Egos Inflating Over Time: A Cross-Temporal Meta-Analysis of the Narcissistic Personality Inventory Jean M. Twenge,1 Sara Konrath,2 Joshua D. Foster,3 W. Keith Campbell,4 and Brad J. Bushman2 1San Diego State University 2University of Michigan 3University of South Alabama 4University of Georgia ABSTRACT A cross-temporal meta-analysis found that narcissism levels have risen over the generations in 85 samples of American college students who completed the 40-item forced-choice Narcissistic Personal- ity Inventory (NPI) between 1979 and 2006 (total n 5 16,475). Mean narcissism scores were significantly correlated with year of data collection when weighted by sample size (b 5.53, po.001). Since 1982, NPI scores have increased 0.33 standard deviation. Thus, almost two-thirds of recent college students are above the mean 1979���1985 narcissism score, a 30% increase. The results complement previous studies finding increases in other individualistic traits such as assertiveness, agency, self-esteem, and extraversion. It is common for older people to complain about ������kids these days,������ describing the younger generation as self-centered, entitled, arro- gant, and/or disrespectful. As a bromide set in a particular time, it is difficult to tell whether these perceptions are a function of age (may- be younger people are more self-centered than older people simply because they are young) or of generation (maybe the younger gen- eration actually is more self-centered than the older generation was at the same age). It is also possible that older people will complain about the younger generation even if young people are actually less self-centered than they were when they were young themselves. Correspondence concerning this article should be addressed to Jean M. Twenge, Department of Psychology, San Diego State University, 5500 Campanile Drive, San Diego, CA 92182-4611 E-mail: jtwenge@mail.sdsu.edu. Journal of Personality 76:4, August 2008 r 2008, Copyright the Authors Journal compilation r 2008, Blackwell Publishing, Inc. DOI: 10.1111/j.1467-6494.2008.00507.x

To study generational change scientifically, it is necessary to sep- arate the effects of generation from age and to measure traits using psychometrically sound questionnaires. This is best accomplished through the time-lag method, which analyzes samples of people of the same age at different points in time. For example, college stu- dents from the 1980s can be compared with college students from the 1990s and 2000s. All samples are of the same age, but are from different generations (otherwise known as birth cohorts). Birth co- hort is a useful proxy for the sociocultural environment of different time periods (Stewart & Healy, 1989 Twenge, 2000). For example, children growing up in the 1970s were exposed to a fundamentally different culture than children growing up in the 1990s. The logic underlying this approach is similar to that used to assess the self- conceptions and personality traits of individuals across different world regions (e.g., Choi, Nisbett, & Norenzayan, 1999 Heine & Lehman, 1997 Markus & Kitayama, 1991), except that individual differences between birth cohorts (instead of cultural groups) are assessed. In support of this idea, several previous studies have found strong birth cohort differences in characteristics such as anxiety, self- esteem, locus of control, and sexual behavior (Twenge, 2000 Twenge & Campbell, 2001 Twenge, Zhang, & Im, 2004 Wells & Twenge, 2005, respectively). These studies used meta-analysis to locate sam- ples of college students and children who completed the same psy- chological questionnaires at different points in historical time. The correlation between mean scores and the year the data were collected were then analyzed, using a method known as cross-temporal meta- analysis (e.g., Twenge, 2000). The present study uses cross-temporal meta-analysis to examine changes in scores on the Narcissistic Personality Inventory, or NPI (Raskin & Hall, 1979, 1981 Raskin & Terry, 1988). The NPI is the most widely used measure of narcissistic personality in the general population. The NPI is not designed as a clinical instrument for measuring narcissistic personality disorder (NPD), and there is no cut-off score for clinically high narcissism (Foster & Campbell, 2007). Narcissism is characterized first and foremost by a positive and inflated view of the self, especially on agentic traits (e.g., power, importance, physical attractiveness: e.g., Campbell, Rudich, & Sedikides, 2002 John & Robins, 1994). Second, narcissism is asso- ciated with social extraversion, although people high in narcissism have relatively little interest in forming warm, emotionally intimate 876 Twenge, Konrath, Foster, et al.

bonds with others (e.g., Campbell, 1999 Carroll, 1987). Third, nar- cissism involves a wide range of self-regulation efforts aimed at enhancing the self. These efforts can range from attention seeking (Buss & Chiodo, 1991) and taking credit from others (e.g., Campbell, Reeder, Sedikides, & Elliot, 2000 Farwell & Wohlwend-Lloyd, 1998 Rhodewalt & Morf, 1995) to seeking high-status romantic partners (Campbell, 1999) and opportunities to achieve public glory (Wallace & Baumeister, 2002). Those high in narcissism also lash out with aggression when they are rejected or insulted (Bushman & Baumeister, 1998 Twenge & Campbell, 2003). Many of these be- haviors can potentially be explained by the link between narcissism and impulsivity (Vazire & Funder, 2006). In a sense, narcissism can be conceptualized as a self-regulating system, where self-esteem and enhancement are sought through a variety of social means but with little regard for the consequences borne by others (for reviews, see Campbell, Brunell, & Finkel, 2006 Morf & Rhodewalt, 2001). The NPI is ideal for a cross-temporal meta-analysis assessing changes in narcissism. First, it is reliable, well validated, and widely used. Second, the NPI is somewhat protected from social desirability influences through its use of forced-choice dyads, and, perhaps as a result, is not correlated with measures of social desirability (Watson, Grisham, Trotter, & Biderman, 1984). For each of the 40 forced- choice dyads on the NPI, participants choose either the narcissistic response (e.g., ������I can live my life anyway I want to������) or the non- narcissistic response (e.g., ������People can���t always live their lives in terms of what they want������). The 40 items are summed together. Higher scores indicate higher levels of narcissism. Previous Literature Most previous studies suggest that narcissistic traits should increase with the generations. Several authors have argued that American culture has increasingly emphasized individualism (e.g., Fukuyama, 1999 Seligman, 1990 Twenge, 2006). Perhaps as a result, previous cross-temporal meta-analyses demonstrate a clear rise in individu- alistic traits. Between the 1970s and the 1990s, both college men and women scored higher on the agentic traits measured by the Bem Sex Role Inventory M scale, such as ������independent,������ ������individualistic, particular to me,������ and ������leadership ability������ (Twenge, 1997). College Change in Narcissism 877

women and���on some scales���college men scored higher on asser- tiveness measures between the 1970s and the 1990s (Twenge, 2001b), and both sexes increased in extraversion (Twenge, 2001a). College students scored higher on the Rosenberg Self-Esteem Scale between the 1960s and the 1990s, and children scored higher on the Coopersmith Self-Esteem Inventory between the 1980s and the 1990s (Twenge & Campbell, 2001). Agentic traits, assertiveness, extraver- sion, and self-esteem are all positively correlated with narcissism (e.g., Campbell et al., 2002). A study of changes in personality with age development shows that younger cohorts increase with age more than older cohorts in social dominance but also in agreeableness and conscientiousness over the young adulthood years between 18 and 40 (Roberts, Walton, & Viechtbauer, 2006). However, this meta- analysis examined personality changes with age instead of mean levels by cohort, so it is not clear how the generations differed in mean levels of these traits. Even more directly related to narcissism, an analysis of teenagers��� MMPI responses showed that in the 1950s, only 12% agreed with the statement ������I am an important person.������ By the late 1980s, 80% agreed (Newsom, Archer, Trumbetta, & Gottesman, 2003). From the 1960s to the 1990s, agreement with California Psychological In- ventory items such as ������I have often met people who were supposed to be experts who were no better than I������ ������I would be willing to describe myself as a pretty ���strong��� personality������ and ������I have a nat- ural talent for influencing people������ (also an NPI item) increased (Gough, 1991 cited in Roberts & Helson, 1997). In addition, a large (n 5 3,445) cross-sectional study of NPI re- sponses found that younger people were more narcissistic than older people, with a significant negative correlation between NPI scores and age (Foster, Campbell, & Twenge, 2003). This difference could reflect developmental changes in narcissism with age, generational shifts in narcissism, or both. A time-lag study like the one we un- dertake here is necessary to determine if NPI scores have increased (or decreased) over the generations. Although most evidence points to increases in narcissism over the generations, an alternative model suggests a decrease in narcissism. Generational theorists Howe and Strauss (1993, 2000 Strauss & Howe, 1991) describe Baby Boomers (in college early 1960s to early 1980s) as inner fixated and self-absorbed they specifically use the word ������narcissistic������ in their description (Strauss & Howe, 1991, pp. 878 Twenge, Konrath, Foster, et al.

56���57, 79, 302). In contrast, they portray Generation X (in college mid-1980s to late-1990s), as ������lacking ego strength������ and having ������low self-esteem������ (Howe & Strauss, 1993 Strauss & Howe, 1991, p. 323). Finally, they describe the ������Millennials������ (in college early 2000s to late 2010s, sometimes called ������GenY������) as outer-fixated, group-oriented, and civically responsible. ������Are they self-absorbed? No. They���re co- operative team players,������ say Howe and Strauss (2000, p. 8). They continue, ������Individualism and the search for inner fulfillment are all the rage for many Boomer adults, but less so for their kids, [who are] not as eager to grow up putting self ahead of community the way their parents did������ (p. 237). However, these descriptions are not based on empirical data collection. Although Strauss and Howe���s portrayal of generations includes many traits that are not related to narcissism, the descriptions above suggest that Baby Boomers should be the highest in narcissism, GenXers the lowest, and ������Millennials������ either just as low or even lower (as Strauss and Howe specifically say that they are not self-absorbed). Thus, their characterization of generations sug- gests that narcissism decreased among college students between the 1980s and the 2000s, or, at the very least, stayed steady after the Baby Boomers left college in the mid-1980s. Overview This article presents a cross-temporal meta-analysis of American college students��� responses to the 40-item, forced-choice version of the NPI. This analysis will examine the correlation between NPI mean scores and the year the data were collected, showing how nar- cissism levels have changed over the generations. The issue of changing college populations is an important concern for studies that examine college student samples across time. How- ever, college populations have not changed as much as one might think. Socioeconomic status has not changed: The median income of college students��� parents, when adjusted for inflation, did not vary by more than $3,000 between 1985 and 2004 (U.S. Bureau of the Cen- sus, 2006). The racial composition of college student samples has differed only slightly over this time period. Black students earned 6% of bachelor���s degrees in 1985 and now earn about 9% Asians increased from 3% to 7% and Hispanics increased from 3% to 7%. Although these represent significant improvements for these specific racial groups, these shifts do not dramatically change the racial Change in Narcissism 879