Shame and Guilt: Relationships of Test of Self-Conscious Affect Measures With Psychological Adjustment and Gender Differences in Iran

Shame and guilt are morally relevant affects that play a central role in interpersonal and social life. As Tangney (2003) emphasizes, for example, “Shame and guilt are typically mentioned in the same breath as moral emotions that inhibit antisocial, morally objectionable behavior” (p. 387). Unsurprisingly, therefore, social scientific interest has often focused on attempts to understand these moral affects. Perhaps most famous were early psychoanalytic efforts to describe guilt as an emotionally disturbing influence of the superego on the ego (e.g., Freud, 1923/1960). As Dost and Yagmurlu (2008) make clear, however, psychological theory and research subsequently suggested that guilt may also be important in the adaptive development of self-responsibility. Shame and guilt have attracted attention in anthropological research as well. In her analysis of Japanese culture, for instance, Benedict (1967) argued that shame represents a public and guilt a private emotional response, a claim that remains the subject of controversy (Dost & Yagmurlu, 2008; Tangney, Miller, Flicker, & Barlow, 1996; Taylor, 1985).

Especially influential in recent research attempts to differentiate between these two moral emotions have been the psychotherapeutic insights of Lewis (1971). She argued that shame results from a global condemnation of the whole self, but guilt reflects instead a more limited negative evaluation of specific behaviors. The Test of Self-Conscious Affects (TOSCA) operationalizes this distinction and yields a consistent empirical picture. The TOSCA Shame Scale displays linkages with maladjustment, whereas the TOSCA Guilt Scale tends to predict adjustment (Tangney, 1995, 2003; Tangney & Dearing, 2002).

Emde and Oppenheim (1995) suggest, “The self-conscious moral emotions, including shame and guilt, are important regulators that are likely to operate differently in different cultures” (p. 432). Kitayama, Markus, and Matsumoto (1995) more specifically distinguish between Western cultures in which independence is in the foreground of personality development with interdependence in the background, in contrast to non-Western cultures in which the prominence of the two types of self-functioning is reversed. Most studies examining the TOSCA Shame and Guilt Scales have used Western samples, and an important question is whether these measures have the same implications in non-Western societies (Tangney, 2003).

Interestingly, therefore, Furukawa, Tangney, and Higashibara (2012) recently discovered mean differences in the self-reported TOSCA Shame and Guilt scores of Korean, Japanese, and American children, but relationships of these affects with other measures seemed largely consistent across samples. Similar correlations across cultures suggest that TOSCA Shame and Guilt Scales have parallel connotations in Western and non-Western societies. On the other hand, Japanese investigations using university age students as the research participants have confirmed that the TOSCA Shame Scale predicts maladjustment, but they have offered little or no support for the idea that the TOSCA Guilt Scale operationalizes adjustment (Hasui et al., 2009; Uji, Kitamura, & Nagata, 2011a,b; Uji, Nagata, & Kitamura, 2012). In short, the meaning of the TOSCA Shame and Guilt Scales across cultures clearly deserves additional research attention, especially in older samples.

The present study sought to extend analysis of the TOSCA Shame and Guilt Scales to another non-Western cultural context, Iran. Procedures also examined the implications of these measures in university-age Iranians. As a formally Islamic society, Iran is more strongly organized around communal religious beliefs and practices than are more “secular” Western societies, and thus should encourage more interdependent forms of self-functioning. Given the recent observation of parallel correlations across Korean, Japanese, and American children (Furukawa et al., 2012), the general expectation was that the TOSCA Shame Scale would predict maladjustment and that the TOSCA Guilt Scale would predict adjustment in Iran, just as it does in university-age research participants in the West (e.g., Tangney & Dearing, 2002). On the other hand, examination of Iranian university students made it possible to explore the possibility suggested in Japanese studies that the TOSCA Guilt Scale might not have more adaptive implications in older non-Western samples (Hasui et al. 2009; Uji et al., 2011a,b, 2012).

Procedures also analyzed possible gender differences. A vast research literature demonstrates that “female participants consistently report greater shame and guilt than their male counterparts” (Tangney & Dearing, 2002, their emphasis). This gender difference perhaps reflects the generally stronger interpersonal sensitivity of women as revealed, for example, in their higher levels of empathy (e.g., Davis, 1983; Jonason, Lyons, Bethell, & Ross, 2013). It was of interest to determine if such relationships would hold true in Iran as well.

Method

Results

As Table 2 makes clear, the TOSCA Shame Scale correlated negatively with integrative self-knowledge and mindfulness and positively with internalized self-criticism, comparative self-criticism, rumination, and impaired control over mental activity.

The TOSCA Guilt Scale predicted higher levels of integrative self-knowledge, mindfulness, and reflection and lower levels of internalized self-criticism, impaired control over mental activity, interpersonal ambivalence, and aggression. All other significant relationships conformed to theoretical expectations. In other words, measures of psychological adjustment correlated positively with each other as did indicators of psychological maladjustment, and negative relationships appeared between these two types of variables.

MANOVA results uncovered overall gender differences, Wilks’ Lambda = .789, F(13, 118) = 2.43, p < .01. Women (4.44 + 0.09), for example, scored higher than men (4.06 + 0.08) on guilt; but men (3.26 + 0.07) and women (3.14 + 0.07) did not differ on shame.

Perhaps more importantly, however, the two genders displayed a different pattern of correlations, Box’s M = 157.768, F (91, 50532.905) = 1.49, p < .01. This outcome pointed toward a need to evaluate possible interactions of shame and guilt with gender in predicting the other constructs. The first step of multiple regression procedures, therefore, used gender and shame to predict each of the other measures, and then the second step examined their interaction. Guilt rather than shame and its interaction then served with gender as predictors in a second set of multiple regressions. Prior to these analyses, standardization of the two TOSCA measures addressed the potential problem of multicollinearity (Aiken & West, 1991).

Table 3 reviews the regression results when shame, gender, and their interaction served as predictors. In these analyses, scoring of the gender variable employed 0 for women and 1 for men. Women, therefore, scored higher on guilt and mindfulness. Significant gender-by-sex interactions appeared for 6 other variables: self-control, rumination, impaired control over mental activity, interpersonal sensitivity, interpersonal ambivalence, and aggression. Figure 1 depicts these significant interactions and demonstrates that the TOSCA Shame Scale predicted psychological functioning in women but not in men. All these significant relationships for women conformed to theoretical expectations. Specifically, shame displayed a negative association with self-control and positive linkages with the five indices of maladjustment. For men, counterintuitive tendencies of shame to display negative linkages with interpersonal sensitivity (β = -.15, p = .22) and interpersonal ambiguity (β = -.19, p = .12) failed to reach statistical significance.

Only one noteworthy new outcome appeared when guilt replaced shame in these multiple regressions. Gender and guilt interacted to predict integrative self-knowledge: ΔR² = .04, p < .05 with the Interaction β = -.35, p <.05. This result occurred because guilt displayed a direct connection with integrative self-knowledge in women, β = .45, p < .001, but not in men, β = .13, p = .28.

Discussion

As in numerous previous Western investigations (Tangney & Dearing, 2002), the TOSCA Shame and Guilt Scales in Iran predicted maladjustment and adjustment, respectively. Researchers often describe shame and guilt as self-conscious affects (e.g., Tangney, 1995), and the present data confirmed the relevance of these two emotions for “self-consciousness” in Iran. Specifically, the TOSCA Shame Scale, as a presumed form of maladaptive self-consciousness, displayed inverse linkages with the psychological benefits of integrative self-knowledge and mindfulness and positive associations with two measures of obsessive self-experience. Conversely, the TOSCA Guilt Scale, hypothesized to reflect adaptive self-consciousness, correlated positively with integrative self-knowledge, mindfulness, and self-reflection and also correlated negatively with the obsessiveness of impaired control over mental activity. Among other things, these relationships with self-knowledge measures were important in establishing the TOSCA instruments as relevant to psychological processes deemed to be crucial in achieving the Iranian Muslim moral ideal (Motahharī, 2000; Shimamoto, 2008).

Other possible linkages with moral functioning seemed apparent as well. Internalized and comparative self-criticism operationalized self-perceived failures to meet standards that have at least indirect implications for morality. Both correlated positively with the TOSCA Shame Scale. A stronger relationship of comparative self-criticism with shame seemed consistent with the assumption that this measure might be closer to the experience of shame than would be internalized self-criticism. In contrast, the TOSCA Guilt Scale exhibited a negative relationship with internalized self-criticism, suggesting perhaps that emotional guilt successfully prevented violations of personal standards. On the other hand, Giner-Sorolla, Piazza, and Espinosa (2011) recently argued that the TOSCA Guilt Scale does not so much measure an emotion as a motivation to put things right after transgressions of personal standards. The negative correlation of guilt with internalized self-criticism perhaps supported this alternative suggestion.

At least in terms of full sample correlations, only limited support appeared for the expectation that TOSCA Shame and Guilt Scales would predict interpersonal maladjustment. The TOSCA Shame Scale failed to correlate with any measure from the Inventory of Interpersonal Problems. As hypothesized, however, TOSCA Guilt did correlate negatively with interpersonal ambivalence and with aggression. This relationship with aggression supported the possibility that in Iran as elsewhere, guilt may have a potential to promote more constructive responses to anger (Tangney, 2003).

Attention to gender differences proved to be important. Iranian women scored higher than Iranian men on guilt, replicating a consistent effect observed in the West (Tangney & Dearing, 2002). However, the two genders did not differ in terms of their average levels of shame. The TOSCA Shame Scale did display a number of expected correlations with other measures in the full sample; so, it appeared to influence psychological functioning in males as well as in females. The failure of women to exhibit higher average levels of shame, nevertheless, stood in clear contrast to previous findings in the West (Tangney & Dearing, 2002).

Especially noteworthy and unexpected were interactions between gender and the two TOSCA measures. These results had at least four implications. First, Iranian women seemed more sensitive and Iranian men more insensitive to shame. Women displayed six shame relationships with other variables that proved to be nonsignificant in men. Hence, Iranian women did fail to display the higher average TOSCA Shame scores that have been observed consistently elsewhere, but significant interactions still confirmed that shame was a more noteworthy feature of female psychological functioning in Iran.

Second, these gender differences could not be interpreted as evidence that women were more specifically vulnerable to the maladjusted potentials of self-conscious affect. This was so because gender also interacted with guilt to predict higher levels of integrative self-knowledge in women but not in men. The suggestion, therefore, was that women were generally more sensitive than men to both the adjusted and maladjusted potentials of self-conscious affect, a conclusion that conforms to previous research findings (e.g., Davis, 1983; Jonason, Lyons, Bethell, & Ross, 2013).

Third, interactions did not mean that Iranian men were wholly insensitive to the influences of shame and guilt. As noted already, significant full-sample TOSCA correlations with at least some of the other scales confirmed the psychological importance of shame and guilt as measured by the TOSCA in men as well as in women.

Finally, interactions produced additional support for the thinking behind the hypotheses of this project. Self-control seemed obviously relevant to moral psychological functioning both generally and also specifically in Iran (Motahharī, 2000; Shimamoto, 2008); yet, in the full sample, it failed to predict either shame or guilt. In women, however, self-control correlated negatively with the TOSCA Shame Scale. In addition, hypothesized shame linkages with interpersonal maladjustment failed to materialize in the full sample, but in women, TOSCA Shame scores did, as expected, correlate positively with interpersonal sensitivity, interpersonal ambiguity, and aggression.

Deserving emphasis is the fact the TOSCA Guilt Scale predicted better mental health in university-age Iranians. Previously, Furukawa et al. (2012) found that TOSCA Shame and Guilt measures displayed similar correlational implications in Korean, Japanese, and American children; however, Japanese studies examining university students suggest the TOSCA Guilt Scale does not display the linkages with adjustment that are often observed in the West. The present data, therefore, demonstrated that in at least some non-Western societies the TOSCA Guilt Scale can have adaptive implications in university student samples. Of course, not all non-Western societies are the same, and when combined with the observations of Furukawa et al. (2012), this apparent contrast between Japanese and Iranian TOSCA Guilt Scale results may point toward important cultural differences in the development of moral affects.

Relationships among the other, non-TOSCA scales conformed to expectations. Significant correlations among intrapersonal adjustment measures proved to be positive, as did the significant relationships among indices of intrapersonal and interpersonal maladjustment. In addition, all significant linkages between adjustment and maladjustment proved to be negative. Data for integrative self-knowledge were perhaps especially noteworthy because this was the only scale that exhibited expected relationships with all other measures. Such consistency seemed to support suggestions that self-knowledge may be a centrally important psychological ideal within Iranian and presumably other Muslim societies (Ghorbani et al., 2011).