and 30 months of age (Lewis, 1992) so there is both a theoretical and an ... mick, Bursik, & Arneklev, 1993; Fischer & Tangney, 1995; Hoglund & Nich-.
PsychologicalReports, 2003,93, 101-126. O Psychological Reports 2003
SELF-CONSCIOUS EMOTIONS AND CRIMINAL OFFENDING ' STEPHEN G. TIBBETTS California Stare University, Son Bernardino Summary.-This study examined the relation of personahty traits-sharne-proneness, guilt-proneness, and pride-on offending behavior. Using survey data from a sample of 224 college students, the consrruct and criterion-related validity of scales of the Shame Proneness Scale, the Test of Self-conscious Affect, and the Personality Feelings Questionnaire-2 were assessed. Regression analyses showed that self-conscious emouons are important in the etiology of criminal offending. Specifically, rated pride was posiuvely correlated with self-reported criminal activity, whereas ratings of guilt were negatively associated with offending. The relation of shame with criminality varied depending on the type of measure used to indicate proneness to shame.
In the past 30 years, there has been a revolution in psychology regarding the study of self-conscious emotions and their influence on human behavior (for an extensive review of pre-1995 studes, see Tangney & Fischer, 1995; for more recent studies, see Harder, 1995; Hoglund & Nicholas, 1995; Tangney, 1996; Tangney, Miller, Flicker, & Barlow, 1996; Tangney, Wagner, Barlow, Marschall, & Grarnzow, 1996; Ferguson & Crowley, 1997a, 1997b; Quiles & Bybee, 1997; Einstein & Lanning, 1998; Leith & Baumeister, 1998; Eisenberg, 2000). A similar, more recent but less recognized trend has occurred in the criminological literature (e.g., Braithwaite, 1989; Grasmick & Bursk, 1990; Grasmick, Burs&, & Cochran, 1991; Grasmick, Bursik, & Kinsey, 1991; Grasmick, Blackwell, & Bursik, 1993; Grasmick, Burslk, & Arneklev, 1993; Nagin & Paternoster, 1993; Paternoster & Simpson, 1993, 1996; Clarke, 1995; Elis & Simpson, 1995; Wortley, 1996; Tibbetts, 1997; Hay, 2001). Self-conscious emotions are dstinguished from other emotions in that they are, by definition, social in the sense of being emotional experiences in social relationships and interactions, in which people evaluate themselves and each other. Self-conscious emotions "are built on reciprocal evaluation and judgement" (Fischer & Tangney, 1995, pp. 3-41. Self-conscious emotions have been dstinguished from other emotions by their requirement of the awareness of self-consc~ousness,which many emotions do not require, such as the same "prmary" emorlons present at birth-pain, joy, or d~sgust(for a review, see Lewis, 1992). Ln contrast, studies indicate that self-awareness is developed in healthy children between 18 and 30 months of age (Lewis, 1992) so there is both a theoretical and an 'The author thanks Melissa Combs for hel in collecting and analyzing the data used herein. Address correspondence to Stephen G. ~ i g b e t t s ,De artment of Criminal Jusuce, California , 92407. Stare University, 5500 University Parkway, San ~ e r n a r j m o CA
S. G. TIBBETTS
empirically observed difference between self-conscious emotions and the primary emotions. Notable self-conscious emotions include guilt, pride, shame, and embarrassment.' Each of these emotions have been found by both chical and empirical research by both criminologists and psychologists to have profound effects on human development, perceptions, and behavior (see Lewis, 1971; Tangney, 1990, 1995; Grasrnick & Bursk, 1990; Lewis, 1992; Grasmick, Bursik, & Arneklev, 1993; Fischer & Tangney, 1995; Hoglund & Nicholas, 1995; Tangney, Wagner, Barlow, Marschall, & Gramzow, 1996). Thus, the intensities and associations of self-conscious emotions are likely to be key in the inclination or inhibition of committing criminal behavior. Although psychologists have developed conceptual frameworks and assessed the psychometrics of many tests used to measure self-conscious emotions, most criminological research has not acknowledged, let alone utilized, this extensive theoretical and empirical literature. On the other hand, psychologists have not recognized the extensive research on shame and embarrassment by workers in criminology (e.g., Braithwaite, 1989; Elis & Sirnpson, 1995; Wortley, 1996; Hay, 2001) as we1 as other studies (Grasmick & Burs&, 1990; Grasmick, Bursik, & Kinsey, 1991; Grasmick, Bursik, & Arneklev, 1993; Clarke, 1995). This lack of consistency across disciphes has resulted in varied findings and conclusions regarding the relationships of selected self-conscious emotions in the etiology of criminal offending, discussed below. Another result is that virtually all recent criminological studies of self-conscious emotions have included measures in which the relationships of shame, guilt, and embarrassment are confounded (Grasmick & Bursik, i990; Grasmick, Bursk, & Cochran, 1991; Grasmick, Burslk, & Kinsey, 1991; Grasmick, Bursik, & Arneklev, 1993; Nagin & Paternoster, 1993; Elis & Simpson, 1995; Wortley, 1996). Perhaps this is why only one clearly conceptualized trait-low self-control-has received adequate attention in the criminological literature (Gottfredson & Krschi, 1990; Gibbs & Giever, 1995; Piquero & Tibbetts, 1996; Tibbetts & Myers, 1999).
'Embarrassment, while generally considered an important self-conscious emotion, was not examined in the present study primarily because embarrassment is most likely ex erienced after a situation that is not intentional (Edelmann, 1987; Lewis, 19921, which woulcf)not generally apply to criminal offending. T o clardy, feelings of embarrassment rypically are experienced due to situational breaches of social norms regarding conduct chat is not immoral nor intentional, yet may diminish a person's persona, i.e., presented self. However, such violations of conduct, e.g., bur ing during a social dinner, do not tend to influence one's image of self and, thus, are not likd to influence, e.g., deter, decisions to commit immoral acrs or criminal offending. Additionally, studies (Edelmann, 1987) have shown that experiences of embarrassment do not result in ermanently ne ative feelings of self because such feelings are simply based o n situational vioyations of socia? folkways, and not a global feelmg of self. Nevertheless, we suggest that research examining the association benveen self-conscious emocions and criminality include measures of embarrassment to see if such episodes of embarrassment lead to criminal acrs.
SELF-CONSCIOUSEMOTIONS AND OFFENDING
Given this lack of coherence regardmg the self-conscious emotions, the purpose of the present study was (1) to examine the psychological and criminological literature for theoretical evidence regarding the conceptuaLzations and associations of self-conscious emotions with criminal offending, (2) to evaluate the construct and criterion-related validity of several established measures of three types of self-conscious emotions (shame, guilt, and pride), and (3) to test the relationships of these emotional constructs to general criminal offending behavior to gain a better understanding of how each of these emotions may be related to reported criminal activity.
Shame Shame is a self-conscious emotion involving intense feelings of worthlessness or weakness that results from global evaluations of self-concept, regardless of the source (internal or external), regarding discrepancies between one's perceptions of self and ideal images of self (Lewis, 1971; for reviews, see Tangney, 1996; Ferguson & Crowley, 1997a). To dustrate, a person feeling shame would Uely express feeltng(s) verbally as, ' ' I am a bad person who d ~ dthat terrible thing(s)," with emphasis on the "I am a bad person" (Tangney, Wagner, Gavlas, & Gramzow, 1991; Lewis, 1992; Tangney, 1996). Also, researchers have suggested that a feeltng of shame, given its global nature, is generally more painful than those of other emotions, e.g., guilt, embarrassment [see below]. Further, research has shown that the phenomenological experience of shame is different than that of other self-conscious emotions, in that it involves a f e e h g of powerlessness or worthlessness (for a review, see Fischer & Tangney, 1995; also see Lewis, 1971; Lewis, 1992; Tangney, 1996; Ferguson & Crowley, 1997b). This type of phenomenological experience is likely to lead to a different reaction, e.g., anger, helplessness, by people than the reactions to other, less self-conscious emotions, e.g., guilt, embarrassment. Psychologists have noted that feelings of shame tend to lead to a concern with others' evaluation of an impaired sense of self, whereas other types of self-conscious emotions, e.g., guilt, see below, lead to feeltngs of concern regarding the influence on others of one's actions (or nonactions). Given the impaired sense of self that is characteristic of shame, psychologists (Lewis, 1971; Harder, Cutler, & Rockart, 1992; Harder, 1995; Hoglund & Nicholas, 1995; Tangney, 1996; Tangney, Wagner, Barlow, Marschall, & Gramzow, 1996; Leith & Baumeister, 1998) have concluded that being prone to shame IS more likely to result in negative or maladaptive attitudes and behaviors than are actual feelings of shame, or other selE-conscious emotions, e.g., guilt, embarrassment. Clinical and empirical research has shown this concept of shame and its hstinction from other self-conscious emotions (for a review, see Tangney &
S. G. TIBBETTS
Fischer, 1995; see also, Harder, et a/., 1992; Tangney, 1996; Tangney, Mdler, Flicker, & Barlow, 1996; Ferguson & Crowley, 1997a). Further, this definition of shame is consistent with recent theoretical developments by psychologists regarding an established conceptual framework of self-conscious emotions (see studies cited above). To clardy, this conceptualization of shame emphasizes the nature of the global evaluation people make toward themselves, regardless of the source, i.e., external or internal, of the evaluation, after accionls) (or nonaction) that they know goes against what is beneficial. It is important to note that this global, self-conscious emphasis of shame is likely to have a relation to decisions to commit criminal offenses, especially in relation to the less self-oriented effects of nonglobal self-conscious emotions, such as guilt or embarrassment. While several psychological models, now considered outdated, have noted that shame was distinguished by "public exposure" (Ausubel, 1955; Piers & Singer, 19711, it is now generally acknowledged by psychological theory and empirical research that shame can occur both in the presence of others and alone (Tangney & Fischer, 1995; Tangney, 1996; Tangney, Mdler, Flicker, & Barlow, 1996; Ferguson & Crowley, 1997b). However, this modern psychological definition of shame is not consistent with the definitions of shame used in most of criminological studies that have emphasized the internal source of shame. Specifically, Grasmick and his colleagues (Grasmick & Bursik, 1990; Grasmick, Bursik, & Cochran, 1991; Grasmick, Blackwell, & Bursik, 1993) and other criminologists (Nagin & Paternoster, 1993; Paternoster & Simpson, 1993, 1996; Elis & Simpson, 1995) defined and operationalized shame as "the guilt you would feel if you committed an offense even if no one found out." Further, criminologists' conceptualizations of shame tend to disregard the emphasis of shame on the self, as opposed to one's action (or nonaction). The one- or two-item measures of shame that are dominant in most of the criminological studies of shame only inquire as to the respondent's feeling of "guilt" if they were to commit a given offense, which does not deheate between the negative evaluation of one's self (shame) and the negative evaluation of one's actions (guilt) or public mishaps (embarrassment). Thus, the associations of shame with offending in the criminological literature is quite M e r e n t from that found in psychological studies. Specifically, virtually all criminological studies of shame report moderate to strong inhibitory relationships with criminality (Grasmick & Bursik, 1990; Grasmick, Bursik, & Cochran, 1991; Grasrnick, Burslk, & Kinsey, 1991; Grasmick, Blackwell, & Burs~k,1993; Grasrnick, Bursdc, & Arneklev, 1993; Nagin & Paternoster, 1993; Paternoster & Simpson, 1993, 1996; Tibbetts, 1997), which is generally not acknowledged in the psychological literature.
SELF-CONSCIOUSEMOTIONS AND OFFENDING
While a few criminological studies have differentiated the various selfconscious emotions, none of these studies incorporate the psychological distinctions between shame, guilt, and embarrassment. For example, an exploratory study (Tibbetts, 1997) deheated the relationships of three types of shame: that due to exposure or without exposure and shame-proneness. Furthermore, Tibbetts attempted to delineate the relationships of shame from other self-conscious emotions, e.g., guilt. The study by Tibbetts found that states of anticipated shame inhibited deviant intentions, but that shameproneness had a positive relationship with attitudes toward committing criminal offendmg. But like most criminological stud~es,chat study did not estimate the correlates of shame while c o n t r o h g for the other self-conscious emotions. Thus, although Tibbetts (1997) reported more specific conclusions regardmg the associations of state versus trait measures of shame with criminality, the estimated correlations are k e l y to be biased because other selfconscious emotions, e.g., guilt, pride, were not taken into account. Such criminological definitions of self-conscious emotions are not consistent with the established psycholog~calconceptualizations of these emotions that have been supported by clkcal and empirical research (Harder, et al., 1992; Harder, 1995; Tangney & Fischer, 1995; Tangney, 1996; Tangney, Mdler, Flicker, & Barlow, 1996). Thus, the relationships of shame, as well as other emotions, with offending found in the criminological literature are based on more traditional definitions (e.g., Piers & Singer, 1971) and the measures based on such defmitions. The lack of dflerentiation between shame and other self-conscious emotions confounds the observed correlates of these emotions, and the conclusions based on such findings are likely to be confusing when compared to findings reported in the psychological literature. One important perspective offered in the psychoIogical literature for distinguishing shame from other self-conscious emotions is attribution theory (see Graham & Weiner, 1986; Weiner, 1986). Although this perspective is largely based on research with children, it has implications for this study because it concerns the process in which emotional reactions are triggered. From the attribution perspective, the primary appraisal involves analyzing the nature of the outcome-success or failure-which elicits a n outcome-dependent emotion, e.g., shame or pride. The secondary appraisal then takes place and involves an evaluation of whether the farlure or success was due to external or internal factors, which may also be seen as uncontrollable or controllable factors (Ferguson & Stegge, 1995). This usually results in assessing which attribution-dependent emotions, such as shame or guilt, are likely to be elicited based on this evaluation. Although attribution theory is not examined in the present study, such a theory and empirical findings provide
S. G. TIBBETTS
an example of how self-conscious emotions have been differentiated and conceptuahzed in the psychological literature. In sum, the conceptualizations of shame utilized by criminologists are inconsistent or even contradictory to those used in psychology. It is apparent that criminologists have not incorporated psychological research that has shown that shame is more than an internally induced emotion and often has an external source. Furthermore, other aspects, e.g., intensity of distress, phenomenological experience, etc., psychologists use to distinguish shame from other self-conscious emotions have not been considered in recent criminological research. At the same time, psychologists have not incorporated findings reported in the criminological literature regarding th_e inhibitory relationships between anticipated shame and criminal offending. These omissions are in hrect opposition to the desired interdisciplinary nature of theory and empirical research and result in biased estimates of the relationship of shame and other self-conscious emotion with criminal offending.
Guilt Guilt is considered by most experts in the study of self-conscious emotions as a feeling in which the self is not the central object of negative evaluation, but rather the thing done or not done is the focus (Lewis, 1992; Tangney, 1996; Tangney, Wagner, Flicker, & Barlow, 1996; Ferguson & Crowley, 1997b) regardless of the source (internal or external). To ~Uustrate, a person feeling guilt would Uely express thei.r feeling(s) verbally as, "I am not a bad person, but I did that terrible thing," with emphasis on "that terrible thing." In contrast to shame, the subjective evaluation by the individual focuses on the thing that was done (or not done) rather than on the self. Psychologists have noted this distinction as the primary ddference between shame and guilt, and empirical studies have supported this proposition (for reviews, see Harder, et al., 1992; Harder, 1995; Tangney & Fischer, 1995; Tangney, 1996). Also, researchers have noted that feeling guilt, given its less global nature, is generally less painful than feehgs of shame. Furthermore, psychological research has shown that the predispositional experience of guilt is different from that of shame in that it involves a feeling of regret and remorse, but not an overall f e e h g of low self-worth (Tangney, 1996; Tangney, Mder, Flicker, & Barlow, 1996; Quiles & Bybee, 1997). This type of phenomenological experience is Uely to lead to a different reaction by the individual than shame, since feelings of guilt are more apt to result in a desire to repair damage that is done or to apologize. Psychologists note that a feeling- of guilt leads to a concern about one's actions on others, that is likely to lead to attempts to repair the damage done, as opposed to feelings of shame that prehspose one to hide, escape, or even strke back (Tangney & Fischer, 1995; Leith & Baumeister, 1998).
SELF-CONSCIOUSEMOTIONS AND OFFENDING
Given the act-oriented and reparative nature of guilt, it is likely that people who are more prone to feel guilty would be less likely to commit criminal offenses. This behef is due to the assumption that individuals who are predisposed to experience guilt (or have experienced guilt) wdl be more likely to avoid circumstances that cause guilt or that such actions (or nonaction) d induce feelings that compel makmg amends to resolve the action (or nonaction). Despite the explicit differences between shame and guilt, most criminological studies (Grasmick & Burslk, 1990; Grasmick, Bursik, & Cochran, 1991; Grasmick, Bursik, & Kinsey, 1991; Hudley, 1991; Grasmick, Bursik, & Arneklev, 1993; Nagin & Paternoster, 1993; Paternoster & Simpson, 1993, 1996; Elis & Simpson, 1995; Wortley, 1996; Tibbetts, 1997) that have tested the relationships of self-conscious emotions have not acknowledged these established psychological distinctions. Therefore, these criminological researchers have used conceptual definitions of emotions that are inconsistent with accepted psychological conceptuahzations of shame and guilt. Specifically, most criminological researchers have only measured, albeit inaccurately, shame or embarrassment (Grasrnick & Burslk, 1990; Grasmick, Burslk, & Cochran, 1991; Grasmick, Bursik, & Kinsey, 1991; Hudley, 1992; Grasmick, Burslk, & Arneklev, 1993; Nagin & Paternoster, 1993; Paternoster & Simpson, 1993, 1996; Elis & Simpson, 1995) and have omitted guilt. One form of misconceptu~zingguilt is seen in the methodological strategy taken by some researchers (e.g., Hudley, 1992) that has allowed participants to define their self-reported feehgs of guilt, without distinguishing feelings of shame or embarrassment in prototypical situations and/or certain attributes of the event. (For a discussion on the disadvantages of this form of operationalization of measures, see Tangney, 1996.) However, most criminologists (Grasmick & Bursik, 1990; Grasmick, Bursik, & Cochran, 1991; Grasmick, Bursik, & Kinsey, 1991; Grasrnick, Burs~k,& Arneklev, 1993; Nagin & Paternoster, 1993; Elis & Simpson, 1995) have tended to operationalize shame as the amount of "guilt" one feels (or would feel) if they were to commit a given act. Thus, not only is guilt neglected as a construct in criminological studies of offending, but it is hkely that the few studies in which relationships of guilt have been estimated are invalid due to confounding guilt with self-reported feehgs of shame, which have been shown by extensive psychological research to be distinguishable.
Embarrassment While embarrassment was not measured in the present study, it is important to define embarrassment and to elaborate on the characteristics which dstinguish it from other self-conscious emotions. The predicted relations of embarrassment in recent criminological studies (see previously cited
S. G . TIBBETTS
studies by Grasmick & Burslk, 1990; Grasmick, Bursik, & Cochran, 1991; Grasrnick, Bursik, & Kinsey, 1991; Grasmick, Burslk, & Arneklev, 1993; Nagin & Paternoster, 1993) are of doubtful accuracy because criminological definitions differ from psychological conceptualizations of embarrassment (Edelrnann, 1987; Lewis, 1992, 1995; Miller, 1995). Psychologists (Miller, 1995, p. 323) have defined embarrassment as feelings that occur "when unwanted public events create acute concern for how one is being evaluated by others . . . that results when one's desired public identity is endangered." Therefore, the subjective experience of embarrassment tends to focus on negative evaluations of one's persona (presented self), as opposed to one's self (shame) or one's actions/nonactions (guilt), which is a primary &stinguishing characteristic of embarrassment from shame and guilt (Edelrnann, 1987). In contrast, criminologists (e.g., Grasmick & Bursik, 1990; Grasmick, Bursik, & Kinsey, 1991; Grasmick, Bursik, & Arneklev, 1993) have defined embarrassment as the Uehhood that "most of the people whose opinions you value lose respect for you" if one was to commit a given act, which is not consistent with psychological conceptuahzations of embarrassment and, further, does not distinguish it from other self-conscious emotions because both shame and guilt can also result after something is detected by others (Fischer & Tangney, 1995; Tangney, 1996; Tangney, Miller, Flicker, & Barlow, 1996; Tangney, Wagner, Barlow, Marschall, & Gramzow, 1996; Quiles & Bybee, 1997; Leith & Baumeister, 1998). Criminological research has reported inconsistent results for the correlations of embarrassment with criminal attitudes and behavior, which is not surprising given the apparent confounded nature of the conceptualization of this construct in this literature.
Pride Psychologists (Mascolo & Fischer, 1995, p. 65) have defined pride as feehgs in which "a person appraises the self as having responsibhty for accomplishing a socially valued outcome or being a socially valued person." Furthermore, psychological researchers (Tangney, 1989; Tangney, Wagner, & Gramzow, 1989; Ferguson & Crowley, 1997a) have identified two types of pride, alpha and beta. Alpha pride is a more self-oriented measure of pride that reflects general pride in self, e.g., "I am a trustworthy person" (Ferguson & Crowley, 1997a). Beta pride, on the other hand, has a more situation-specific focus that reflects pride in one's behavior, e.g., "You would think: 'I feel that my hard work paid off' " (Ferguson & Crowley, 1997a). Pride may be the most neglected self-conscious emotion in crirninological models of offending (for exceptions, see Shields & Whitehall, 1991; Simourd, 1997). An extensive survey of the criminological literature identified only a few empirical studies of the relationships of ride with offending.
SELF-CONSCIOUS EMOTIONS AND OFFENDING
Specifically, Hudley (1992) found that incarcerated youths reported more attributions of pride for individual accomplishments than did secondary school youths. However, the methodological strategy for measuring, thereby defining, pride was based on participants providing prototypical situations that caused them to feel proud (Hudley, 1992). Such a measure of pride, or other emotions, is questionable (see Tangney, 1996). In another approach, Shields and Whitehall (1991) developed the Pride in Dehquency scale, which is a 10-item self-report test that taps an individual's comfort (pride vs shame) about engaging in specified criminal behaviors, e.g., beating up a child molester. However, this instrument is vulnerable to criticisms regarding the tautological nature of the items that make up the scale. There has been only one published study with this test (Simourd, 1997) in assessing its accuracy in predicting offending and recidivism. Simourd (1997) reported that the Pride in Delinquency scale was useful for assessment and that scores were positively associated with several measures of offending. However, this measure of pride depends on respondents' attitudes toward specific criminal behaviors, as opposed to more established psychological measures of pride, e.g., the Test of Self-conscious Affect (Tangney. et a[.. 1989) that tap feehgs regarding more general behaviors. In sum, the Test of Self-conscious Affect has previously published psychometr~ccharacteristics and appears to be a more vahd measure of pride than the Pride in Delinquency scale (Shields & Whitehall, 1991) as well as measures of pride on which participants define their feehgs of pride in nonoffending behaviors (cf. Hudley, 1992). It is worthwhile to note that limited research using the Pride in Dehquency scale (Shields & Whitehall, 1991; Simourd, 1997) suggests pride is positively related to attitudes and behaviors favorable to c r h ~ n a offending. l Applying Established Measures of Self-conscious Emotions to Criminal Offending The above review indicates that criminologists have used different conceptual models and measurements of self-conscious emotions in their correlates with criminal offending than psychologists have. This may be the reason why inconsistent conclusions have been made by psychologists and crirninologists regarding the relationships of self-conscious emotions with offending behavior. For example, criminological researchers (Grasmick & Bursrk, 1990; Grasrnick, Bursik, & Cochran, 1991; Grasmick, Bursrk, & h s e y , 1991; Grasmick, Burslk, & Arneklev, 1993; Nagin & Paternoster, 1993; Elis & Simpson, 1995; Paternoster & Simpson, 1996; Piquero & Tibbetts, 1996) have consistently found that anticipated states of shame are associated with fewer decisions to commit criminal offenses, while psychological research has generally shown that shame-proneness is positively correlated with maladap-
S. G. TIBBETTS
tive behaviors (for a review, see Tangney, Burggraf, & Wagner, 1995). On the other hand, most criminological research does not include guilt as a separate construct (rather it is used to measure shame). Most psychological research has supported guilt as an important correlate to inhibit offending behavior (see Hoglund & Nicholas, 1996; Tangney, 1996; Tangney, Wagner, Barlow, Marschall, & Gramzow, 1996; Quiles & Bybee, 1997; Leith & Baumeister, 1998). The only assumed characteristic of self-conscious emotions that criminologists and psychologists agree on is that pride predisposes individuals to criminal activity. Most of the ddferences between the conclusions of criminologists and psychologists are Wcely due to how these constructs are measured. In the criminological research, virtually all measures of self-conscious emotions consist of one- or two-item measures (all studies by Grasmick & Bursik, 1990; Grasmick, Bursik, & Cochran, 1991; Grasmick, Bursik, & Kmsey, 1991; Grasmick, Bursik, & Arneklev, 1993; Nagin & Paternoster, 1993; Elis & Simpson, 1995; Piquero & Tibbetts, 1996) that have not been vahdated, whereas virtually all of the psychological measures of shame, guilt, and pride are well-vahdated, multiple-item, psychometric vddated scales (Harder & Zalma, 1990; Harder, et al., 1992; Harder, 1995; for a detaded discussion, see Tangney, 1996). The criminological measures tend to emphasize anticipated feehgs of shame (or embarrassment) for behaviors that they are likely to feel for doing criminal acts. However, most psychological measures of shame and guilt tend to emphasize feelings that respondents have felt (or would feel) were they to find themselves in certain situations (Tangney, 1996) that usually do not involve criminal behavior. Further, criminological measures of self-conscious emotions do not distinguish self-conscious emotions, e.g., shame from guilt, whereas psychological measures appear to have made significant advances in distinguishing one self-conscious emotion from the others. Finally, criminologists have not acknowledged the differential correlates of emotional states from emotional traits (Tibbetts, 1997), which are &ely to alter the relationships of emotions with offending behavior. This last point requires further discussion. Criminological measures of self-conscious emotions have generally only examined one form of emotion, anticipated states of feelings. However, psychological theory and empirical research has established at least two forms of emotional factors, namely, states and traits (see M a n , Gilbert, & Goss, 1994; Goss, Gilbert, & M a n , 1994; Fischer & Tangney, 1995, Tangney, 1996; Dorahy & Schumaker, 1997). Emotional states are the feeLngs of a given emotion an individual experiences before or after a particular s~tuation, whereas an emotional trait is the intensity of a given emotion an individual experiences on a long-term basis. For example, an individual may have rela-
tively low guilt-proneness, i.e., guilt trait, but may feel high guilt for doing, or not doing, a particular act, i.e., guilt state. On the other hand, another individual may have relatively high shame-proneness, i.e., shame trait, but not experience high shame for committing (or not committing) a particular act, i.e., shame state. Although criminological research has not incorporated this distinction in their measures, psychological research has shown this distinction between state and trait is vital in estimating the relations of self-conscious emotions and often leads to differential associations with offending behavior. As reported by Tangney and Fischer (1995), many clinical studies regarding the relation of self-conscious emouons with particular forms of chronic deviance, e g., domestic violence, alcoholism, etc., have shown that shame-proneness or gudr-proneness, as traits, are k e l y to predispose an inhvidual to offendmg Such studies are based on emotional traits, i.e., longterm dispositions, which psychological studies indicate are maladaptive in individuals' attempts to adapt to their environment. On the other hand, numerous studies reviewed by Tangney and Fischer (also see M a n , et al., 1994; Tangney, 1996; Dorahy & Schumaker, 1997) have shown chat situational experiences (states) of self-conscious emotions are not necessarily associated with chronic, maladaptive behaviors. Instead, situational or statebased experiences of shame or guilt are assoc~atedwith beneficial actions, i.e., reparative behavior, that inhibit an indiv~dualfrom offending. Thus, the distinction between state- and trait-based measures of selfconscious emotions is key to understanding the relationships of shame, guilt, and pride with criminal offending (Allan, et al., 1994; Goss, et al., 1994; Dorahy & Schumaker, 1997). Psychological research has generally found that measures which rely on adjective lists of prevalent attitudes and behaviors by the respondent (e.g., Personality Feehgs Questionnaire-2; Harder & ZaLna, 1990; Harder, 1995) measure traits of emotions, whereas measures that elicit participants' reactions to hypothetical situations, e.g., Test of Selfconscious M e c t (Tangney, 1989; Tangney, et al., 1989) Shame Proneness Scale (Tibbetts, 1997), tend to tap more experiential (or state) aspects of an emotion. However, the criminological research has not incorporated this diEferentiation between emotional states and traits. The present study distinguished these types of emotional experiences for shame and guilt. Further, the present study examined the correlates of self-conscious emotions with a general index of criminality among a sample of college students who have not been pursued in studies of deviant behavior. Prior conclusions drawn from criminological studies that have estimated the relations of pride, shame, and guilt are in question because they differ from theoretical models and measures regularly uthzed in psychological research.
S. G. TIBBETTS
In light of the evidence from the studies reviewed above, six primary hypotheses were formulated for this study. The first hypothesis is that scores on the various measures of shame will be relatively consistent and strongly correlated, i.e., higher than .50, particularly the state-based measures of shame, e.g., TOSCA-shame and Shame Proneness Scale. The second hypothesis is that scores of guilt wdl be consistent, but moderate, i.e., between .3O and .50, given the nature of the measures: TOSCA-guilt is state-based; Personality Feelings Questionnaire-2 is trait-based. The third hypothesis is that participants who score higher on measures of trait-based measures of shame-proneness, i.e., PFQ-2-shame, will be more hkely to report engaging in criminal behavior after c o n t r o h g for other variables. The fourth hypothesis is that individuals who score higher on measures of state-based measures of shame-proneness, i.e., Shame Proneness Scale and TOSCA-shame, wdl be less likely to report engaging in criminal behavior even after taking into account other variables. The fifth hypothesis is that individuals who score higher on measures of guilt, particularly state-based measures, i.e., TOSCAguilt, will be less hkely to engage in criminal behavior after accounting for other variables. The sixth hypothesis is that individuals who score high on the pride measure d be more likely to engage in criminal behavior after c o n t r o b g for other variables. Finally, the seventh hypothesis is that the variance explained by the self-conscious emotions will explain most of the demographic variables' association (sex, GPA, and membership in a fraternity or sorority) with criminal offending.
participants We surveyed student volunteers enrolled in six undergraduate introductory behavioral science courses at a large southeastern state university. An anonymous questionnaire was distributed to 232 students who were in attendance on the day of the survey. A completed questionnaire was returned by 224 students (108 women and 116 men, M age=21.4 yr.); the remaining eight cases were not included in the analysis as information was missing. All six courses fulfilled general university requirements, and the respondents reported a range - of academic majors. However, first-year students (38% of the sample) were over-represented as would be expected in a s a m p h g from general introductory courses. Comparisons of sample characteristics with those for the entire undergraduate population on the campus were carefully examined. For instance, 22% of the participants reported they had declared a major in the area of social sciences, whereas approximately 19% of the undergraduate student bodv had declared maiors in this area. Other reported academic majors among the participants showed similar comparisons with estimates for aU
SELF-CONSCIOUSEMOTIONS AND OFFENDING
undergraduates at the institution. Other sample characteristics also tended to reflect similar representations found in the general student body. For example, the sample included 48% women and 52% men, which was similar to representation across campus (51% women and 49% women). Also, 15% of the participants reported fraternity or sorority affiliation; this mirrored the university's estimate of 15% fraternity/sorority affihation among the undergraduate student body. Notably, the sample contains few nonwhite minorities (5%), which was similar to the student body (3%). The use of student samples in criminological research has been criticized (Williams & Hawkins, 19861, the major objection being that college populations do not contain a high proportion of serious offenders. However, recent studies (e.g., Nagin & Paternoster, 1993; Tibbetts & Herz, 1996) have shown that many (32-48%) college students commit the offenses examined in the present study, e.g., driving while intoxicated, larceny, marijuana use, etc. Further, the use of introductory courses that fulfill university requirements is the best way to sample a cross-section of college students. Fin d y , it is important to note that many of the measures of the key variables in this study-the Shame Proneness Scale and all of the Test of Self-conscious Affect scales-were specifically designed for college-age respondents. Procedure College students were asked to participate in a project examining their offendmg behavior. After obtaining voluntary consent, the questionnaires were distributed and completed durmg regular class periods i n six different lecture courses. Students were told to respond to the questionnaires according to the instructions provided. A staff member reviewed [he instructions and answered questions about the research. The questionnaires asked respondents to complete several scales, including the Test of Self-conscious Affect (Tangney, et af., 1989), the Shame Proneness Scale (Tibbetts, 1997), and the Personahty Feelmgs Questionnaire-2 (Harder & Zalma, 1990) described below, as well as to report prior offending behavior and demographic information. Measures Dependent variable.-The dependent variable was Criminal Offending as measured by a scale created by summing participants' responses (coded O=no, 1 or more=number of times) to questions regarding the frequency with which they had committed certain behaviors over the past 6 months. These acts included how many times the respondent had (a) used marijuana, (b) used illegal drugs other than marijuana, (c) taken items worth between $1 and $50 that did not belong to them, (d) taken items worth $50 or more that did not belong to them, (e) purposely damaged or destroyed public or private property that did not belong to them, (f) illegally entered or at-
S. G . TIBBETTS
tempted to enter a building or car, (g) driven after drlnking alcohol when they were probably over the legal limit, (h) driven after using illegal drugs, and (i) seriously threatened to or actually slapped or hit someone else, i.e., assault. Seventy-nine percent of respondents' sco1:es on the offendmg scale were nonzero; however, the number of incidents reported by participants varied drastically, from O to 136. Thus, the scores on the scale items were standardized and then summed for each respondent. Coefficient alpha for this scale was .76. A principal components factor analysis of the items included in the scale and a scree plot showed that the largest gap between eigenvalues was between the first and second factor, which suggests thac the scale was unidimensional and, for our purposes, an indicator of offending among the participants. Descriptive statistics for variables are presented in Table 1 below. It is important to note that, although most respondents had nonzero values for the offending measure, the scores were positively skewed. Thus, TOBIT regression coefficients were estimated to examine the excent to which this skewness affected the estimated ordinary least squares regression estimates. Results were substantively the same, with no coefficients switching direction or gaining/losing significance. Therefore, only the OLS regression results are reported below. Independent variables.-Independent variables were various measures of self-conscious emotions including Shame, Guilt, and Pride, as well as demographic control variables. The Test of Self-conscious Affect (Tangney, et al., 1989) is a scenario-based measure of several self-conscious emotions, in which participants rate on a series of >-point scales the hkelihood of responding to 15 situations in ways that have been precoded to indicate guiltand shame-proneness as well as other self-conscious dispositions, e.g., pride, externahation, etc. This scale includes measures of guilt- and sharne-proneness, as well as pride.' The Person&ty Feelings Questionnaire-2 (Harder & Zalrna, 1990) is a 16-item adjective checklist on which respondents rate their experience of specified feelings designated as guilt or shame on a 5-point scale with anchors of 0: you never experience the f e e h g and 4: you experience the feeling continuously or almost continuously, reflecting how often they have experienced such f e e h g s (also see Ferguson & Crowley, 1997a). The Shame Proneness Scale (Tibbetts, 1997) is a scenario-based measure of shame-proneness, on which the respondents report (on a >-point scale) how they would feel or react to 35 hypothetical situations and attitudinal propen-
'The Test of Self-conscious Affect also includes measures of detachment, i.e., a disposition to respond in such a way that diminishes the im ortance of a situation, and externalization, i.e., a for the event ourwardly, which were nor exdisposition that one tends to deflecr responsi!ility amined in this study.
SELF-CONSCIOUSEMOTIONS AND OFFENDING
sities. These measures were chosen because they represent divergent ways of assessing self-conscious emotions and are commonly used measures of selfconscious emotions in the psychological and criminological literature. Three measures of shame-proneness were used: the Test of Self-conscious Affect, the Shame Proneness Scale, and the Personality Feehgs Questiomaire-2. As mentioned above, the first two (with 15 and 35 items, respectively) are scenario-based measures of shame-proneness, whereas the Person&ty Feehgs Questionnaire-2 is an adjective checkhst (with 10 items pertaining to shame) reflective of how often respondents experience f e e h g shame. For example, the latter simply lists words or phrases, such as "Sadness." On the other hand, the Test of Self-conscious Affect asks respondents their feelings on scenarios, such as "You make plans to meet a friend for lunch.. . At 5 o'clock, you realize you stood him u p . . . You would think 'I'm inconsiderate'." Previous studes (e.g., Tangney, et al., 1989; Harder & Zalma, 1990; Harder, et al., 1992; Harder, 1995; Tibbetts, 1997) have shown these measures to be reliable and valid indicators of shame-proneness. Item responses to the Test of Self-conscious Affect items were measured on a 5-point scale with anchors of 1: not Uely and 5: very likely. Responses to the items of the Shame Proneness Scale were coded on a 5-point scale that varied according to the particular item (see Tibbetts, 1997); nevertheless, all items were coded in a positive drection toward shame-proneness. Responses to the Personahty Feelings Questionnaire-2 items used anchors of 0: never experienced the f e e h g and 4: experience the f e e h g continuously or almost continuously. Coefficients alpha for the shame-proneness scales of the three measures were .79, .93, .76, respectively, which support the internal consistency among items in each scale measuring shame-proneness. Further principal components factor analyses and corresponding scree plots were examined for the items of each scale. These analyses showed that responses to virtually all items of each of the respective shame scale items had high loadings on one factor, which indcates that the scales were unidirnensional, measuring one factor believed to be shame-proneness based on prior srudies. Mean scores on the shame-proneness measures were compared to average scores reported by previous studies that uthzed these measures among similar samples, i.e., college students or adults primarily between the ages of 17 to 30 years. As reported in Table 1 below, the mean score was 43.1 (SD= 10.1) on the TOSCA-shame subscale for our sample. This score was similar to the average score (M=42.8, SD=8.1) reported by Tangney, Wagner, and Grarnzow (1992) for a similar sample of 264 college students. The mean score (M = 16.4, SD = 5.6) on the Personahty Feehgs Questionnaire-2 shame-proneness measure was similar to that (M= 16.1, SD =4.5) reported by Harder and Zalma (1990) for college students ages 17 to 22 years. Fi-
S. G. TIBBETTS
nally, the average score (M= 108.9, SD=20.3) on the Shame Proneness Scale was similar to the mean score (M= 106.7, SD= 15.8) reported by Tibbetts (1997) in a study of 604 college students. Statistical analyses (t tests) showed that none of the mean shame-proneness scores of our sample differed significantly from the reported scores in recent studies, which supports the v&dity of these measures across time and place. Two measures were used to assess guilt-proneness, the Test of Self-conscious Affect and the PersonaLty Feehgs Questionnaire-2, with 15 items and 6 items, respectively, measuring guilt-proneness. These two were chosen because they represent Avergent and commonly cited measures of guiltproneness in the psychological literature, whereas the criminological literature does not include a multiple-item measure of guilt-proneness. An example item for this scale was the response to the scenario-of "You make plans to meet a friend for lunch. At 5 o'clock, you realize you stood him up"; the "c" question was "You would try to make it up to him as soon as possible." Item responses to these items were measured on a >-point scale anchored by 1: not likely and 5 : very likely. Responses to the Personahty Feelings Questionnaire-2 items used anchors of 0: never experienced the f e h g and 4: experience the feeling continuously or almost continuously. Coefficient alpha for the TOSCA scale was .77 and for the PFQ-2 scale was .74. Principal components factor analyses and corresponding scree plots showed that the corresponding items of each scale loaded on one factor, which we assumed to be guilt-proneness. Mean scores on the guilt-proneness measures were compared to average scores reported by previous studies that utilized these measures among college students. As reported below in Table 1, the mean score was 58.6 (SD= 8.2) on the TOSCA guilt-proneness scale for our sample; this score was similar to the average score (M =58.3, SD= 6.6) reported by Tangney, et a/. (1992) for 264 college students. The mean score (M = 10.1, SD= 3.9) on the PFQ-2 guilt-proneness measure was similar to that (M = 9.9, SD = 3 . l ) reported by Harder and Zalrna (1990) for 37 college students. The t tests showed the mean scores d ~ dnot differ significantly from the reported scores in recent studies for either measure. This supports the validity of these measures across samples drawn from different places. The Test of Self-conscious Affect includes 10 items measuring two types of pride: Alpha pride and Beta pride, with 5 items measuring each concept of pride. As discussed above, alpha pride is a more self-oriented measure of pride that reflects general pride in self, e.g., "You would think: 'I am a trustworthy person'," whereas beta pride is a more situation-specific measure of pride that reflects pride in one's behavior, e.g., "You would think: 'I feel that my hard work paid off' " (Ferguson & Crowley, 1997b). Item responses were measured on a 5-point scale anchored by 1: not likely
SELF-CONSCIOUSEMOTIONS AND OFFENDING
and 5: very Uely. For the present purpose items indicating both Alpha and Beta pride were summed to form one index score. Coefficient alpha for the Pride scale was .79, and those for the 5-item scales of Alpha and Beta pride were .65 and .53, respectively, which indicates chat they should not be used as independent scales. Furthermore, a principal components factor analysis and a scree plot of the scale items showed that the 10 pride items loaded on one factor we assumed to be a tendency to experience feelings of pride. Mean scores on the Pride measure were compared to the average score reported by prior studies of college students. As reported in Table 1 below, the mean score on Pride was 41.1 (SD=5.5) on the TOSCA shame-proneness scale; this score was similar to the average score (M=40.4, SD=5.3) reported by Tangney, et al. (1992) for 264 college students. t tests showed that the scores did not differ significantly across studies. Demographic and control variables.-Three control variables were measured because studies (e.g., Stannard & Bowers, 1970; Hagan, Simpson, & Gillis, 1987; Bunn, Caudill, & Gropper, 1992; Tibbetts & Herz, 1996; Gfroerer, Greenblatt, & Wright, 1997; Wechsler, Dowdall, Maenner, Gledhill-Hoyt, & Lee, 1998) have shown such characteristics are often correlated with deviance in college samples: (a) sex, (b) grade point average (GPA), and (c) fraternity membership. Sex was coded 1 (female student) or 2 (male student). GPA was measured by one item that asked respondents to report the grade point average for the previous year. Coding was GPA = 1 (0-1.991, 2 (2.00-2.49), 3 (2.5-2.991, 4 (3.00-3.49), and 5 (3.504.00). Fraternity status was indicated by whether respondents were involved with the fraternity or sorority system (1=non-Fraternity and 2 =Fraternity). RESULTS Bivariate correlation coefficients (see Table 1) were examined to check for multicollinearity among the independent variable measures and assess construct and criterion-related validity of the various measures of self-conscious emotions. Examination showed some associations among the independent variables were above .60; specifically, the correlations of SPS-shame with PFQ-2-shame and TOSCA-shame were .63 and .67, respectively. However, estimated &agnostic statistics, e.g., variance inflation factors, tolerance, etc., indicated that none of the independent variables had such high colhearity with the other predictors that estimates in h e a r regression analyses would be misspecified so all variables were included. Bivariate correlations (see Table 1) among the Test of Self-conscious Affect, Personality Feelings Questionnaire-2, and Shame Proneness Scale measures of shame-proneness were .56, .63, and .67, respectively, indicating a robust association among the shame measures. These supported the criteri-
S. G. TIBBETTS
on-related validity of these shame-proneness measures. Thus, the first hypothesis was supported by the strong correlations, particularly the two statebased shame measures-TOSCA-shame and Shame Proneness Scale. TABLE 1 CORREUTION MATRIXFORALLVARUBLES ( N = 224) Variablet
1. TOSCA-shame 43.14 10.12 16.35 5.58 .56* 2. PFQ-2-shame 3. SPS-shame 108.88 20.26 .63* .67* 4. TOSCA-guilt 58.56 8.22 .56* .19* .44* 5. PFQ-2-guilt 10.07 3.89 .46* .61* .58* .37* 6. TOSCA-pride 41.13 5.53 -.I2 -.20* -.02 -.03 -.21* 7. Gender 1.52 .50 -.37* -36' -.28* -.31* -.15* .12 3.40 1.02 .10 -.14* -.04 .14* .02 -.I0 -.16* 8. GPA 9. Greekstatus 1.15 .36 .05 -.07 -.03 -.04 -.09 -.08 -.I1 -.I2 10. Offending .OO 5.28 -.22* -.09 -.28* -.44* -.27* .15* .22* -.16* .22* tTOSCA = Test of Self-conscious Affect; PFQ-2 = Personahty Feelings Questionnaire; SPS = Shame Proneness Scale. * p < .05, two-tailed.
The coefficient between the Test of Self-conscious Affect and PFQ-2 guilt-proneness scores was 3 7 . This correlation supported the second hypothesis that the scores on guilt measures would be consistent but the correlation was s m d . The low association is likely due to one being scenariobased whereas the other is an adjective checklist. Likely these tap ddferent aspects of guilt, respectively, state versus trait experiences of guilt (see Tangney, 1996). Nevertheless, the criterion-related validity of the guilt-proneness measures was generally supported by the correlation between the guilt scores as the positive correlations were in the predicted direction. Positive, significant associations were found for TOSCA-pride, sex, and fratern~tystatus. Negative correlations with criminal offending were found with TOSCA-shame, SPS-shame, TOSCA-guilt, PFQ-2-guilt, and GPA. Nevertheless, bivariate correlations do not control for the associations of other variables and, therefore, may imply a relation where one does not exist. Thus, multivariate regression models were estimated to test hypotheses of relations of self-conscious emotions with offending behavior. The ordinary least squares regression analysis of offending behavior is reported in Equation 1 of Table 2. Here five variables explained a significant amount of variance in offending behavior when other variables were controlled. Specifically, PFQ-2-shame, TOSCA-pride, and fraternity status had positive correlations with offending, whereas SPS-shame and TOSCA-guilt had inverse relationships. Ln the multivariate analysis, the estimated relationships of TOSCA-shame, PFQ-2-guilt, sex, and GPA with offending behav-
SELF-CONSCIOUS EMOTIONS AND OFFENDING
ior were not significant. Therefore, Hypothesis 3 was supported, given that the trait-based measure of shame (PFQ-2) had a positive relation with offending as predicted. Hypothesis 4 had mixed support in the sense that another measure of shame, the state-based Shame Proneness Scale, had a significant, inverse correlation with criminal offending. Full support was lacking because the other state-based measure of shame had no significant correlation with offendmg when other factors were controlled. TABLE 2 SUMMARY O F ORDINARY LEASTSQUARES REGRESSIONANALYSIS OF REPORTED CRIMINAL OFFENDING ( N = 224)
Equation 1 SEb Beta
Equation 2 SE b Beta
TOSCA-shame PFQ-2-shame SPS-shame Shame Index TOSCA-guilt PFQ-2-guilt Guilt Index TOSCA-pride Gender
GPA Greek Status (Constant) Note.-For Equation 1 R2= .3O; For Equation 2 R2= .27. tTOSCA =Test of Self-conscious Mecr; PFQ-2 =Personality Feelings Questionnaire; SPS = Shame Proneness Scale. Dashes indicate no applicable estimate. *p < .05, two-tailed.
Hypothesis 5 was generally supported, given that one measure of guilt (trait-based PFQ-2). had an inverse, significant correlation with criminal offendmg. Also, Hypothesis 6 was supported by the positive correlation between pride and criminal offending Fmally, Hypothesis 7 was generally supported as there was a notable decrease m the correlations between two of the three demographic variables' and criminal offending. Specifically, al- though the bivariate correlations between sex, GPA, and fraternity/sorority status with criminal offending were significant, the multivariate regression model of all variables showed that, when the scores on self-conscious emotion measures were taken into account, the relationships between sex and GPA were no longer significant. To examine the cumulative relationships of the self-conscious emotions -shame and guilt-with criminal offending, analyses were based on each respondent being assigned a composite score based on the three measures of shame, as well as a score on the two measures of guilt based on factor load-
S. G. TIBBETTS
ings from a factor analysis. Specifically, factor analyses were conducted for each category (shame and guilt) in which one factor was extracted from the three measures of shame. Next, the factor loadings of each of the separate measures on the extracted factor were then multiplied by the standardized scores on the corresponding variable for each case, and summed with terms from the other measures to obtain a composite measure for both shame and guilt. These composite measures were called shame index and guilt index. As reported in Equation 2 of Table 2, this analysis showed respondents who scored high on the guilt index score were significantly less likely to report engaging in criminal offending, which supports Hypothesis 5 . That Hypotheses 3 and 4 may be supported by this analysis cannot be confirmed. To clarify, the shame index measure had no significant correlation with criminal offending, perhaps reflecting the positive relation of state-based shame (SPS shame) and trait-based shame (PFQ-2-shame) canceling each other out. Finally, Hypothesis 5 was supported, as individuals who scored high on pride measures were more likely to report engaging in criminal offendmg even while accounting for these indexes. DISCUSSION This study examined the construct and criterion-related validty of various measures of self-conscious emotions and, also, the shared variance of various traits of self-conscious emotions-shame, guilt, and pride-with selfreported offending behavior. Analyses showed that the measures of self-conscious emotions appeared generally to be reliable and vahd measures of the constructs that they were intended to measure. Further, regression analyses indcated that scores for pride had a positive correlation with self-reported criminal activity, whereas one measure of guilt (TOSCA-guilt) was negatively associated with offending. Finally, fraternity/sorority status had a significant association with criminal offending even after controlling for self-conscious measures. The relationships of shame with criminal offending varied with the type of measure used to indicate shame-proneness, while controlling for other variables. Specifically, a chronic, i.e., trait, measure of shame (PFQ-2-shame) had a positive correlation with criminal offending. However, a more adaptive, state-based measure of shame (Shame Proneness Scale) had a negative correlation with offending, and another adaptive measure of shame (TOSCA-shame) had a nonsignificant one. More generally, diagnostic statistics estimating the overall fit of the models (R2 = .3O) indicated some support for the proposition that self-conscious emotions are of use in explaining offending behavior. The relations of self-conscious emotions seemed to account for at least some of the associations of demographic variables with offending, except for fraternity/sorority status.
SELF-CONSCIOUS EMOTIONS AND OFFENDING
More specifically, findmgs showed that the Test of Self-conscious Affect, Personality Feehgs Questionnaire-2, and Shame Proneness Scale were, for the most part, reliable and valid indicators of the self-conscious emotions that they were designed to measure. The internal reliability of each of these measures was supported by coefficients in the present study. The criterionrelated validity of the shame and guilt measures was supported by the moderate to strong correlations among the various measures of each emotion. Construct v&&ty of the shame and guilt measures was generally supported by correlations between shame and guilt measures with sex. Finally, the construct and criterion-related validity for the pride measure was supported. The positive relationship between offending and trait-based shame is consistent with recent findmgs (Harder, et al., 1992; Allan, et al., 1994; Tangney, 1996) that has shown that individuals who exhibit trait levels of sharne-proneness are more likely to be disposed to exhibit maladaptive behavior. On the other hand, both inverse and n d findings for rated shame with offending observed for test items on more situational or state-oriented shame also support previous conclusions (Grasrnick & Bursik, 1990; Grasmick, Burslk, & Cochran, 1991; Grasmick, Bursik, & h s e y , 1991; Grasmick, Blackwell, & Bursik, 1993; Grasmick, Burs&, & Arneklev, 1993; Nagin & Paternoster, 1993; Paternoster & Sirnpson, 1993, 1996; M a n , et al., 1994; Tangney, 1996). These state-based measures seem to reflect the adaptive nature of shame as a self-conscious emotion in constraining or not contributing to individual criminahty when feehgs of shame are not more longterm or chronic. Consistent with these findmgs, the observed significant inhbitory relationship between situationally based guilt with criminal offending supported prior work on state-oriented gullt as measured by the Test of Self-conscious Affect (Tangney, et al., 1995; Tangney, 1996; Tangne~,Wagner, Barlow, Marschall, & Gramzow, 1996). On the other hand, the null relation for the trait-based measure of guilt is consistent with recent psychological literature (for reviews, see Tangney, et al., 1992, 1995; Tangney, 1996; Dorahy & Schumaker, 1997) wherein individuals who exhibit chronic gdt-proneness are less &ely to commit criminal activity. Thus, the former, state-based measure can be interpreted as reflecting the reparative or adaptive nature of guilt as a self-conscious emotion in constraining criminal behavior. The latter form of g d t , as measured by the Personality Feelings Questionnaire-2, represents a more trait form of guilt which (Harder, et al., 1992; Harder, 1995; Tangney, et al., 199.5; Tangney, 1996; Dorahy & Schumaker, 1997) is less &ely to be reparative or inhibitory of committing criminal offenses. These findings are also consistent with recent psychological conceptual frameworks (for an extensive review, see Tangney & Fischer, 1995) that suggest chronic traits of self-conscious emotions of guilt and shame are &ely to be correlated with
S. G. TIBBETTS
negative human perceptions and behavior, whereas &spositions to anticipate feelings of guilt or shame in specific situatiorls or states of self-conscious emotions tend to encourage positive attitudes and adaptive social behaviors. Also important were the correlations with criminal offendmg and rated pride. The idea that offenders are more disposed to feelings of pride is consistent with studies and theoretical models of criminality (Katz, 1988; Gottfredson & Hirschi, 1990). These models claim that low self-control and incitement to commit crime represent a basic association in predicting which individuals will engage in risk-taking and impulsive behaviors, as well as theoretical models which propose individuals commit offenses because they desire a f e e h g of risk or thr~Ufrom offending. Individuals who have a tendency to feel pride are also ltkely to blame their deviant behavior on external sources or to remove themselves from blame in their commission of offenses (Sykes & Matza, 1957; Katz, 1988). The positive correlations of pride with criminal offending in the present study support those reported in the few empirical studies of this relationship (Katz, 1988; Shields & Whitehall, 1991; Simourd, 1997). The consistency of the findings across studies suggests that a disposition toward feelings of pride is important in the etiology of offending behavior and, therefore, warrants more attention from theorists in criminology. Furthermore, individual differences in pride may condition association of other predictors, e.g., demographic, attitudinal, etc., of criminality. Researchers should examine the interactions between pride and other predictors, as well as clarify the differential offending of two aspects of pride. Alpha pride is a more self-oriented measure that reflects general pride in self, i.e., more chronic nature of pride, whereas beta pride is a more situation-specific measure that reflects pride in one's behavior (Tangney, et al., 1989; Ferguson & Crowley, 1997b); see the Method for elaboration. Among correlations of the control variables with offending behavior, one significant relation in the regression models was for fraternity/sorority status. This finding is consistent with studies (e.g., Stannard & Bowers, 1970; Gfroerer, et al., 1997; Wechsler, et al., 1998) that show that fraternity members tend to report more offending, perhaps with increased opportunity and peer influences to engage in such behavior. Fraternity affihation should be examined further, particularly for relations of peer associations with varying dispositions of self-conscious emotions and perhaps higher hkelhood of reportmg. The regression equations showed that the associations of sex and GPA with offending were nonsignificant as in prior work (e.g., Lutwak & Ferrari, 1996; Tibbetts & Herz, 1996) wherein relations of demograhpic variables with criminal offending can often be accounted for when attitudinal or dispositional measures are included. It is &ely that ddferences in rated self-con-
SELF-CONSCIOUS EMOTIONS AND OFFENDING
scious emotions account for disparities between relatively good and bad students, as well as between men and women. Several limitations should be considered when examining present results. Importantly, the study design was a cross-sectional, self-report survey, and all data were simply correlational, so any causal interpretations are not defensible. The sample was one of convenience drawn from college students at one institution; therefore, the generalizabhty of the results must be established. However, the prevalence of the specific offending events tends to be high among college students. In addition, this study rehed on self-reported behavior; however, the reliabhty and vddity of such self-reports has been supported (for a review, see Hindelang, Hirschi, & Weis, 1981). Finally, not all self-conscious emotions, namely, embarrassment, were examined but need to be assessed in the relation with criminality. Despite these h i t a t i o n s , the findings can be interpreted as evidence that higher ratings of certain self-conscious emotions are related to the behavior of persons in commission of dlegal offenses. Individual ddferences in dispositions toward self-conscious emotions account for the covariance of certain demographic variables, such as sex And, some reported ddferences in offending between men and women as well as those between above-average vs below-average students, may reflect propensities to anticipate feelings of self-conscious emotions. This study has shown that the dispositions to experience various selfconscious emotions are related to self-reported criminal offending. Further, the type of self-conscious emotion, e.g., guilt versus pride, and the nature of such inchations, e.g., states versus traits, have relations of varying magnitude and direction with individual differences in propensities to engage in criminal activity. It is clear that studies on prehctors and inhibitors of committing deviant behavior should include measures of self-conscious emotions, particularly dspositions to experience shame, guilt, and pride, to understand their associations with criminal offending. REFERENCES ALLAN. S., GILBERT, P., &GOSS.K. (1994) An exploration of shame measures: 11. Psychopathol-
ogy. Personalify and lndiuidzral D~fference.r,17, 719-722. AUSUBEL. D. I? (1955) Relationships between shame and guilt in the socializing process. Psychological Reuiezu, 62, 378-390. BRAITHWAITE, 1. (1989) Crime, shame, and reinfegrafio~z.New York: Cambridge Univer. Press. BUNN, D., CAUDILL, S., &GROPPER, D. (1992) Crime in the classroom: an economic analysis of undergraduate studenr cheating behavior. Research in Economic Educafion, 23, 197-207. CLARKE, R. (1995) Siruational crime prevention. In M. Tonry & D. Farrington (Eds.), Bui[diizg a safer sociefy: sfrafegicapproaches ro crime prevenfio~z.Chicago, IL: Univer. of Chicago. DORAHY, M. J., &SCHUMAKER, J. F. (1997) Dissociative functioning and its relationship to state and trait guilt in the non-c11n1cJpopulation. Personality and Indioidual Drfferences, 23, 967-972. EDELMANN, R. J. (1987) The psychologv of embarrassment. Chichester, Eng.: Wiley.
S. G . TIBBETTS
D., &LANNING, K. (1998) Shame, guilt, ego development, and the five-factor model EINSTEIN, of personality. Journal ofPersonality, 66, 555-582. EISENBERG, N. (2000) Emotion, regulation, and moral development. Annual Review of Psycholonv. - 5 1, 665-697. Eus. L. A., & SIMPSON, S. S. (1995) Informal sanction threats and corporate crime: additive versus multiplicative models. Iournal of Research in Crime and Delinquency, 32, 399-424. FERCUSON, T. J., &CROWLEY,S. L. (1997a) Gender dfierences in the organization of guilt and shame. Sex Roles, 37, 19-44. FERGUSON, T. J., &CROWLEY, S. L. (1997b) Measure for measure: a multitrait-multimethod analysis of guilt and shame. Journal of Personality Assessment, 69, 425-44 1. FERGUSON, T. J., &STEGGE,H. (1995) Emotional states and traits in children: the case of guilt and shame. In J. P. Tangne & K. W. Fischer (Eds.), Self-conscious emotions: shame, guilt, eembarrassmenf, and p r i d . New York: Guilford. Pp. 174-197. FISCHER, K. W., & TANGNEY, J. P. (1995) Self-conscious emotions and the affect revolution: framework and overview. In J. P. Tangney & K. W. Fischer (Eds.), Self-conscious emotions: shame, guilt, embarrassment, and pride. New York: GuiKord. Pp. 3-23. GFROERER, J. C., GREENEMIT,J. C.. &WRIGHT. D. A. (1997) Substance use in the US college-age population: differences according to educational status and living arrangement. American Journal of Public Health, 87, 62-65. GIBBS,J., & GIEVER,D. (1995) Self-control and its manifestations among universi students: an empirical test of Gottfredson and Hirschi's general theory. Justice Quarter$, 12, 231255. Goss, K.. GIL~ERT. P., &ALLAN.S. (1994) An ex loration of shame measures: I. The other as , 713-717. shamer scale. Persorzali~yand Individual ~ z g r e n c e s 17, GOTTFREDSON. M., & HIRSCHI,T. (1990) A general theory of crime. Stanford, CA: Stanford Univer. Press. GRAHAM,S., & WEINER,G. (1986) From an attributional theory of emotion to development psychology: a roundtrip ticket? Social Cognition, 4, 152-179. GRASMICK, H., BLACKWELL, B., &BURSIK, R. (1993) Changes over time in gender dfierences in perceived risk of sanctions. Low and Society Review, 27, 679-705. GRASMICK, H.. &BURSIK, R. (1990) Conscience, significant others, and rational choice: extending the deterrence model. Low and Society Review, 24, 837-861. GRASMICK, H.. BURSIK. R., &ARNEKLEV, B. (1993) Reduction in drunk driving as a response to increased threats of shame, embarrassment, and legal sanctions. Criminology, 31, 41-67. GRASMICK, H., BURSIK, R., &COCHRAN, J. (1991) 'Render unto Caesar what is Caesar's: religiosity and taxpayers' inclinations to cheat. Sociological Quarlerly, 32, 251-266. GRASMICK, H., BURSIK,R , & KINSEY,K. (1991) Shame and embarrassment as deterrents to noncompliance with the law. Environment and Behavior, 23. 233-251. b c m , J., SIMPSON, J., &GILLIS,A. R. (1987) Class in the households: a power-control theory of gender and delinquency. American Journal of Sociology, 92, 788-816. HARDER, D. W. (1995) Shame and guilt assessment, and relationshi s of shame and guilt proneness to psycho athology. In J. P. Tangney & K. W. ~ i s c i e r(Eds.), &lf-consnozdi emotions: shame, gui( embarrassment, and pride. New York: Guilford. Pp. 368-392. HARDER,D. W., CUTLER, L., &ROCURT, L. (1992) Assessment of shame and guilt and their relationships to psychopathology ]ozrrnal of Personality Assessment, 59, 584-604. HARDER, D. W., &ZALMA.A. (1990) Two promising shame and guilt scales: a construct validity comparison. Journal ofPersorrafi/y Assessment, 55, 729-745. h v , C. (2001) An explorato test of Braithwaite's reintegrative shaming theory. Jozrrnal of Resea* in Crime and ~ x i f ~ z r e n38, c ~132-153. , HINDELANG, M., HIRSCHI. T., &WEIS,J. (1981) Measuring delinqirency. Beverly W s , CA: Sage. HOGLUND, C. L., &NICHOLAS, K. B. (1995) Shame, guilt, and anger in college students exposed to abusive family environments. Journal of Family Violence, 10, 141-157. HUDLEY, C. A. (1992) Attributions for pride, anger, and guilt among incarcerated adolescents. Criminal Justice and Behavior, 19, 189-205. KATZ, 1. (1988) The sedtrction of crime: moral and sensual attraction of doing evil. New York: Basic Books.
SELF-CONSCIOUS EMOTIONS AND OFFENDING
LEITH,K. I?, &BAUMEISTER, R. F. (1998) Empathy, shame, guilt, and narratives of interpersonal conflicts: guilt-prone people are better at perspective takmg. Journal of Personality, 66, 1-37. H. B. (1971) Shame and grrilt in neurosb. New York: International Universities Press. LEWIS, LEWIS,M. (1992) Shame: the exposed re& New York: Macmillan. LEWIS.M. (1995) Embarrassment: the emotion of self-exposure and evaluation. In J. P. Tangney & K. W. Fischer (Eds.), Self-conscious emofions: shame, guih, embarrassment, and pride. New York: Guilford. Pp. 198-218. L ~ A KN.,, &FERRARI, 1. R (1996) Mord affect and c o g n i t i v e ~ s s e s differenuaiig : shame from guilt among men and women. Personality and Indivr ual Dzfferences,21, 891-896. MASCOLO, M., & FISCHER, K. (1995) Developmental transformations in a praisals for pride, shame, and guilt. In J. P. Tangne & K. W. Fischer (Eds.), ~egonscibzrsemotions: shame, guilt, embarrassment, and piL. New York: Guilford. Pp. 64-113. MILLER, R. S. (1995) Embarrassment and social behavior. In J. P. Tangney & K. W. Fischer (Eds.), Self-consriorrs emofions: shame, gnilf, embarrassmenf, and pride. New York: Guilford. Pp. 322-339. NAGIN,D., &PATERNOSTER, R. (1993) Enduring individual diFferences and rational choice cheories of crime. Law and Society Rmiew, 27, 467-496. PATERNOSTER, R., & SIMPSON, S. (1993) A rational choice theory of corporate crime. In R. Clarke & M. Felson (Eds.), Advances in m'minological theory. Vol. 5. New Brunswick, NJ: Transaction Press. Pp. 37-58. PATERNOSTER. R.,&SIMPSON, S. (1996) Sanccion threats and appeals to moraliry: testing a rational choice model of corporate crime. Low and Sociefy Review, 30, 378-399. M. B. (1971) Shame and guilf. New York: Nolton. PIERS.G., &SINGER. A,, &TIBBE~TS, S. (1996) Specifymg the direct and indirect effects of low self-control PIQUERO, and situational factors in offenders' decision making: toward a more complete model of rational offending. Jtistice Quarterly, 13, 481 -510. QUILES,Z. N., & BYBEE,1. (1997) Chronic and predis ositional guilt: relations to mental health, prosocial behavior, and religiosity. Journal o/Jersonalify Assessmenf, 69, 104-126. SHIELDS, I. W., &WHITEHALL, G. C. (1991) The Pride in Delinquency scale. Paper presented at the eastern Onrario Correctional Psychologists' winter conference, Burritts Rapids, Canada. SIMOURD. D. J. (1997) The Criminal Sentiments Scale-Modified and Pride in Delinquency scale: psychometric propetties and construct validity of rwo measures of criminal attitudes. Crimzrzal]ustice and Behauior, 24, 52-70. STANNARD, C., & BOWERS,W. (1970) The college fraternity as an opportunity structure for meeting academic demands. Social Problems, 17, 371.390. SYKES.G. M., & &TU, D. (1957) Techniques of neutralization: a theory of delinquency. American Sociological Review, 22, 664-670. TANGNEY, J . I? (1989) Shame-proneness, guilt-proneness, and inrerpersonal processes. Paper presented at the annual meeting of the Society for Research in Child Development, Kansas City, KS. J. I? (1990) Assessing individual differences in proneness to shame and guilt: develTANGNEY, opment of the Self-conscious M e c t and Attribution Inventory. Journal of Personality and Social Psychology, 59, 102-111. 1. l? (1995) Shame and guilt in interpersonal relationships. In J. P. Tan ney & K. W. TANGNEY, Fischer (Eds.), Self-conscious emotions: shame, gzrili, embarrassment, and p n i . New York: Gullford. Pp. 114-139. TANGNEY, J. I? (1996) Conce rual and methodological issues in the assessment of shame and guilt. Behavioral ~esearcETherapy,34, 74 1-754. TANGNEY. 1. I?, BURGGRAF, S., &WAGNER, I? (1995) Shame-proneness, guilt- roneness, and psyemofions: chological symptoms. In J. P. Tangney & K. W. Fischer (Eds.), ~e~consnbrrs shame, guilt, embarrassment, and pride. New York: Guilford. Pp. 343-367. TANGNEY, J. I?, & FrscHER, K. W. (1995) SeU-conscious emotions: shame, grtilf, embarrassment, and pride. New York: Gu~lford. TANGNEY. J . I?, MILLER, R. S., FLICKER, L., &BARLOW, D. H. (1996) Are shame, guilr, and embarrassment distinct emotions? Jorrrnal ofPersonality and Sociol Psychologv, 6, 1256-1269.
S. G . TIBBETTS
TANGNEY, J. l?, WAGNER, P. E., BARLOW, D., MARSCHALL, D., &GRAMZOW, R. (1996) The relation between shame and guilt to constructive vs. destructive responses to anger across the lifespan. Journal of Personality and Sonal Psychology. 70, 797-809. TANGNEY, J. I?, WAGNER, I? E., GAVIAS,I., & GRAMZOW, R. (1991) The Test of Self-consczous Affect for Adolescents (TOSCA-A). Fairfax, VA: George Mason Univer. J. l?, WAGNER, P. E., &GRAMZOW, R. (1989) The Test of Self-conscious Affecf. Fairfax, TANGNEY, VA: George Mason Univer. R. (1992) Proneness to shame, proneness to guilt, TANGNEY, 1. I?, WAGNERP E , &GRAMZOW, and psychopathology )ournal of Abnormal psycho log^, 103, 469-478. TIBBE~TS, S. G. (1997) Shame and rational choice in offendmg decisions. Criminal Justice and Behavior, 24, 234-255. T r e s ~ r r s S. , G., & HERZ,D. (1996) Gender differences in factors of social control and rational choice. Deuiatrf Behavior, 17, 183-208. T~BBET-rs, S. G., &MYERS,D. (1999) LOWself-control, rational choice, and student test cheating. American Journal of Criminal Jtrsfice,23, 179-200. WECHSLER, H., DOWDALL, G. W., MAENNER, G., GLEDHILL-HOYT, J . , &LEE,H. (1998) Changes in binge drinking and related problems among American college students between 1993 and 1997. The Journal of American College Health, 47, 57-68. B. (1986) An attribufional theory of motivation nnd emotion. New York: Springer-VerWEINER, lag. WILLIAMS, K. R., &HAWKINS, R. (1986) Perceptual research on general deterrence: a critical review. Low and Society Review, 20, 545-561. WORTLEY, R. (1996) Guilt, shame and situation crime prevention. In R. Homel (Ed.), The politics and practice of situational crime preuetztion. New York: Criminal Justice Press. Pp. 115-132.
Accepted ~Moy27, 2003