Because of its multidimensional nature, empathy as defined by Davis a**esses a much wider spectrum of behavior, from prosocial to antisocial, and acts as a happy medium within the tangled web of previously established definitions and conceptualizations. In fact, Eisenberg and Miller (1987) concluded from their metaan*lysis, “It is clear . . . the degree of positive a**ociation between measures of empathy and prosocial behavior varies depending on the method of measuring empathy,” and furthermore, in reference to a cluster of measures that includes the Davis IRI, that “the relations [are] strongest for selfreport indices” (p. 113). Prosocial Correlates Most research on the correlates of the IRI has been conducted using self-report measures. Nonetheless, we can draw a number of meaningful conclusions, particularly regarding EC, because this is the most commonly used subscale and arguably the most prototypical conception of empathy. People scoring high in EC score higher in shyness and social anxiety but at the same time display less loneliness and fewer negative agentic traits (e.g., boasting, verbal aggression; Davis, 1983c). They are also slightly more emotionally reactive (e.g., feeling a sense of emotional vulnerability; Davis, 1983c) yet higher in self-control (Tangney, Baumeister, & Boone, 2004). Taken together, those scoring high in EC appear to be a little bit nervous around other people, but they care about being liked, and perhaps they use their selfcontrol to defer their own gratification in lieu of others'.
Importantly, the emotional sensitivity and self-control a**ociated with high EC translates into more prosocial attitudes and behaviors. For example, Taylor and Signal (2005) found that higher EC scores are strongly correlated with more positive views about animals as well as self-reported vegetarian practices. However, the prosocial consequences of a high EC extend beyond the treatment of nonhuman animals: Participants who score higher on the EC subscale indicate more continuous volunteer hours per month (Unger & Thumuluri, 1997), choose to participate in experiments that will knowingly evoke feelings of sympathy and compa**ion (Smith, 1992), and are more likely to have returned incorrect change, let somebody ahead of them in line, carried a stranger's belongings, given money to a homeless person, looked after a friend's plant or pet, and donated to charity within the preceding 12 months (Wilhelm & Bekkers, 2010). In one of the original demonstrations of the prosocial behaviors a**ociated with high EC scorers, Davis (1983b) found a strong correlation between those participants and the likelihood of both watching and contributing time and money to the annual Jerry Lewis muscular dystrophy telethon.
In addition, these effects appear to be specific to the EC subscale. Davis (1983a) asked participants to listen to an audiotape of someone in need and then indicate the number of hours they were willing to volunteer to help the person. For half of the participants, the appeal for help was structured in emotional terms; for the other half, it was a cognitive appeal. Davis found that the participants with a higher EC score indicated a greater number of volunteer hours, despite the type of appeal and their scores on other IRI subscales. Thus, dispositional empathy, and particularly EC, is a better predictor of a self-reported desire to volunteer than other empathic subscales or situational factors. Similarly, Davis et al. (1999) found that high EC scores were strongly correlated with people's initial willingness to be involved in situations that might require volunteerism and that this link was mediated by the expectation that the participants would feel sympathy and other positive emotions. These trends were based on the responses of a college population as well as community adults who volunteered at a local governmentfunded volunteering agency over a 5-year period.
Although limited to mostly self-report correlations, a few studies demonstrate behavioral prosocial implications of the IRI. By far the most salient behavioral correlate within the literature is volunteerism. For instance, volunteers for crisis and intervention help lines had significantly higher EC and PT scores than a matched nonvolunteer control group (Paterson, Reniers, & Vollm, 2009). Oswald (2003) further found that high scores on the PT subscale could reliably predict who would volunteer time to counsel working adults who were considering returning to college. Supplementing these findings, Litvack-Miller, McDougall, and Romney (1997) demonstrated the link between the IRI and volunteerism. Canadian grade-school children read a series of vignettes about people in need and indicated how they would respond in each situation. The children were then presented with a video of an actual family in need and were asked to indicate how much time and money they would donate. In both cases, the children with higher EC scores, and to a lesser extent higher scores on the PT subscale, were most likely to indicate prosocial responses and donate the most resources.
PT is also related to prosocial outcomes. For example, it is a**ociated with low social dysfunction (e.g., shyness, loneliness, social anxiety, boasting, verbal aggression) and more other-oriented sensitivity (Davis, 1983c). Those who score high in PT are better able to match target individuals with their self-descriptions, which is a cognitive type of task (Bernstein & Davis, 1982). Intrapersonally, PT is a**ociated with higher self-esteem and lower self-reported anxiety (Davis, 1983a). People scoring high in PT help others when they are reminded to take other people's perspectives but not necessarily in other situations (Davis, 1983a).
Those scoring high in FS are more emotionally vulnerable and more sensitive to other people's perceptions of them (e.g., public self-consciousness, other directedness; Davis, 1983c). Given that the FS subscale measures people's ability to be imaginatively transported by fictional material, it is not surprising that it does not appear to be related to prosocial behavior (e.g., Davis, 1983b, 1983c). Similarly, one would not expect to find a relationship between PD and prosocial behavior because this subscale is a**ociated with higher social dysfunction (e.g., shyness, social anxiety, introversion), lower self-esteem, and a greater concern with what others think about the self (Davis, 1983c). In short, PD involves more self-oriented than other-focused reactions to other people's distress. Antisocial Correlates Much of the remaining literature on the IRI highlights the negative, antisocial consequences of those who score low in empathy. Having empathy is an important factor in the motivation and ability to inhibit harmful behaviors because imagining the potential harm that one might cause deters antisocial behaviors. Studies on the antisocial characteristics of people with low empathy typically focus on a specific sample or subgroup. For instance, bullying within youth populations is negatively correlated with IRI (Ireland, 1999), and actively helping a victimized schoolmate as measured via self-report is positively correlated with IRI (Gini, Albiero, Benelli, & Altoè, 2007). In addition, low IRI scores are linked to aggressive behavior among the inebriated (Giancola, 2003), those who have committed s**ual offenses (Burke, 2001), and those who have been accused of child abuse (Wiehe, 2003). A recent meta-an*lysis found that criminal offenders scored lower on PT than nonoffenders (Jolliffe & Farrington, 2004). However, no relationship was found between EC and criminal offending (Jolliffe & Farrington, 2004). On the whole, the correlation between low empathy and violent behavior is so strong that Bova**o, Alterman, Cacciola, and Rutherford (2002) strikingly concluded that “violent crime may be predicted by traits, such as empathy . . . over and above the a**essment of prior antisocial behavior” (p. 371).