Limitations Limitations. One of the limitations of an*lyzing self-report data is that they might be influenced by people's tendencies to respond in a socially desirable fashion. However, although the EC, PT, and PD subscales of the IRI have been shown to be related to social desirability (Watson & Morris, 1991), social desirability has not changed during a similar time period as this study (Twenge & Im, 2007). This makes it unlikely that our results can be accounted for by changes in socially desirable responding over time. This study also limits its conclusions to American society because there are not much data available over time from other countries. There is also not much work examining cross-cultural similarities and differences in empathy. Future work might examine whether empathy has also been declining in other countries or whether these changes are only occurring in the United States, a finding that would help clarify some of our speculation about causes of decreasing empathy. For example, if one cause involves changing media consumption, we could compare empathy scores between countries that have relatively more or less media consumption of various kinds. The data are also limited to college student populations, and future research might examine shifts in empathy in other populations.4 However, the IRI is commonly given to college students, and their relative h*mogeneity over time is precisely why they are a good population in which to examine temporal trends. Some noncollege populations might not be as comparable over time (e.g., community samples, clinical samples). This study also cannot determine whether the changes we found in PT and EC are a cohort effect or a time-period effect. Any time-lag study that includes people of only one age group does not allow researchers to determine if other age groups also changed in a given characteristic. It is possible that both younger and older Americans became less empathic from the late 1970s to 2009. It is also possible that older Americans did not change at all or even became more empathic over time. However, given the relative stability of empathy (e.g., Davis & Franzoi, 1991; Eisenberg et al., 2002), much of the shift is probably a generational rather than a time-period effect. Conclusion We dare not conclude by suggesting that empathy is declining and nothing can be done about it. Just as we speculate that certain situations lead empathy to decrease, other situations that can increase people's empathy. One promising intervention, the Roots of Empathy, has been successfully implemented in elementary schools by teaching children empathy through multiple structured interactions with a developing infant from their community. This work has found decreases in aggressive behavior and increases in prosocial behavior such as sharing and helping in children randomized to the treatment group compared to those in the control group (see Gordon, 2003). Other experimental work also finds that empathy is teachable in children and young adults (Feshbach, 1983; Feshbach & Cohen, 1988; Hatcher et al., 1994) through a variety of methods. So although there has been no meta-an*lytic work specifying which elements of empathy training are effective in changing particular behaviors in specific groups of people, initial work suggests that declines in empathy appear to be changeable. We recommend more work on examining potential causes, consequences, and remedies of increases in self-focus (e.g., Twenge et al., 2008) and decreases in other-focus. For example, if technology and social networking are indeed significant contributors to empathy decline, perhaps a simple intervention could be to spend 20 or 30 minutes each day personally interacting with family and friends while (emotionally and cognitively) taking their perspective. To summarize, the present research examined changes in empathy over time, based on speculation that related trends and correlates (e.g., increasing narcissism and individualism) reflect a diminishingly empathic society. We found that dispositional empathy—as measured by the IRI (Davis, 1980a), a widely used and validated measure of the trait— declined over time among American college students, particularly on the EC and PT subscales and since 2000. This finding is troubling, as dispositional empathy is linked with higher prosociality and lower antisociality, but it opens the door for research on the causes and consequences of living in a potentially less empathic society.