor- sized the importance of ritual activity. Goffman oughly examined the framing of situations. saw interaction as characterized by rituals of This study investigates how the internal iden- openings and closings, entrances and exits, tity process and framing of surrounding situ- and corrections for deviant behavior. Interac- ations are associated with moral behavior and tion is also composed of particular frames moral feelings. (Goffman 1974). Frames provide meaning for individuals during interaction, organizing and determining appropriate roles and behaviors Background to be enacted. Unlike Durkheim, Goffman Broadly speaking, morality represents cul- emphasized the self in interactions (Goffman tural codes that specify what is right or 1959). His analysis of the self shows how wrong, good or bad, or acceptable or unac- individuals work at presenting a strategic ceptable in a society (Turner 2010; Turner image of themselves to others, in a manner and Stets 2006). Traditionally, as seen in the similar to how actors play out a script on work of Durkheim, Goffman, and Collins, the stage (Goffman 1959). In playing their part, sociology of morality emphasizes how moral- individuals inform others that they are adher- ity binds people together through a common ing to the cultural script. When individuals system of rules and expectations during inter- commit blunders and role playing fails, they actions. Durkheim (1965) revealed how feel embarrassed and modify their behavior, morality emerges from the collective effer- realigning actions with cultural expectations vescence that occurs when people congregate that preserve social interaction. In Goffman’s in groups. In his studies of Aboriginals, he work, actors are attuned more to cultural and noted that religious and moral practices stem moral codes than to their identities and stable from ritual worship of cultural symbols self-views. Individuals are motivated to (totems) that are attributed a sacred quality behave in a certain way to effectively play out and arouse strong emotions in interaction. a cultural script rather than—as we will The power of cultural symbols resides in their argue—to verify their identities. ability to define appropriate moral behavior. Collins (2004) extends these analyses of Actions conforming to these expectations are ritual interaction. Ritual interactions include virtuous; actions violating these expectations elements such as the physical co-presence of cause outrage. individuals, a common focus of attention, a For Durkheim (1961), morality controls common emotional mood, symbolic represen- and integrates members of a society and tations of a common focus, and a sense of thereby generates social solidarity. Society is moral virtue about symbols that mark group 122 American Sociological Review 77(1) membership. As ritual activity continues in an moral principles, affects moral behavior. For interaction, emotions are aroused and build. two decades, Blasi advanced an understand- Repeated and highly rhythmic ritual activity ing of the role of the self and identity in moral among group members produces collective action, and this theory has gained appeal effervescence—including positive emotions among many psychologists (e.g., Aquino et and moral feelings—that generate group soli- al. 2009; Aquino and Reed 2002; Frimer and darity. Walker 2009; Hardy 2006; Lapsley and Like Durkheim, Collins takes seriously the Narvaez 2004; Narvaez and Lapsley 2009). role of emotions in the moral order. He main- Recent work involves studying how central or tains that the motive underlying behavior in important the moral identity is in determining interaction is to experience and maximize moral action (Aquino et al. 2009; Frimer and emotional energy. Positive emotional energy Walker 2009). mobilizes individuals to initiate interactions; In the current research, we present a model negative energy fails to instigate interactions. of moral action that examines the cognitive Collins sees interactions that provide the most (i.e., identity) and affective (i.e., emotional) emotional energy as fostering the moral order, parts of the self. Indeed, Blasi (1999) but we will argue that the self and identity acknowledged that moral emotions motivate help build and maintain the moral order moral behavior and the moral identity. For through moral actions and moral emotions. example, empathy motivates altruistic behav- This is consistent with Weber’s ([1922] 1968) ior whereas shame and guilt inhibit harmful view that we need to examine patterns of behavior. More recently, psychologists (Haidt individuals’ actions—patterns that form the 2001; Hardy 2006; Hoffman 2000) and neu- basis of social structures. roscientists (Greene et al. 2001) have been Weber (1978) saw moral values as residing investigating the relationship between moral within individuals. People carry beliefs about judgments, moral behavior, and emotions. what is good/bad or right/wrong, and these However, to advance the study of morality, beliefs influence their behavior. Weber we need a theoretical model that explicitly believed that scientists can investigate indi- links the cognitive, behavioral, and emotional viduals’ internal meanings about morality and dimensions of the self; and we must then put their corresponding actions, but that scientists that theoretical model to the test. We do that cannot determine whether certain moral in this research, and we use the identity model beliefs ought to exist. In the same way Weber in sociology. conceived moral values as subjective and In identity theory (Burke and Stets 2009; inspiring action, we see moral identity as Stryker [1980] 2002), when people have a guiding and regulating action and emotions moral identity, being (more or less) moral will within and across situations. influence their behavior in a situation. Other Psychologists rather than sociologists have individuals will then react to their behavior. looked to the individual as the source of When individuals think others interpret their moral action. For years, moral psychology behavior in the same way they intended, they addressed the question of how moral reason- feel good. When individuals think others ing is related to moral behavior (Kohlberg interpret their behavior differently (as either 1981; Piaget [1932] 1965). Over time, schol- less or more moral than how it was intended), ars discovered that a strong relationship they feel bad. These ideas are consistent with between advanced moral reasoning and moral Blasi’s early idea on the link between the behavior does not exist (Blasi 1980). Instead, moral identity and moral action through self- scholars saw a moral identity, rather than consistency (i.e., matching behavior to people’s moral reasoning, as crucial (Blasi 1984). How identity) (Blasi 1984). It also incorporates his people see themselves along the moral dimen- later idea that emotions play a role in moral sion, rather than their adherence to particular functioning (Blasi 1999). Stets and Carter 123

Figure 1. Identity Model

Some situations may be defined in moral the study of morality, an articulated and testable terms to a greater degree than others. For theory that integrates the moral identity, behav- example, attending a church service may make ior, and emotions is needed, as well as a consid- morality more relevant than attending a party. eration of the situations within which they are Recently, psychologists have studied how embedded. For this, we turn to identity theory. priming morality (e.g., having individuals review the list of Ten Commandments) may activate a moral self-view, and in turn, moral Identity Theory behavior (Aquino et al. 2009). However, we In the control systems approach of identity the- need a theoretical model that integrates the ory, when an identity is activated in a situation, (internal) moral identity process with the a feedback loop is established (see Figure 1) (external) moral relevance of a situation. Iden- (Burke and Stets 2009). This loop has six com- tity theory provides this integration. ponents: (1) the identity standard (the meanings In general, while the study of morality of an identity), (2) output (behavior), (3) input appears in early sociological writings, there is (how people think others see them in a situation no adequate conceptualization of the self, which [i.e., reflected appraisals]), (4) a comparator is needed to theorize about variability in moral (which compares the input with the identity action and moral feelings. Goffman comes standard), (5) emotion (that results from the close, but we find his research on frames, which comparison process), and (6) situation mean- provides meaning as to how a situation should ings (which vary in the degree of correspon- be interpreted, to be most relevant. While psy- dence with identity standard meanings). We chologists have attended to self and identity in apply each component to the moral individual. 124 American Sociological Review 77(1)

Identity standard. The identity standard enterprise, as when our commitment and is the moral identity. This standard contains the responsibility to others generates a “we- meanings an individual associates with being a feeling” and a collective sense of self. In study- moral person. Meanings are individuals’ ing justice and care, we do not claim that they responses when they reflect upon themselves characterize all meanings of morality, only that (Burke and Stets 2009). For example, a woman they represent important meanings. may consider herself submissive when she thinks about how dominant she is, she may see Output. When an identity is activated in a herself as efficient when she thinks of herself situation, it guides behavior (Burke and Stets as a worker, and she may define herself as 2009; Stryker and Serpe 1982). More specifi- dependable in her church group. Submissive- cally, meanings of the behavior (output) ness, efficiency, and dependability are the should correspond to meanings held in the meanings that help define her, and she will identity standard. For example, an individu- control these self-meanings when interacting al’s self-meanings of being more just and with others so that they are maintained at a caring should be related to behavior in a situ- level she sets (whether high or low). ation that carries meanings of being fair and For the moral person, meanings of justice supportive to others. Alternatively, self-mean- and care are implicated (Gilligan 1982; Haidt ings of being less just and caring should be and Kesebir 2010; Kohlberg 1981). While most associated with behavior that carries mean- scholars agree that justice and care are funda- ings that the person is acting unjust and is not mental to being moral, some argue that morality being helpful. On average, moral behavior is goes beyond justice and care and includes a conduct instilled with the expectation to do broader spectrum of meaning. For example, what is right or good based on social consen- some scholars have identified cross-cultural sus, such as behaving in a just and caring moral meanings of being autonomous, commu- manner (Turner and Stets 2006). Given this, nity-oriented, and acknowledging the divine our first hypothesis is as follows: (Shweder et al. 1997). Others have identified the meaning dimensions of harm/care, fairness/reci- Hypothesis 1: The higher a person’s moral iden- procity, in-group/loyalty, authority/respect, and tity score, the more likely the person will be- purity/sanctity as moral foundations (Haidt and have morally. Graham 2009; Haidt and Kesebir 2010).1 We focus on the degree to which individu- Input. Input consists of situational mean- als are just and caring for several reasons. ings that influence how a person behaves and First, all cultures appear to promote moral feels. These meanings have two sources: (1) values based on being just and caring (Hauser reflected appraisals and (2) the definition of 2006). Second, our sample, while ethnically the situation. First, when a person enacts diverse, shares values associated with West- behavior, others react to the meaning of that ern culture; we thus follow Western concep- behavior, and the person interprets this reactions of morality that tend to value protecting tion. This is reflected appraisal. It is how a individuals’ welfare and autonomy (Haidt and person thinks others perceive her in the situa- Graham 2009). While this minimizes an tion. For example, a woman may see herself approach to the moral person that emphasizes as just and kind (meanings in her moral iden- ties to groups—which may describe many tity standard). She may enact behaviors non-Western countries—justice and care do consistent with these meanings, such as treat- not exclude the collective. Justice can be a ing another person fairly and being supportive. collective enterprise, particularly when it is Others in the situation may react to her behav- imbued with meanings of cooperation and ior by telling her that she is fair and caring.2 consensus within a group (Hegtvedt and She will then interpret this reaction as consis- Scheuerman 2010). Care can also be a collective tent with how she sees herself. Stets and Carter 125

A second source of input is how one intensity to signal the meaning that she is defines a situation. In defining a situation, caring. If the same person is perceived as a 10 one interprets the setting as containing mean- in terms of care, she might reduce her display ings that characterize a particular social occa- of caring behavior toward others because sion or frame of interaction (Goffman 1974). others’ perceptions reveal excessive caring Embedded within this frame of interaction are compared to her identity standard.3 meanings related to goals to be achieved, Identity theorists typically examine gener- roles to be played, identities to be verified, alized positive and negative emotions to the behaviors to be enacted, and feelings to be neglect of specific emotions. We study two expressed. Individuals use these definitions specific moral emotions: guilt and shame. of situations to guide behavior and feelings. While many feelings might constitute moral emotions (Haidt 2003; Turner and Stets Comparator. Reflected appraisals and 2006), most research focuses on guilt and definitions of situations are continuously fed shame (Tangney and Dearing 2002b; Tang- into the comparator. The comparator relates ney, Stuewig, and Mashek 2007). Shame is input meanings from both sources with stored more intensely felt than guilt. Shame is tied to identity standard meanings. It produces an violations of moral standards in which the error signal, which is the difference between entire self is evaluated negatively, that is, the the input and the identity standard meanings. whole self is judged responsible for a viola- In thinking about this numerically, if an iden- tion. Individuals reflect upon themselves as tity standard is set at 7 (on a scale of 0 to 10) horrible agents. They feel worthless to others for being caring, and others indicate that the and want to hide, escape, or strike back. Guilt person is acting 7 in terms of being caring, involves behavior that is judged as bad. there is a perfect match between input and Rather than focusing on a bad self as in identity standard meanings. This is identity shame, guilty people focus on their bad verification. However, if others perceive the behavior, allowing a good self to still exist. person as acting a 5 in terms of being caring, Guilt leads people to feel remorse and regret, there is a mismatch between input meanings and this motivates them to confess, apologize, and identity standard meanings. This is iden- and repair wrongdoings. The distinction tity non-verification. Similarly, if meanings in between shame and guilt can be summarized a situation are defined in terms of something as follows: someone feeling shame would other than what is relevant for an identity, a say, “I did that bad thing”; someone feeling discrepancy emerges between identity mean- guilt would say, “I did that bad thing.”4 ings and situation meanings. For example, if Theoretically, we are interested in the meanings in a situation are about acquiring commonalities between guilt and shame.5 wealth rather than behaving morally, one will Both are intimately social because they have difficulty verifying the moral identity. involve others who may witness a person’s actions and evaluate them. When feedback Emotion. In identity theory, identity veri- from others is discrepant with a person’s self- fication is associated with positive feelings view, this discrepancy will be associated with and identity non-verification is related to neg- feelings of shame for not living up to one’s ative feelings (Burke and Stets 2009). own standards as well as guilt for having Negative emotions drive the identity control done something, the meanings of which sig- system to reduce the inconsistency by behav- nal an inconsistency with one’s identity stand- ing differently so as to change meanings in the ard meanings. Ultimately, guilt and shame situation. In the earlier example, if others per- keep people integrated into society through ceive the person’s level of caring as a 5 rather internal monologues with the self and feed- than a 7, the person might take the time to back from others (Turner 2010).6 This leads help others more often and with increased to our second hypothesis: 126 American Sociological Review 77(1)

Hypothesis 2: When there is a discrepancy be- moral filter, defining what actions are appro- tween one’s moral identity standard mean- priate in a situation. In identity theory, peo- ings and how moral a person thinks others ple’s morality is the meanings in their moral see him or her in a situation, the person will identity. Moral context is how individuals be more likely to report moral emotions. interpret a situation as containing moral meanings. If individuals perceive meanings Situation meanings. Identity theorists of good/bad or right/wrong behavior, a situa- have not closely examined how people’s defi- tion is high in moral content because there are nitions of situations influence behavior and moral meanings in the situation to attend to. emotions in those situations. Rather, they In turn, this should influence moral behavior. have focused on how changes in a situation This leads to our third hypothesis: may change the meaning of an identity. For example, researchers have examined how the Hypothesis 3: The more a person defines a birth of a child changes one’s gender identity, situation as containing moral meanings, the with women becoming more feminine and more likely the person will behave morally. men becoming more masculine (Burke and Cast 1997). Alternatively, scholars have stud- Situation meanings also carry an affective ied how salient identities are maintained in aspect in the form of feeling rules that specify the face of situational changes, such as retain- emotions individuals ought to experience, ing an athlete identity when one goes to such as feeling sad at a funeral or happy at a college (Serpe and Stryker 1987). party (Hochschild 1983). Feeling rules, or We study the degree to which a situation is emotion norms, indicate the direction of an interpreted as relevant to moral considera- emotion (positive or negative), its intensity tions because this interpretation may be (from strong to weak), and its duration (from related to moral action and moral feelings. We fleeting to lasting). They are not codified for- label this the cognitive (interpretive) aspect of mally but are learned and reinforced in inter- situation meanings. Individuals likely draw action. Like the cognitive aspect of moral on shared symbols and definitions derived situations, we expect people to draw on cul- from culture to identify the degree to which tural expectations about emotions that ought situations contain moral meanings. For exam- to be experienced when moral rules are pres- ple, deciding whether a friend should drive ent. The feeling rules in moral situations home drunk or whether to donate to a charity involve the idea that individuals should feel are more likely to be interpreted as containing guilt or shame when moral codes are violated. moral meanings than deciding between two Knowledge of these feeling rules may redi- different detergent brands. rect a person away from immoral behavior to Researchers are beginning to investigate avoid negative feelings. the influence of situational factors on moral Our two aspects of moral situations—cog- behavior (e.g., moral priming activating a nitive and affective—are similar to Hochs- moral self-view) (Aquino et al. 2009), com- child’s (1979) discussion of framing rules and peting meanings in situations (e.g., financial feeling rules in situations. Framing rules indi- incentives weakening moral behavior), and cate which interpretations and meanings indi- the presence of moral exemplars strengthen- viduals should give to situations, and feeling ing moral behavior (Aquino and Freeman rules specify how individuals ought to feel 2009). Wikstrom’s (2010) Situational Action given the particular interpretation made by Theory (SAT) of moral action is particularly the framing rules. Consequently, we expect relevant. Wikstrom addresses people’s moral- the following: ity as well as the moral setting and argues that both are relevant in guiding moral actions. Hypothesis 4: The more a person reports One’s morality and moral context serve as a that one should experience moral emotions Stets and Carter 127

following immoral behavior, the more likely to work to verify their level of morality in the the person will behave morally. situation. Under the guidance of feeling rules, individuals should be more likely to feel guilt Finally, we anticipate that a person attuned or shame when they behave inconsistently. to moral meanings and feeling rules will be Thus, while the identity process involves more inclined to experience moral emotions individuals acting to control perceptions of when enacting immoral behavior. This indi- themselves in situations to match their moral cates an interaction effect. Awareness of identity standard, this process exists in an meanings along the moral dimension in situa- environment that carries varying degrees of tions and the corresponding feeling rules moral meanings and emotion norms about associated with moral behavior should make morality. individuals more attentive to their behavior. A We test these assumptions on more than violation in cultural expectations of behavior 350 students at a large southwestern univer- should produce guilt or shame. Therefore, we sity. Different moral issues may be relevant to hypothesize the following: students than to non-students or older adults. However, as is clear in identity theory, one Hypothesis 5: The more a person defines a situ- should study meanings as they are understood ation as containing moral meanings but be- by the people being investigated. Given our haves immorally, the more likely the person sample, we examine moral issues that are will report moral emotions. relevant in students’ lives.

Hypothesis 6: The more a person reports that one should feel moral emotions for behav- Method ing immorally and engages in immoral be- We recruited participants for this study from havior, the more likely the person will report undergraduate sociology classes at a large moral emotions. southwestern university during the 2007 to 2008 academic year. Classes were a mixture Summary. We see individual variation in of lower and upper division as well as general moral behavior and moral emotions arising education and elective courses. For their par- from two sources. First are the moral mean- ticipation, we offered respondents extra credit ings in the identity standard. Identity standards in their classes and a chance to win $100 in a are always arrayed on a continuum of mean- lottery. The response rate was 86 percent for ing. Having a moral identity does not mean a total of 369 individuals. that one has meanings of being a good person; This was a two-part study. First, partici- rather, one has meanings that fall within a pants responded to a survey that measured range, for example, of being very uncaring their moral identity, moral behavior, and and unjust to very caring and very just. The moral emotions. Three months later, they goal is to live up to one’s self-view, however responded to a second survey that measured that view is arranged across the moral contin- moral meanings and feeling rules in situa- uum. When the meanings of one’s behavior tions. We delayed administering the second based on feedback from others are inconsis- survey to reduce the potential influence the tent with the meanings in one’s identity first survey might have had on their responses. standard, the person will feel bad. All respondents completed both surveys, so A second source of individual variation is there was no attrition. situation meanings that have cognitive and In the first survey, we obtained a measure of affective components. When individuals respondents’ moral identity. Respondents identi- interpret situations as moral and are attentive fied their behavior and emotions across a variety to the corresponding feeling rules for violat- of situations in which they had the choice to do ing moral codes, they should be more likely the good or right thing (see the Appendix). 128 American Sociological Review 77(1)

Respondents were asked to think about the last Table 1. Percentage of Respondents Report- time they were in each situation and report what ing Moral Behavior they did and how they felt following their Moral actions. If they never experienced the situation, Situation Behavior (%) they were to imagine themselves in the situation and respond as to what they would do and how Did not copy 70 they would feel following their actions. Our Did not drive drunk 82 Did not take item 88 analyses include only respondents who experi- Gave to charity 57 enced the situation. Most respondents reported Did not allow student to copy 59 experiencing each situation: (1) copied a stu- Did not allow friend to 87 dent’s answers (75 percent), (2) drove home drive drunk drunk (42 percent), (3) took an item (66 per- Returned lost item 83 cent), (4) gave to charity (93 percent), (5) Returned money to cashier 47 allowed a student to copy one’s answers (75 percent), (6) let a friend drive home drunk (67 percent), (7) returned a lost item (84 percent), and (8) returned money to a cashier (78 per- Table 1 reports the percentage of respond- cent). In the second survey, respondents were ents who behaved morally across the situa- asked how they would rate each scenario in the tions. The relatively high percentages Appendix in moral terms, and how they thought reporting good actions suggest that respond- individuals ought to feel given particular courses ents may be portraying themselves in a of action in each situation. socially desirable way. To investigate this, we The scenarios were a useful methodologi- consulted data on these same individuals cal procedure to get individuals to respond to from a laboratory study that simulated a test- a common set of experiences. To ensure sce- ing situation where participants had the narios represented situations individuals opportunity to cheat to get a higher score on a likely experienced, we administered a pretest test (Stets and Carter 2011). If respondents a year earlier to approximately 150 under- answered truthfully in the survey, individuals graduates in similar classes. Respondents who reported they were more likely to copy a identified three recent situations they experi- student’s answers in the survey would be enced where they had a choice between doing more likely to cheat in the laboratory study. the right or wrong thing. They described the This is what we found. The odds of cheating situation in detail, including what choice they in the laboratory study increased by 30 per- made and why. For this study, we used the cent for respondents who reported copying a situations most frequently listed. We thus student’s answers in the survey compared to used situations relevant to the population those who had not reported copying (odds from which we drew our sample. Past research ratio = 1.30, p < .01). While not definitive, has tended to rely on a class of moral dilem- these findings suggest our respondents were mas known as the trolley car problems to not simply answering in a socially desirable assess individuals’ moral decision making manner. (Greene et al. 2001; Hauser 2006). Some scholars argue that we need more research on real-life dilemmas (Walker and Pitts 1998; Measures Walker et al. 1995), which is what we have Moral Identity attempted here. We do not claim that the scenarios we study represent the full domain of For the moral identity, we examined how moral situations for all groups, only that they people view themselves along the justice and are situations students typically identify as care dimensions. We gave respondents a list moral situations in their everyday lives. of 12 bipolar characteristics used in prior Stets and Carter 129 studies: honest/dishonest, caring/uncaring, Table 2. Principle Components Factor unkind/kind, unfair/fair, helpful/not helpful, Analysis of the Moral Identity stingy/generous, compassionate/hardhearted, Items Factor Loading untruthful/truthful, not hardworking/hardworking, friendly/unfriendly, selfish/selfless, Honest .59 Caring .65 and principled/unprincipled (Aquino and Kind .67 Reed 2002; Stets and Carter 2006; Walker Fair .58 and Hennig 2004). Respondents were told to Helpful .63 think about what kind of person they thought Generous .55 they were for each pair of characteristics and Compassionate .54 to place themselves along a continuum Truthful .62 between the two contradictory characteristics. Hardworking .54 A value of 1 reflected agreement with one Friendly .56 bipolar characteristic, 5 reflected agreement Selfless .57 with the other characteristic, and 3 placed the Principled .52 respondent halfway between the two. Eigenvalue 4.14 This measurement procedure used to capture the meanings of many identities (Burke Ω .87 and Stets 2009) follows the Osgood measurement technique of using a semantic differential to measure meaning (Osgood, Suci, and Tan- nenbaum 1957). Characteristics thought to untruthful/truthful, not hardworking/hard- capture relevant underlying meaning dimen- working, and principled/unprincipled. The sions of an identity are placed on a scale of correlation between the two scales was .60. polar opposites, and individuals respond to the This suggests that care and justice are two characteristic as they take themselves as an facets of an underlying phenomenon: moral- object. These characteristics are not traits, ity. From an identity perspective, a care iden- which are usually conceptualized as habitual tity and a justice identity are very close in dispositions to act in a certain way that are semantic space, such that measuring one acquired through learning. Rather, they are identity would capture important meanings in meanings or sets of responses people provide the other. Measuring both justice and care when they think about themselves (Burke and thus provides a more complete and accurate Stets 2009). Individuals control these mean- measure of individual morality. ings at a particular level. Traits are not subject Because identity meanings should corre- to control in the same way that people manage spond to behavior meanings, we also exam- the particular level of meaning of an identity. ined whether behaviors in the scenarios were The idea of individuals managing the level of differently associated with the care and justice identity meanings is a core assumption in the identities. Both identities predicted behavior self-regulating identity model. equally well in half of the scenarios. In the To examine whether the 12 bipolar items other half, the care identity was a slightly bet- characterize care and justice, we rotated the ter predictor, but no clear pattern emerged for first two factors of a factor analysis using why this occurred.7 Taken together, these anal- oblique rotation. Factor one represented the yses suggest that care and justice are empiri- following items we labeled care: caring/ cally inseparable. Future research should uncaring, unkind/kind, helpful/not helpful, examine whether a care identity influences stingy/generous, compassionate/hardhearted, behaviors that are distinguishable from a jus- friendly/unfriendly, and selfish/selfless. Fac- tice identity as this is a limitation of the current tor two consisted of the following items we study. In this study, we used the factor structure labeled justice: honest/dishonest, unfair/fair, from the first unrotated factor. Table 2 presents 130 American Sociological Review 77(1)

Figure 2. Histogram of Moral Identity Scores factor loadings. We reverse coded negatively (+5). We transformed these response catego- worded characteristics and summed items with ries into a 0 to 10 scale (0 = Extremely a high score representing a more moral com- Immoral to 10 = Extremely Moral). We pared to less moral identity. For the analysis, reverse coded negatively worded characteris- we standardized the scale (mean = 0; standard tics. The mean was high (mean = 7.52; stand- deviation = 1). ard deviation = 1.22); thus, respondents In Figure 2, we provide a histogram of identified these items as strongly character- the moral identity scores in the original 1 to izing morality. 5 metric units. Most people have a moral identity above the midpoint of the scale; Moral Identity Discrepancy that is, most people identify themselves as relatively good or moral individuals. Even The moral identity discrepancy measure com- the lowest moral identity scores are at least pares individuals’ moral identity standard with at the midpoint of the moral–immoral how they think others see them (i.e., reflected dimension. Throughout our analyses, hav- appraisals) along the morality dimension to ing a more moral or less moral identity is determine the degree of difference (Burke and thus relative. Respondents with a lower Stets 2009). Operationally, this is individuals’ score are by no means immoral but simply moral identity standard minus how moral they lower on this scale compared to others in the think others see them in each situation. After sample. individuals reported how they behaved and felt Finally, we examined whether individuals in each scenario, they reported how much they in our sample identified the 12 bipolar items thought others saw them as being likeable, intel- as characteristic of morality. In the second ligent, moral, dominant, spiritual, attractive, survey, respondents were asked to rate each reliable, loving, and competitive (from “Not at of the characteristics in terms of how moral all” to “Very Much” [coded 0 to 6]). Responses they thought each item was, however they to how they thought others saw them in moral defined morality. Response categories ranged terms served as the reflected appraisals mea- from Extremely Immoral (–5) to Neither sure. Other characteristics were filler items to Immoral nor Moral (0) to Extremely Moral mask our interest in the moral dimension. We Stets and Carter 131 standardized the moral identity measure and the morally meaningful. Half the scenarios were reflected appraisal measure for each situation. worded in terms of doing the bad action, and Then, we subtracted participants’ standardized the other half were worded in terms of doing reflected appraisals measure for each situation the good action. Response categories ranged from their standardized moral identity measure. from “Extremely Immoral” to “Extremely We squared each value so that a departure from Moral” (coded –5 to 5). When respondents zero in either a negative or positive direction evaluated good actions as extremely moral meant an increased discrepancy.8 (+5) or bad actions as extremely immoral (–5), we took the absolute value of this variable to capture the potency of the moral Moral Behavior imperative. A high absolute value represents a For moral behavior, we used respondents’ situation that had very high moral relevance. reports of their conduct in each of the eight A low absolute value represents a situation situations. We coded this 0 to 1, with 1 that is neither moral nor immoral, that is, reflecting moral behavior, and 0 reflecting morality is not relevant in the situation. We immoral behavior. Across the scenarios in the coded the variable from 0 (morality is not Appendix, we coded the following behaviors relevant) to 5 (morality is highly relevant).10 as 0: copied a student’s answers (#1), drove Respondents were presented these scenar- home drunk (#2), took an item (#3), did not ios and asked to identify how someone ought provide a donation (#4), let a student copy to feel following particular actions. Again, one’s answers (#5), let one’s friend drive half the scenarios were worded in terms of home drunk (#6), did not return a lost item doing the bad action, and the other half were (#7), and did not return money to the cashier worded in terms of doing the good action. We (#8). We coded the alternative to each of the used the previously described six emotion above behaviors as 1. measures. We examine individuals’ responses to guilt and shame. This measure operational- izes emotion norms in a situation because Moral Emotions individuals reported the degree to which peo- After participants reported how they behaved in ple should feel guilt and shame following a each situation, they reported how they felt. The bad or immoral behavior. The correlation survey listed the following six emotions: happy, between guilt and shame was high (r = .95, p fearful, angry, sad, shameful, and guilty. < .01). We summed the two items to create an Happiness, fear, anger, and sadness are four overall emotions norm variable. primary emotions, and shame and guilt are two critical secondary, moral emotions (Tangney et al. 2007; Turner and Stets 2006). Response cat- Analysis egories for each emotion ranged from “Not at Because participants’ moral behavior in the all” to “Very Intense” (coded 0 to 6). We focus survey is coded 0/1, we estimate this equation on the moral emotions of guilt and shame. The using logistic regression. We provide the odds correlation between respondents’ guilt and ratio for a one-unit change in the independent shame was high (r = .90, p < .01). We thus variable. We estimate moral behavior to be a summed the two items with a higher score mea- function of a higher score on moral identity suring more intense moral emotions.9 (Hypothesis 1), reports of moral framing rules (Hypothesis 3), and reports of moral feeling rules (Hypothesis 4). Because the moral emo- Situation Meanings: Moral Meanings tion variable is continuous, we estimate this and Feeling Rules equation using ordinary least squares regres- In the second survey, respondents were asked sion. We anticipate that individuals will be more to think about each scenario and indicate the likely to report guilt and shame when they extent to which they saw the situation as experience a moral identity discrepancy 132 American Sociological Review 77(1)

Table 3. Means, Standard Deviations, and Ranges of Variables Variable Mean SD Min Max

Sex .33 .47 0 1 Moral Identity .00 1.00 −3.38 1.83 Moral Behavior .66 .47 0 1 Moral Emotions .00 1.00 −.67 2.79 Moral Identity Discrepancy .00 1.00 −.69 7.23 Moral Meanings .00 1.00 −1.81 1.12 Feeling Rules .00 1.00 −1.22 1.24

(Hypothesis 2), define a situation as moral but (33 percent) participated. While the gender behave immorally (moral behavior × meanings) distribution is different from the university (Hypothesis 5), and when they think people from which this sample is drawn (52 percent should experience moral emotions for immoral women, 48 percent men), other characteris- behavior and behave immorally (moral behav- tics of our sample are similar to the university ior × feeling rules) (Hypothesis 6). population. For example, respondents are The unit of analysis in this study is an ethnically diverse (13 percent White, 33 per- individual’s response to each scenario. cent Latino/Chicano, 29 percent Asian, and Because participants could respond to eight 25 percent other) as is the university popula- different scenarios, there are eight responses tion (17 percent White, 29 percent Latino/ per person, or a possible 2,952 responses. In Chicano, 40 percent Asian, and 14 percent standard regression models, errors for indi- other). Average age in this sample is 21, viduals are assumed to be uncorrelated. which is the average age of students at this Because individuals are responding to multi- university. Average parental income is ple scenarios, errors for these scenarios are $35,000 to $49,000, which is somewhat lower assumed to be correlated. In our estimation than the average university student ($48,000 procedure, we used the cluster option in Stata to $96,000). to take into account these correlated errors. Table 4 presents zero-order correlations of We controlled for several key background the variables. To correct for individuals’ characteristics that might influence moral responses to multiple scenarios, we use bon- behavior and moral emotions, including ferroni significance levels. Table 4 shows that respondents’ sex, race/ethnicity, age, and moral behavior is associated with a high score income. We estimated the main effects of these on moral identity (r = .14, p < .05) and a low demographic factors as well as their interaction score on moral emotions of shame and guilt (r effects. Across the moral behavior and moral = –.58, p < .05). Moral emotions also are emotions equations, sex was the only variable of positively related to a discrepancy between significance, and it was only significant for the people’s moral identity standard and how moral emotions equation.11 Therefore, we pre- they think others see them along the morality sent findings with sex (coded 0 for female and 1 dimension (r = .30, p < .05). Not all people for male) included in this equation. with a high score on moral identity defined situations as containing moral meanings (r = .15, p < .05). A great deal of individual varia- Results tion exists in interpreting situations as con- Table 3 presents means and standard devia- taining moral meanings. However, when tions for the variables. Most variables have situations are defined as containing moral been standardized (mean = 0; standard devia- meanings, individuals are more likely to tion = 1). More women (67 percent) than men behave morally (r = .24, p < .05). Stets and Carter 133

Table 4. Correlations among Variables (1) (2) (3) (4) (5) (6) (7)

(1) Sex 1.00 (2) Moral Identity −.10 1.00 (3) Moral Behavior −.07 .14* 1.00 (4) Moral Emotions −.03 −.07 −.58* 1.00 (5) Moral Identity Discrepancy .05 −.09 −.25* .30* 1.00 (6) Moral Meanings −.04 .15* .24* −.10 −.04 1.00 (7) Feeling Rules −.07 .04 .11* .12* .01 .14* 1.00

*p ≤ .05; **p ≤ .01 (two-tailed tests).

Table 5. Logistic and OLS Regression Analyses of Moral Processes on Moral Behavior and Moral Emotions (N = 369) Dependent Variables

Moral Behavior Moral Emotions

Independent Variables Odds Ratio β

Sex −.35* Moral Identity 1.28** .02 Moral Behavior −.43** Moral Identity Discrepancy .13** Moral Meanings 1.61** .18** Feeling Rules 1.19** .42** Moral Behavior × Moral Meanings −.32** Moral Behavior × Feeling Rules −.27** Sex × Feeling Rules −.08* Sex × Moral Behavior .31** R2 .06** .45**

*p ≤ .05; **p ≤ .01 (two-tailed tests).

Table 5 presents analyses for our hypotheses. (Hypothesis 3). Finally, a one standard devia- In column 1, the moral identity is positively tion increase in being aware of the feeling rules associated with enacting moral behavior (odds of moral situations increases the odds of behav- ratio = 1.28, p < .01). The odds of doing a good ing morally by almost 20 percent (odds ratio = thing increase 28 percent for a one standard 1.19, p < .01) (Hypothesis 4). In general, iden- deviation increase in moral identity. This sup- tity meanings and situation meanings signifi- ports Hypothesis 1 and is consistent with the cantly and independently relate to moral idea that meanings in one’s identity standard behavior. influence behavior that implies the same mean- Column 2 of Table 5 presents results for ing. Furthermore, and independently, interpret- reports of guilt and shame.12 The moral identity ing situations as containing moral meanings is not associated with moral emotions directly; increases the odds of behaving morally by at however, there is an indirect association by way least 60 percent (odds ratio = 1.61, p < .01) of moral behavior. Thus, a mediation model is 134 American Sociological Review 77(1)

Table 6. Means of Moral Emotions by Moral Results in Table 5 also reveal that when Behavior and Moral Meanings individuals engage in bad behavior, and they Moral Behavior report that people should feel guilt and shame for immoral behavior, it is associated with an Moral Meanings No Yes increase in reports of moral emotions (β = 0 .46 −.28 .42, p < .01) (Hypothesis 6). When individu- 1 .58 −.28 als enact good behavior, this feeling rule is 2 .82 −.38 associated with a decrease in reports of moral 3 .94 −.38 emotions (β = [.42] + [–.27] = .15, p < .01). 4 1.05 −.47 Thus, it is not only morally charged situations 5 1.13 −.49 combined with moral behavior that is linked to reduced reports of moral emotions, but also the feeling rules associated with these situations that, when combined with moral behav- operating. The moral identity is related to moral ior, are related to a decline in reports of moral behavior, and moral behavior, in turn, is linked emotions. to moral emotions. We also find that a discrep- We highlight additional findings in our ancy in the moral identity is related to an moral emotions equation. First, reports of increase in reports of moral emotions (β = .13, good behavior are negatively associated with p < .01) (Hypothesis 2). reports of guilt and shame (β = –.43, p < .01). When individuals define a situation in We also find two significant sex interaction moral terms and engage in moral behavior, effects: sex × feeling rules and sex × moral we see a decrease in reports of guilt and behavior. The first interaction indicates that shame (β = [.18] + [–.32] = –.14, p < .01) when considering the degree to which one (Hypothesis 5). Table 6 shows the interaction should feel bad for immoral behaviors, men more clearly. Among respondents who are less likely than women to report moral defined a situation with strong moral mean- emotions for average levels in feeling rules ings, there is a large difference in the experi- (β = –.35, p < .01). For one standard deviation ence of moral emotions given moral and increase in feeling rules, women report more immoral behavior. Respondents who acted moral emotions than do men (β = .42 [women] immorally when the situation was defined as versus β = [–.35] + [.42] + [–.07] = 0 [men]). having strong moral meanings had a large Feeling rules are thus related more to wom- increase in the average level of moral emo- en’s than to men’s reports of moral emotions. tions, while those who acted morally had a The second interaction reveals that men are strong reduction in the average level of moral less likely than women to report moral emo- emotions. This difference diminishes as the tions for immoral behavior (β = –.35, p < .01). perceived level of moral meanings in the situ- We find no gender differences in reports of ation diminishes. Even at the lowest level of moral emotions for enacting moral behavior moral meanings, where respondents did not (β = [–.35] + [–.43] + [.31] = –.47 [men] and view the situation in moral terms, there is still β = –.43 [women]). a significant difference in the level of moral In general, our hypotheses are supported. emotions between respondents who did and Consistent with identity theory, the moral did not engage in moral behavior. We must identity is positively associated with moral remember, however, that almost all respond- conduct. Additionally, situations imbued with ents have a level of moral identity above the moral meanings along the cognitive and midpoint on the scale. This is apparently affective dimensions are associated with enough to make individuals feel bad for enactment of moral behavior. We find that behaving immorally even in situations not reports of moral emotions are less likely to seen to have strong moral significance. occur when individuals engage in moral Stets and Carter 135 behavior, interpret a situation as morally meanings are the cultural milieu in which the meaningful, and are attentive to emotion moral identity process operates. norms that shore up negative emotions for The moral identity standard locates indi- immoral behavior. Additionally, reports of viduals along a range of meanings from low to moral emotions are positively related to iden- high. Wherever individuals are located on this tity non-verification. These moral emotions continuum, they act with the goal of verifying emerge not only when individuals claim a their identity meanings. We found that indi- relatively high moral identity and then per- viduals with a high moral identity score were ceive that others do not agree that they are more likely to behave morally, while those behaving in a good manner, but also when with a low moral identity score were less likely individuals claim a relatively low moral iden- to behave morally. However, it is not simply tity and find that others do not view their moral identity meanings that guide behavior, behavior as bad. Both ends of the identity but the relationship between these meanings meaning continuum may be inconsistent. and the perceived meanings of who one is in a situation. What is crucial is that there is a match in the two sets of meanings. Respond- Discussion ents who received feedback from others that Studying the moral self is opportune given did not verify their moral identity standard the unregulated practices of stock brokers, were more likely to report guilt and shame than investment advisors, and mortgage lenders those whose identities were verified. whose behavior facilitated the recent eco- In identity theory, it does not matter whether nomic recession in the United States. The cost identity non-verification is due to individuals of their irresponsible practices has touched failing to live up to their identity standard or the lives of many innocent victims, as wit- exceeding their identity standard. In either nessed in the loss of individuals’ retirement case, individuals see they are not morally savings, homes, and jobs. The fact that a few acceptable given the standard they have set for greedy actors have the potential to damage themselves. Exceeding one’s moral standard the lives of many (as evidenced in the Bernie may generate the view that one is a moral Madoff case) brings issues of right and wrong, fraud, while failing to meet a moral standard good and bad, and just and unjust to public may result in feeling morally inadequate. Both awareness. To understand the illicit behavior situations will be associated with feelings of of some, we need to study the moral dimen- shame for not meeting one’s standard and guilt sion of the self and what makes some indi- for having enacted behavior inconsistent in viduals more dishonest than others within and meaning with the moral identity. across situations. Individuals’ identities are enacted within Identity theory provides an explanation for situations that carry their own shared meanings understanding this variability—variability that about how to behave and feel. Individuals was not discussed by Durkheim, Goffman, or interpret the cultural context in ways that make Collins. While psychologists address this indi- their identities and their surroundings mean- vidual variability, they lack an integrated the- ingful to them. We found that when individuals ory that identity theory provides, which brings defined situations as morally meaningful, both together moral cognition, moral behavior, cognitively and affectively, these definitions moral emotions, and situation meanings. were associated with acting morally; when Moral cognition is the control of perceptions respondents behaved immorally, they were in a situation to match internal moral identity more likely to feel guilt and shame. Actions meanings. Moral behavior is action in the ser- and feelings are thus embedded in particular vice of moral identity meanings. Moral emo- frames that provide additional meanings rele- tions are feelings following the lack of vant to one’s identity. Internal identity mean- verification of one’s moral identity. Situation ings and external cultural meanings coalesce to 136 American Sociological Review 77(1) influence moral action and feelings within and findings reveal that behaving immorally when across situations. a situation is defined as morally relevant is The fact that the results reveal that the associated with guilt and shame. However, identity process operates similarly across key these feelings may be tempered if important demographic variables such as race, age, and others (e.g., family and friends) define the income suggests that identity theory is a gen- situation in other, non-moral terms. High sta- eral theory of action. However, we did find a tus others may also be influential in how a significant difference between men and situation is defined, making moral meanings women regarding their experience of moral more or less relevant. The nature and degree emotions. Women were more likely than men of influence others have in framing situations to report guilt and shame when behaving as moral is an important avenue for future immorally. Research does reveal that women research. have a tendency to report guilt and shame Future work should also study other mean- more than men (Tangney and Dearing 2002a). ings of the moral person than justice and care. Women’s greater vulnerability to feelings of Morality may entail a focus on community, guilt and shame may be due to their tradi- loyalty, and respect for others. These mean- tional gender roles of caring for others or their ings underscore ideas of interdependence, tendency to be more interpersonally oriented cohesiveness, and attachment to groups. than men (Brody and Hall 2008). We also When individuals do not adhere to moral found that feeling rules of a situation were codes, identity non-verification may produce more likely to be associated with women’s feelings of guilt and shame in the violator, but reports of guilt and shame. Given women’s group members may also feel moral emotions tendency to be interpersonally oriented, of anger, contempt, or disgust (Haidt 2003). women may be more likely than men to take Because identities exist in mutually verifying into account a situation’s feeling rules so as contexts with each actor attempting to verify not to disrupt social relationships. Explana- the identity of the other (Burke and Stets tions for these sex differences are speculative; 2009), immoral behavior by one person will future research needs to investigate them likely disrupt other group members’ identity more closely. verification. In turn, this can lead to negative There are shortcomings to this research. Data feelings in other group members, such as are cross-sectional so we do not know how the anger or disgust. Empirical research is needed identity process unfolds in real time. For exam- regarding these interpersonal and emotional ple, according to identity theory, if people dynamics. repeatedly experience identity non-verification, Finally, more research is needed to iden- they may change their behavior (output) or their tify the source of moral identity meanings. perceptions of self-in-situation meanings (input) Exposure to particular social contexts and to achieve verification. However, while one or individuals may encourage a higher moral both of these are occurring, changes in one’s identity. For example, when parents are identity standard are also happening, although at involved in their children’s lives, their chil- a much slower rate (Burke and Stets 2009). dren are more likely to recognize moral val- Future research is needed to examine how indi- ues (Hardy, Padilla-Walker, and Carlo 2008). viduals respond to moral identity disruptions, Schools can also sensitize individuals to including how moral identity meanings may moral meanings by providing an atmosphere change over time. that fosters justice, virtue, and volunteering Another shortcoming is that we did not (Atkins, Hart, and Donnelly 2004). Religious examine how others in a situation may influ- traditions that promote reflection on moral ence how an interaction is framed. In the case issues and foster charitable work also help of morality, others may augment or reduce a individuals recognize moral meanings (Hart situation’s moral meanings. For example, our and Atkins 2004). Stets and Carter 137

In advancing the sociology of morality, we wallet, a book and so forth).” You “did have constructed an argument that is different not return the lost item” or “returned from the perspectives of Durkheim, Goffman, the lost item.” Collins, and Weber. Rather than seeing moral 8. “A cashier returns more money to you action and emotions as emerging from cul- than what is owed.” You “did not return the money” or “returned the money.” tural expectations in interaction or from individuals’ internal meanings about morality, both internal meanings and situation mean- Funding ings are the source of moral behavior and feelings. Individual variability in moral This research was supported by a National Science Foun- dation Grant (SES-0617022) to the first author. behavior and feelings results from the com- plex interplay between variation in internal moral identity meanings people seek to verify Acknowledgments and variation in definitions of moral mean- We would like to thank members of the Social Psychol- ings embedded in a situation. Thus, the con- ogy Seminar at the University of California, Riverside for trol system approach of identity theory in comments on an earlier draft. sociology becomes an important way to understand moral action and moral emotions. Notes 1. Other scholars claim that we need to focus more on naturalistic conceptions of morality that are rooted in Appendix everyday understandings and experiences (Walker and Pitts 1998; Walker et al. 1995). 2. Independent of others’ reactions, individuals may Scenarios evaluate whether they are behaving in a manner consistent with their identity meanings (Burke 2006). 1. “You have the opportunity to copy This is the self-appraisal process. Future research needs to examine this as an alternative source of another student’s answers during an identity discrepancy. exam.” You “did not copy the student’s 3. Behavior change is not the only way identity verifi- answers” or “copied the student’s cation may occur. Individuals might also engage in answers.” cognitive strategies to change meanings in a situa- 2. “You are drunk and need to drive tion. For example, individuals can use psychosocial home.” You “did not drive myself maneuvers to disengage from immoral behavior such home” or “drove myself home.” as reconstructing the conduct so that it is not viewed 3. “You have the opportunity to take an as immoral, minimizing one’s role in the behavior, item (for example, merchandise, minimizing consequences of the action, or blaming money etc.) that doesn’t belong to the victim (Bandura 1999). 4. One might expect shame to be about public trans- you.” You “did not take the item” or gressions and guilt to involve private transgressions, “took the item.” but research fails to support this distinction (Tangney 4. “You are asked to donate to a charity.” et al. 2007). Shame and guilt are equally public in You “did not provide a donation” or terms of others being present and aware of one’s you “provided a donation.” transgression. What is different is that following the 5. “You have the opportunity to allow transgression, shameful individuals focus on how another student to copy your answers others evaluate them; guilty people are concerned during an exam.” You “did not let the with how their behavior might have harmed another student copy your answers” or “let the (Tangney et al. 2007). In shame, individuals are self- student copy your answers.” focused; in guilt, they are other-focused. 5. In self-discrepancy theory, a discrepancy between 6. “A friend of yours is drunk and wants attributes a person actually possesses and the ideal to drive home.” You “did not let your state that a close other hopes the person attains influ- friend drive home” or “let your friend ences shame, while a discrepancy between a person’s drive home.” actual state (from her own viewpoint) and the state 7. “You find an item that does not belong the person feels she ought to attain influences guilt to you (for example, a cell phone, a (Higgins 1987). However, when this is tested, these 138 American Sociological Review 77(1)

specific self-discrepancies are not differentially and Identity, edited by D. K. Lapsley and D. Narvaez. related to shame and guilt (Tangney et al. 1998). Mahwah, NJ: Lawrence Erlbaum. 6. Future research should examine other moral emotions Bandura, Albert. 1999. “Moral Disengagement in the such as the other-condemning emotions of anger and Perpetration of Inhumanities.” Personality and Social disgust, the other-suffering emotion of compassion, Psychology Review 3:193–209. and positive emotions that emerge from good deeds Blasi, Augusto. 1980. “Bridging Moral Cognition and such as elevation and gratitude (Haidt 2003). Moral Action: A Critical Review of the Literature.” 7. These results are available upon request. Psychological Bulletin 88:1–45. 8. Squaring the value reflects the identity theory assump- Blasi, Augusto. 1984. “Moral Identity: Its Role in Moral tion that the emotional outcome to an identity Functioning.” Pp. 128–39 in Morality, Moral Behav- discrepancy is curvilinear rather than linear (Burke and ior, and Moral Development, edited by W. M. Kur- Harrod 2005). In a separate analysis, we examined tines and J. L. Gewirtz. New York: Wiley. whether there was a linear relationship between moral Blasi, Augusto. 1999. “Emotions and Moral Motivation.” identity discrepancy and guilt and shame. The findings Journal for the Theory of Social Behaviour 29:1–19. were not significant. Results are available upon request. Brody, Leslie R. and Judith A. Hall. 2008. “Gender and 9. We ran separate analyses of guilt and shame to inves- Emotion in Context.” Pp. 395–408 in Handbook of tigate whether the identity process operated Emotions, edited by M. Lewis, J. M. Haviland-Jones, differently for these two emotions. Across all equa- and L. F. Barrett. New York: The Guilford Press. tions, results were not statistically different. Results Burke, Peter J. 2006. “Perceptions of Leadership in are available upon request. Groups: An Empirical Test of Identity Control The- 10. In a separate analysis, we examined whether the rela- ory.” Pp. 267–91 in Purpose, Meaning, and Action: tionship between moral meanings that run from good Control Systems Theories in Sociology, edited by K. to bad (rather than relevant/present to not relevant/ A. McClelland and T. J. Fararo. New York: Palgrave absent) were associated with moral behavior or MacMillan. moral emotions. Findings were not significant and Burke, Peter J. and Alicia D. Cast. 1997. “Stability and are available upon request. Change in the Gender Identities of Newly Married 11. Effects of race, age, and income were nonsignificant Couples.” Social Psychology Quarterly 60:277–90. and are available upon request. Burke, Peter J. and Michael M. Harrod. 2005. “Too Much 12. We examined ethnic variations in reports of guilt and of a Good Thing?” Social Psychology Quarterly shame because there is evidence that Western cul- 68:359–74. tures are more guilt-based and Eastern cultures are Burke, Peter J. and Jan E. Stets. 2009. Identity Theory. more shame-based (Markus and Kitayama 1991). We New York: Oxford University Press. found no significant racial differences in reports of Collins, Randall. 2004. Interaction Ritual Chains. Prince guilt or shame. Results are available upon request. ton, NJ: Princeton University Press. Durkheim, Emile. 1961. Moral Education: A Study in the Theory and Application of the Sociology of Educa- References tion. New York: Free Press. Abend, Gabriel. 2010. “What’s New and What’s Old Durkheim, Emile. 1965. The Elementary Forms of the about the New Sociology of Morality.” Pp. 561–86 Religious Life. New York: Free Press. in Handbook of the Sociology of Morality, edited by Fourcade, Marion and Kieran Healy. 2007. “Moral Views S. Hitlin and S. Vaisey. New York: Springer. of Market Society.” Annual Review of Sociology Aquino, Karl and Dan Freeman. 2009. “Moral Identity in 33:285–311. Business Situations: A Social-Cognitive Framework Frimer, Jeremy A. and Lawrence J. Walker. 2009. “Rec- for Understanding Moral Functioning.” Pp. 375–95 onciling the Self and Morality: An Empirical Model in Personality, Identity, and Character: Explora- of Moral Centrality Development.” Developmental tions in Moral Psychology, edited by D. Narvaez and Psychology 45:1669–81. D. K. Lapsley. New York: Cambridge University Press. Gilligan, Carol. 1982. In a Different Voice: Psychological Aquino, Karl, Dan Freeman, Americus Reed II, Vivien Theory and Women’s Development. Cambridge, MA: K. G. Lim, and Will Felps. 2009. “Testing a Social- Harvard University Press. Cognitive Model of Moral Behavior: The Interactive Goffman, Erving. 1959. The Presentation of Self in Influence of Situations and Moral Identity Central- Everyday Life. Garden City, NY: Doubleday. ity.” Journal of Personality and Social Psychology Goffman, Erving. 1967. Interaction Ritual: Essays on 97:123–41. Face-to-Face Behavior. Garden City, NY: Anchor Aquino, Karl and Americus Reed II. 2002. “The Self- Books. Importance of Moral Identity.” Journal of Personality Goffman, Erving. 1974. Frame Analysis: An Essay on the and Social Psychology 83:1423–40. Organization of Experience. New York: Harper and Atkins, Robert, Daniel Hart, and Thomas M. Donnelly. Row. 2004. “Moral Identity Development and School Greene, Joshua D., R. Brian Sommerville, Leigh E. Nys- Attachment.” Pp. 65–82 in Moral Development, Self, trom, John M. Darley, and Jonathan D. Cohen. 2001. Stets and Carter 139

“An fMRI Investigation of Emotional Engagement in Narvaez, Darcia and Daniel K. Lapsley. 2009. Person- Moral Judgment.” Science 293:2105–2108. ality, Identity, and Character. New York: Cambridge Haidt, Jonathan. 2001. “The Emotional Dog and Its University Pres. Rational Tail: A Social Intuitionist Approach to Moral Osgood, Charles E., George J. Suci, and Percy H. Tan- Judgment.” Psychological Review 108:814–34. nenbaum. 1957. The Measurement of Meaning. Haidt, Jonathan. 2003. “The Moral Emotions.” Pp. 852–70 Urbana: University of Illinois Press. in Handbook of Affective Sciences, edited by R. J. Piaget, Jean. [1932] 1965. The Moral Judgment of the Davidson, K. R. Scherer, and H. H. Goldsmith. New Child. New York: Free Press. York: Oxford University Press. Sayer, Andrew. 2010. “Class and Morality.” Pp. 163–78 Haidt, Jonathan and Jesse Graham. 2009. “Planet of the in Handbook of the Sociology of Morality, edited by Durkheimians: Where Community, Authority, and S. Hitlin and S. Vaisey. New York: Springer. Sacredness Are Foundations of Morality.” Pp. 371–401 Serpe, Richard T. and Sheldon Stryker. 1987. “The in Social and Psychological Bases of Ideology and Construction of Self and Reconstruction of Social System Justification, edited by J. T. Jost, A. C. Kay, Relationships.” Pp. 41–66 in Advances in Group and H. Thorisdottir. New York: Oxford. Processes, edited by E. Lawler and B. Markovsky. Haidt, Jonathan and Selin Kesebir. 2010. “Morality.” Greenwich, CT: JAI Press. Pp. 797–832 in Handbook of Social Psychology, Shweder, Richard A., Nancy C. Much, Manamohan edited by S. T. Fiske, D. T. Gilbert, and G. Lindzey. Mahapatra, and Lawrence Park. 1997. “The Big New York: John Wiley and Sons. ‘Three’ of Morality (Autonomy, Community, and Hardy, Sam A. 2006. “Identity, Reasoning, and Emotion: Divinity), and the Big ‘Three’ Explanations of Suffer- An Empirical Comparison of Three Sources of Moral ing.” Pp. 119–69 in Morality and Health, edited by A. Motivation.” Motivation and Emotions 30:207–215. Brandt and P. Rozin. New York: Routledge. Hardy, Sam A., Laura M. Padilla-Walker, and Gustavo Stets, Jan E. and Michael J. Carter. 2006. “The Moral Carlo. 2008. “Parenting Dimensions and Adoles- Identity: A Principle Level Identity.” Pp. 293–316 cents’ Internalization of Moral Values.” Journal of in Purpose, Meaning, and Action: Control Systems Moral Education 37:205–223. Theories in Sociology, edited by K. McClelland and Hart, Daniel and Robert Atkins. 2004. “Religious Par- T. J. Fararo. New York: Palgrave MacMillan. ticipation and the Development of Moral Identity Stets, Jan E. and Michael J. Carter. 2011. “Understand- in Adolescence.” Pp. 157–72 in Nurturing Morality, ing the Moral Self: Applying Identity Theory.” Social edited by T. A. Thorkildsen and H. J. Walberg. New Psychology Quarterly 74:192–215. York: Kluwer Academic/Plenum. Stryker, Sheldon. [1980] 2002. Symbolic Interactionism: Hauser, Marc. 2006. Moral Minds: How Nature Designed A Social Structural Version. Caldwell, NJ: Blackburn Our Universal Sense of Right and Wrong. New York: Press. Harper Collins. Stryker, Sheldon and Peter J. Burke. 2000. “The Past, Hegtvedt, Karen A. and Heather L. Scheuerman. 2010. Present, and Future of an Identity Theory.” Social “The Justice/Morality Link.” Pp. 331–60 in Hand- Psychology Quarterly 63:284–97. book of the Sociology of Morality, edited by S. Hitlin Stryker, Sheldon and Richard T. Serpe. 1982. “Commit- and S. Vaisey. New York: Springer. ment, Identity Salience, and Role Behavior: A Theory Higgins, E. Tory. 1987. “Self-Discrepancy: A Theory and Research Example.” Pp. 199–218 in Personality, Relating Self and Affect.” Psychological Review Roles, and Social Behavior, edited by W. Ickes and E. 94:319–40. S. Knowles. New York: Springer-Verlag. Hochschild, Arlie R. 1979. “Emotion Work, Feeling Tangney, June P. and Rhonda L. Dearing. 2002a. “Gen- Rules, and Social Structure.” American Journal of der Differences in Morality.” Pp. 251–69 in The Psy- Sociology 85:551–75. chodynamics of Gender and Gender Role, edited by Hochschild, Arlie R. 1983. The Managed Heart: Com- R. F. Bornstein and J. M. Masling. Washington, DC: mercialization of Human Feeling. Berkeley: Univer- American Psychological Association. sity of California Press. Tangney, June Price and Rhonda L. Dearing. 2002b. Hoffman, Martin L. 2000. Empathy and Moral Develop- Shame and Guilt. New York: The Guilford Press. ment: Implications for Caring and Justice. New York: Tangney, June Price, Paula M. Niedenthal, Michelle Cambridge University Press. Vowell Covert, and Deborah Hill Barlow. 1998. Kohlberg, Lawrence. 1981. The Philosophy of Moral “Are Shame and Guilt Related to Distinct Self- Development. San Francisco: Harper and Row. Discrepancies? A Test of Higgins’s (1987) Hypoth- Lapsley, Daniel K. and Darcia Narvaez. 2004. Moral eses.” Journal of Personality and Social Psychology Development, Self, and Identity. Mahwah, NJ: Law- 75:256–68. rence Erlbaum. Tangney, June Price, Jeff Stuewig, and Debra J. Mashek. Markus, Hazel R. and Shinobu Kitayama. 1991. “Cul- 2007. “Moral Emotions and Moral Behavior.” Annual ture and the Self: Implications for Cognition, Review of Psychology 58:345–72. Emotion, and Motivation.” Psychological Review Turner, Jonathan H. 2010. “Natural Selection and the 98:234–53. Evolution of Morality in Human Societies.” Pp. 125–45 140 American Sociological Review 77(1)

in Handbook of the Sociology of Morality, edited by Wikstrom, Per-Olof H. 2010. “Explaining Crime as S. Hitlin and S. Vaisey. New York: Springer. Moral Actions.” Pp. 211–39 in Handbook of the Turner, Jonathan H. and Jan E. Stets. 2006. “Moral Emo- Sociology of Morality, edited by S. Hitlin and tions.” Pp. 544–66 in Handbook of the Sociology of S. Vaisey. New York: Springer. Emotions, edited by J. E. Stets and J. H. Turner. New York: Springer. Jan E. Stets is Professor of Sociology and Co-Director of Vaisey, Stephen. 2007. “Structure, Culture, and Com- the Social Psychology Laboratory at the University of munity: The Search for Belonging in 50 Urban Com- California, Riverside. During 2008 to 2010, she was munes.” American Sociological Review 72:851–73. Sociology Program Director at the National Science Walker, Lawrence J. and Karl H. Hennig. 2004. “Differ- Foundation. Recent publications include “The Social ing Conceptions of Moral Exemplarity: Just, Brave, Psychology of the Moral Identity” in Handbook of the and Caring.” Journal of Personality and Social Psy- Sociology of Morality 2010; Identity Theory (with Peter J. chology 86:629–47. Burke) 2009; and “Consistency and Enhancement Pro- Walker, Lawrence J. and Russell C. Pitts. 1998. “Natural- cesses in Understanding Emotions” (with Emily K. istic Conceptions of Moral Maturity.” Developmental Asencio) Social Forces 2008. Her work continues to Psychology 34:403–419. develop identity theory, the sociology of emotions, and Walker, Lawrence J., Russell C. Pitts, Karl H. Hen- the sociology of morality. nig, and M. Kyle Matsuba. 1995. “Reasoning about Morality and Real-Life Moral Problems.” Pp. 371–407 in Morality in Everyday Life: Developmental Per- Michael J. Carter is Assistant Professor of Sociology at spectives, edited by M. Killen and D. Hart. New York: California State University, Northridge. His work exam- Cambridge University Press. ines how the moral identity influences behavior and Weber, Max. [1922] 1968. Economy and Society. Trans- emotions in small groups and across social contexts. lated by G. Roth and C. Wittich. Los Angeles: Uni- Recent publications include “The Moral Self: Applying versity of California Press. Identity Theory” (with Jan E. Stets) in Social Psychology Weber, Max. 1978. “Value-Judgments in Social Science.” Quarterly 2011, and “Moral Identity, Status, Moral Emo- Pp. 69–98 in Weber: Selections in Translation, edited tions, and the Normative Order” (with Jan E. Stets, by W. G. Runciman. New York: Cambridge University Michael M. Harrod, Christine Cerven, and Seth Abrutyn) Press. in Social Structure and Emotion 2008.