ELITES REVISITED: AMERICAN SOCIAL AND POLITICAL LEADERSHIP IN THE

International Journal of Public Opinion Research Vol. n No. 2 0954-2892/99 S300 ELITES REVISITED: AMERICAN SOCIAL AND POLITICAL LEADERSHIP IN THE 1990s Stanley Rothman and Amy Black ABSTRACT Although the work of leaders in business, law, government, and media profoundly impacts American society and culture, few surveys examine the attitudes and beliefs of these men and women in positions of greatest influence. Reporting data from a new study of elites in the United States, this paper compares recent findings with those from a national study in the s to examine both the continuity and the change in elite demographics and political opinions over the past decades. Although women and minorities are represented in greater numbers in the late 1990s than in previous decades, white men continue to dominate the major leadership groups in the United States. The data, however, reveal that American leaders do not constitute a unified, monolithic 'overclass'. Our results reveal that elites from various leadership sectors come from different backgrounds and differ sharply on a variety of issues, revealing patterns of polarization and differentiation between various elite groups. Cultural elites remain well to the left of more traditional leadership groups on most issues. ELITE THEORY AND REALITY Arguments about the structure of power in the United States have a long history and range from Marxist and Marxist influenced populist interpretations such as those of C. Wright Mills (1956) and William Domhoff (1983) through various 'new class' interpretations fathered by Irving Kristol and supported by many neo-conservatives (see, e.g., Kristol 1978) to pluralist theories of the kind advanced by Robert Dahl (1961), Arnold Rose (1967), Suzanne Keller (1991) and Eva Etzioni-Halevy (1993) among many others. Despite their differing ideological perspectives, Mills, Domhoff, and Kristol share a similar perception of American politics as dominated by one or two unified elites dedicated to maintaining or increasing their power. To Mills it is the military-industrial complex, to Domhoff a ruling class, and to Kristol a new class competing to seize power from the traditional bourgeoisie. More recently Michael Lind () has popularized the notion of an 'overclass' consisting of affluent white males World Association for Public Opinion Research lggg

170 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH who, despite minor disagreements, share the common values which enable them to dominate the society. Lind's writing resembles, to some extent, that of some radical feminists who see domination as located at the intersection of race, class and gender (Enloe 1993, Hartmann 1992, Brown ). According to this view, America is ruled by an affluent white patriarchy. Various pluralist theorists, on the other hand (Dahl 1961, Lerner et al. 1996, Verba and Orren 1985) see power as relatively dispersed (though not equally) among a number of leadership groups which strive to increase their power and to implement policies which they support as right, just, or simply advantageous to them. In the United States, such elites consist of high-level business leaders, journalists, political figures, and bureaucrats, among a number of other groups, who, while sharing some values, also disagree about many issues.' Given the extensive debate about these matters, scholars have completed surprisingly few empirical studies of elites. In fact, only two major surveys have been conducted in recent decades, those by Sidney Verba (Verba and Orren 1985, Verba et al. 1987) and by Lerner, Nagai and Rothman (i996). 2 The groups chosen in each of these studies differ somewhat, and the purpose of the research also varies, but, overall, one finding remains constant. Although consisting primarily (but not entirely) of relatively affluent white males, the views of the members of various leadership groups in the United States in the early s differed sharply in a number of key areas. The theory underlying the project directed by Rothman (Lerner el al. 1996) was derived from the pluralist tradition, supplemented by the work of such scholars as Max Weber (1930) and, more recently, Daniel Bell (1973, 1978). Weber discussed the general radicalism of intellectuals in the early twentieth century, and Bell applied these insights (in a modified fashion) to the United States, arguing that the particular bourgeois Protestant culture of the United States was eroding in the face of affluence and religious decline. Rothman, then, expected to find sharp differences among white male members of various elites on social, cultural, and political issues. When comparing the opinions of members of different elite groups, businessmen would fall on one end of various dimensions of ideology, and journalists and motion picture and television producers (in a broad sense cultural elites) would group toward the opposite end of these scales. By and large, the data supported these hypotheses. The findings show little ' For our definition of elites see Lerner, Nagai and Rothman (1996) chapter one, which also contains a more detailed historical analysis of the development of elite theory. 2 Domhoffand others rely on samples drawn from various volumes such as Who's Who and the biographies therein. Thomas Dye (1976, 1983, 1990) supplements such data by making use of published articles by Rothman and his colleagues, as well as the work of Verba and Orren (1985). The biographical data Domhoff uses tell very little about the attitudes of members of various elite groups, and even lack some relevant demographic material.

AMERICAN LEADERSHIP IN THE 1990S I71 evidence of an overarching overclass or dominant elite sharing common background and outlooks, although no amount of data can resolve the ongoing debate over such issues. This essay, however, tests aspects of these contending theories of power with data from the 1990s. 3 We developed a number of hypotheses predicting in which ways the backgrounds and values of elites might have changed in the intervening 15 years since the last empirical analysis of leadership groups in America. Specifically, we expected to find that the proportion of women and African Americans in elite groups has increased. Given their relative decline in numbers and the growth of intellectual and cultural concerns among other segments of the population, we also expected to find a relative reduction in the disproportionate number of Americans of Jewish background who are members of elite groups as compared to the elite survey. We also assumed that the collapse of the Soviet Union and other Communist states has reduced the hostility of cultural elites to capitalist markets since the s. On purely economic issues, therefore, we expected to find something of a convergence of attitudes with regard to the economic system and economic policy. However, we also hypothesized that cultural elites would be just as suspicious of the power of businessmen as they were in the early s. Finally, we also expected to find a convergence in attitudes of the various elite groups with regard to social and cultural issues. In this area, however, it was our view that the factors described by Daniel Bell were still operative and that the evidence would point to a continued erosion of traditional Protestant bourgeois values. 4 Our contention is, therefore, that the adversary culture had, for whatever reason, emerged victorious in the so-called culture wars. As we shall see, the results support some of these hypotheses. 3 iviost of the elite (neo Marxist) theorists of the past saw America as dominated by an elite united by mutual interest as well as class background. While other issues can be raised, our data suggest that businessmen and women today are not a homogenous group from almost exclusively traditional WASP backgrounds, and that corporate lawyers are not simply blind supporters of business. Further, our data show a business elite composed of individuals with varied social and political views. Some neo Marxists today would point to the increased popularity of the market as evidence of the power of capitalists. However, the simpler hypotheses would accord a very large role to the failure of the Soviet Union and other socialist states. We do not examine various feminist theories of power stratification in this essay. We plan to do so in future essays. Other attempts to 'prove' the existence of a capitalist ruling class by outcomes (especially economic outcomes), also lead one to antinomies which do not readily fit in with power elite theories. After all, the highest per capita income by race in the U.S. is that of Japanese Americans and, by ethnicity, Americans of Jewish background. Do these two groups, then, compose our power elite or ruling class? (Lipset and Rabb,, pp. 26, 36, 80, 82; Hacker, 1997, p. 161). 4 We developed a number of additional hypotheses. To list them all here would make for a very top-heavy article. For details see Lerner el al. (1996) and Rothman (1996). The decision to collect certain kinds of data, however, was not solely determined by our hypotheses. In some cases we simply believed that the collection of such data about important groups has intinsic interest. For example it is useful and important to examine the changing backgrounds of various significant leadership groups even if one has not developed hypotheses about these changes.

172 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH THE ELITE DATA Between 1979 and 1984, Lerner, Nagai, and Rothman directed a personal interview study of 1,861 members of strategic elite groups in the United States, the results of which are reported in American Elites (Lerner et al, 1996). In, we replicated the study in computer-assisted telephone interviews with members of nine strategic elite groups. The following analysis and tables examine the points of intersection between the two studies, comparing only replicated or similar questions for the seven elite groups included in both studies. For the purposes of simplification, all data reported herein will refer to the American Elites study as the study and the more recent study as the study. 5 DIFFERENCES IN SAMPLE DESIGN The sampling frames from the two studies do differ in some important ways. Most significantly, the format of the two studies demanded different sampling techniques. Because the investigators administered the Thematic Apperception Test to respondents, the study had to rely upon personal interviews of the respondents conducted by a geographically dispersed group of survey organizations and coordinated by Response Analysis of Princeton, New Jersey. To facilitate the ease of interviewing the subjects, the researchers geographically stratified the samples. By, the enormous cost of personal interviewing precluded us from conducting in-person interviews in the new study. Instead, the study was a computer-assisted telephone (CATI) design conducted by the National Opinion Research Center, so the researchers did not face the constraint of geographic limitations. Thus, all of the samples are national in scope, designed to give all members of a particular elite population equal chance of inclusion in the random sample regardless of geography. By its very nature, sampling elite populations is a difficult task requiring many subjective decisions. The most important decision was defining the population from which to draw the samples. One of the seven groups, federal s In order to measure the continuity and change in elite attitudes and opinions over time, we will continually compare our data with the results of the earlier survey project published in the book American Elites (Lerner et at., 1996). The data for the study we reference are available through the Roper Center. The data from the Study of Leadership Characteristics will be archived at the Roper Center within the next two years. For a detailed discussion of the sampling procedures in the earlier study see Lerner et al. (1996), pp. 141 4. We do not pretend to have sampled all possible elite groups. However, only professional ideologues, we believe, would deny the importance of these groups in American social and political life. Our sample is very similar to that of Verba and Orren, although we conducted the first ever significant survey of the producers, directors, and writers of prime time television and motion pictures. On the other hand, we did not include leaders of the major political parties. If we had the time and funds, we would have included them and a number of other groups as well.

AMERICAN LEADERSHIP IN THE 1990S I73 judges, has a clearly defined population all individuals with lifetime appointments to the federal judiciary, including judges in federal district courts, circuit courts of appeals, and the Supreme Court. For many of the other samples, however, the principal investigators had to define the population of each elite group. In most instances, an industry-recognized publication provided constructive criteria for defining which individuals constitute the 'elite' of a given sector. Publications such as Fortune, American Lawyer, and Variety served as sources for identifying the population of industry-wide leaders, from which we then drew random samples. In other instances, we relied upon measures such as market domination, job rankings, and published directories to define and draw our samples. 6 We believe that we exercised sufficient care in defining our elite populations, fulfilling two primary criteria: applying a consistent definition of strategic elites across leadership groups and providing methodology to facilitate replication of our sampling frames. We also believe that our sampling techniques are sufficiently rigorous to allow for useful generalizations about the attitudes, opinions, and composition of the leadership sectors we studied. The sample of business leaders surveyed in differs more substantially from the sample survey than that of any other of the groups studied. The samples of corporate lawyers, on the other hand, differ only slightly, and the journalist, bureaucratic, Hollywood, and judicial samples followed almost identical sampling designs in both studies. Comparative data from the two studies do not indicate that either the slight sampling differences or the variation in interviewing techniques significantly affected our most important findings. As we shall see, in both samples business leaders are the most conservative group, while journalists and the Hollywood elite, though differing slightly from each other, are the most liberal groups in the sample. Further, our respondents provide responses markedly similar to the results of Verba and Orren's (1985) elite study. Although one may wish to exercise some caution in making definitive statements about change in elite opinions over time, the data from the two elite studies are useful for comparing the structure and nature of American leadership groups in the past decades. WHO ARE THEY? A PORTRAIT OF LEADERSHIP IN THE UNITED STATES What does the comparison of elite studies over time reveal about the sociodemographics, attitudes, and opinions of those men and women in positions of 6 For more specific descriptions of the sampling techniques and sample sizes for the and studies, see Appendix A.

174 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH TABLE I Selected demographic characteristics and samples Year White Race Black Other Percent Male Mean Age Bureaucrats Business Judges Lawyers Media Religion TV/Movie 94 92 97 98 90 87 99 95 95 89 87 99 95 Source: Study of Leadership Characteristics ; Lerner et al. 1996. greatest influence over American society? In what ways have elite opinions changed over the past 20 years and in what ways have they stayed the same? The following analysis compares and contrasts the results of the and national surveys of elites to provide a portrait of leadership in the United States over the past decades. RACE AND GENDER As in the study, American leadership groups today still consist predominantly of white males. Table 1 compares the racial, gender, and age composition of the elite groups in the two studies. As the data indicate, the representation of blacks has increased among Hollywood elites, journalists, and lawyers but has declined as a percentage of judges, bureaucrats, religious leaders, and top business leaders. Although the percentages only range between 1 and 6 percent, the representation of other minority groups increased in every sample group except business leaders. Given the very small numbers of respondents of color in both surveys, it is unwise to make many generalizations about the nature of changes in racial composition over time. We can say with certainty, however, that the number of minorities in various leadership groups, while on the rise, remains quite small. 6 5 2 11 1 7 1 3 5 11 6 1 2 1 4 2 1 6 1 3 2 6 2 3 1 3 93 82 93 94 95 89 98 88 79 64 85 85 95 87 49 52 45 53 61 64 49 47 43 45 53 56 49 47

AMERICAN LEADERSHIP IN THE 1990S 175 Women are another matter. Although their numbers have not increased among business or religious elites, the representation of women has risen sharply in the highest levels of the bureaucracy, the legal profession, and the media; sectors where women were already beginning to make progress in the early s. Today, over a third of top-level journalists are women. Not surprisingly, women elites are younger than their male colleagues. The results of independent sample ^-tests reveal that women bureaucrats, business leaders, judges, and journalists are all significantly younger than their male counterparts. 7 Although female lawyers, religious leaders, and Hollywood elites are, on average, younger than their male colleagues, the differences are not statistically significant. The mean age difference is greatest among judges and business leaders, among whom women are an average of eleven and nine years younger than the men, respectively. The age difference reflects, one suspects, the newness of their arrival within the federal judiciary and among the high ranks of the business sector as compared to other groups. RELIGION Comparing the religious preference and background of the elites in our studies, we discover interesting patterns of variation among the leadership groups. Large majorities of bureaucrats, business leaders, and judges in both samples are Protestant or Catholic, reflecting a pattern similar to that in the population as a whole. Different patterns emerge for other elite groups. In both the and samples, a plurality of Hollywood elites are Jewish (41 percent and 37 percent, respectively), with about another third (35 percent in and 32 percent in ) claiming no religious affiliation. Compared to the data, elites in the sample are more likely to be Protestant and are more likely to claim a religious preference 'other' than the offered categories. Protestants are most heavily represented in the religious, business, and judicial samples in both the and studies; they are least represented in the television/movie sample. The distribution of Catholics is quite different between the two studies. The percentage of Catholics included in the sample of religious leaders is 30 points higher than that in, 8 and the percentage of Catholic lawyers more than doubled between the two samples, increasing from 11 percent in to 24 percent in. The percentage of Catholics also rose significantly among journalists, increasing from 12 percent in the earlier sample to 19 percent in the new study. 7 Due to space limitations, we do not report the results here. The data are available upon request from the first author. 8 This probably reflects the different sampling techniques of the two studies.

176 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH The proportion of Americans of Jewish background has significantly declined in the film industry and among judges, lawyers, and bureaucrats. During the same time frame, in contrast, the number of Jews increased slightly in the business world (from 6 percent to 9 percent) and in the elite news media (from 14 percent to 19 percent). In all of these areas, however, Jews are still significantly over-represented in elite positions. Comprising slightly more than 2 percent of the population, Americans of Jewish background in the study constitute 37 percent of the television and motion picture elites, 22 percent of elite lawyers, and 19 percent of the media elite. In general, the proportion of elites in all sectors who claim no religious preference declined between and, with the largest decline among elite journalists. Journalists denying any religious preference dropped from 50 percent in to 22 percent in. We found this last finding somewhat surprising, though anecdotal evidence does support it (Frum 1996). In addition to questions about religious affiliation, both the and studies ask respondents the frequency of their church attendance. Respondents in the sample are somewhat more likely to attend religious services than their counterparts, but the order of the groups remains much the same. As in the study, members of the media and the television and movie elite are the least likely to attend religious services, while business leaders and judges report the highest levels of church attendance (Table 2). Most groups, however, register at least some change toward greater participation in formal religious activities. The largest change occurs among judges. Six of ten judges in the earlier study attended religious services a few times a year or less. By, a similar percentage (62 percent) report attending services at least once or twice a month. Conversely, the percentage of businessmen and women who never attend church rose from 13 percent to almost two out of ten. While journalists are much more likely than business leaders to report that they never attend church, the number so reporting dropped almost 10 percent between the and samples, decreasing from almost 50 percent to under 40. Creators of television programs and motion pictures remain the least religious group in the sample, with the proportion of those never attending religious services dropping from almost 53 percent to a little more than four out often. Our data suggest that observers who perceive an increase in religiosity among the Hollywood elite seem to be right. However, the shift is quite modest. EDUCATIONAL BACKGROUND Michael Lind's assertion () that key positions in America are dominated by an overclass composed of a homogenous, Ivy League educated elite is simply wrong. Our study rated colleges and universities on a scale based on the

AMERICAN LEADERSHIP IN THE IO.O.OS 177 TABLE 2 Attendance at religious services and Year Every week 1 2 times a month a year Never Bureaucrats Business Judges Lawyers Media Religion TV/Movie 25 26 35 25 17 33 15 17 9 11 92 4 9 12 21 15 30 20 29 16 26 Source: Study of Leadership Characteristics ; Lerner et at. 1996. work of Ladd and Lipset (1975). Their conception of elite institutions consisted of those which ranked highest on a seven point selectiveness scale, a category of colleges and universities broader than just the Ivy League schools. Nevertheless, a majority (in this case, 68 percent) of only one group in the sample, lawyers, graduated from elite institutions. The percentage of other elites who graduated from the most selective institutions ranged from a high of 47 percent for judges to a low of 19 percent for religious leaders. Between one-fifth and one-quarter of the respondents in most groups in the sample graduated from elite colleges (Lerner et al. 1996, p. 20). We also tested Lind's statements against data from our study. Contrary to Lind's description of American elites, an average of only 18 percent of the leaders in the sample received their highest academic degree from an Ivy League institution. The percentages of Ivy League graduates were highest among lawyers and journalists and quite low among business leaders, motion picture and television creators, judges, and bureaucrats (Table 3). One group does seem to have benefited from an Ivy League education to an astonishing degree. If the shift in Ivy League admission criteria from that based on 'character', wealth and alumni status to a more meritocratic admission policy assisted anyone from the late 1940s to the mid-1960s, it aided Jewish Americans. 6 19 6 7 2 13 33 22 37 27 19 46 29 37 2 * 36 30 13 19 12 19 24 29 49 39 * 53 43

178 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH TABLE 3 Elites' educational background and i goo Elite schools* J 99S Ivy League** All Elites Bureaucrats Business Judges Lawyers Media Religious Leaders TV/Movie Makers Source: Study of Leadership Characteristics ; Lerner el al. 1996. * Elite Schools are those ranked in the highest category of research universities and liberal arts colleges, as determined by the Carnegie Commission's system of ranking educational prestige. ** Ivy League schools include colleges and universities in the Ivy League and the Seven Sisters. Some 55 percent of Americans of Jewish background who are members of elite groups graduated from an Ivy League college or university. SOCIAL BACKGROUND The sample included several measures of the social background of various leadership groups. Respondents were asked their parents' occupations and educational background to estimate the elites' family income during their childhood. Surprisingly, in the s, over 30 percent of business leaders came from blue-collar backgrounds, as did 31 percent of high level bureaucrats. Lawyers tied with journalists as coming from the most affluent backgrounds in the sample. The parents of a vast majority of respondents in all sample groups did not attend college. While just over 26 percent of the fathers of business leaders graduated from college or received a graduate degree, over four of ten fathers of journalists obtained a bachelor's degree or more. Overall, the proportions of mothers graduating from college were smaller than that of respondents' fathers. The differences in educational attainment are greater for the mothers of businessmen and women than for mothers of journalists; that is, only 15 percent of the former versus over one third of the latter earned at least a college degree. Although the study does not include as many background questions as the study, both surveys include questions about parental educational background. Relying upon education as an imperfect measure of income and social status, we can compare the social background of elites in the two studies. 9 4i 24 25 47 68 37 19 25 18 13 15 18 33 24 8 12 ' We recognize that socio-economic background and parental education do not correlate perfectly, but many studies show a strong relationship between the two variables.

AMERICAN LEADERSHIP IN THE 1990S 179 The data on parental educational background suggest that both lawyers and journalists still come from more affluent backgrounds than top-level business executives. Thirty-six percent of the fathers of business leaders completed four or more years of college, as compared to close to 56 percent of the fathers of lawyers and journalists. In contrast, the fathers of six of ten religious leaders in the sample have a high school diploma or less formal schooling, as do 55 percent of bureaucrats' fathers and 48 percent of judges' fathers. The educational attainment of mothers is comparable. Twenty-four percent of the mothers of business leaders completed college or continued for graduate education as compared with 48 percent of the mothers of the legal elite and 40 percent of the journalists. On the other side of the scale, two of ten mothers of the corporate elite did not complete high school, compared with only 4 percent of the mothers of lawyers and one of ten mothers of journalists. As discovered in the earlier elite study, it appears that business is still a more important source of social mobility in the United States than is the legal profession or journalism. WHAT DO THEY BELIEVE? When Rothman and his colleagues published initial results of the study dealing with elite journalists, many critics were skeptical. In their view, this picture of elite journalists as predominantly liberal and cosmopolitan in their orientation was simply inaccurate and distorted (see, e.g., Gans 1985). By the early 1990s, however, the prevailing view changed. It was true, critics maintained, that the elite media most likely held the political opinions described in the studies in the s, but journalists had changed as the country had changed. By the 1990s, even some relatively conservative critics were arguing that journalists had become more conservative and the so called 'culture wars' had considerably diminished in intensity (Frum 1996, Scott 1997).'" In contrast, Weaver and Wilhoit's (1996) national survey of journalists in what they describe as the 'mainstream news media' reports findings more consistent with our empirical data. In their studies conducted in 1982 and 1992, they find evidence of increased liberalism among members of the journalism profession, lending further support to our hypothesis. The elite study revealed sharp divisions among the various elite groups on a variety of social and political issues. A factor analysis of a series of opinion questions identified three underlying dimensions of the respondents' ideology, two of which are also present in the study. Following the terminology 10 Of course most journalists deny that their ideological perspectives affect the way they describe the world to their readers or viewers; see, e.g., Dennis, 1997.

l8o INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH described in Lerner et al. (1996) we will refer to the two common factors as 'collectivist liberalism', borrowing a term from John Dewey (1930) and 'expressive individualism', borrowing the concept from Robert Bellah et al. (1985). The labels of the factors name the liberal in the modern U.S. sense of the word end of two-dimensional scales. The other end of these name the traditional roots of American society. Max Weber (1930), Louis Hartz (1955), and Seymour Martin Lipset (1996) have all observed that America's uniqueness lies in her religious-cultural ethos. Thus, the traditional American value system is based on the psychological consequences of a Protestant ethic which gave rise to a previously unknown degree of personal autonomy and a remarkable discipline of the passions that enabled the individual to achieve the goals of this world. This ethic also fostered the psychological autonomy and self-restraint that led to the development of a strong but disciplined sense of self, thereby providing the psychological foundation of modern American liberalism and capitalism." The goal, in the secular version of this creed, was the pursuit of happiness, but it was a pursuit limited by strict cultural prohibitions especially in the area of sexuality. In contemporary liberalism the goal is still the pursuit of happiness, but the inhibitions have greatly eroded. The first dimension places 'rugged' laissez-faire individualism, derived from America's Calvinist heritage, on one end and 'collectivist liberalism' on the other. Collectivist (or welfare state) liberalism emerged from the Depression Era (see, e.g., Shils ). Often seen as the major source of contemporary American ideology (see, e.g., Verba and Orren 1985; Feldman 1983), this concept rests on the belief that the government should ameliorate the economic inequalities of the capitalist system. Standing in opposition to this view is the traditional 'rugged individualist' perspective that economic well being stems from individual effort and personal achievement. The second dimension places traditional Puritanism on one end and 'expressive individualism' on the other. The latter is characterized by a shift in the meaning of the individual from a 'being' to a 'self;' a shift from restraint of impulse to its free expression. The contrast of expressive individualism with prior bourgeois culture is stark. The Protestant ethic stressed sobriety over playfulness, restraint over expression, and self-restraint over emotion. Expressive individualism tends to reverse that emphasis. 12 How do American leadership groups compare on these two ideological dimensions today as compared to 10-15 years ago? A direct comparison is not possible primarily because we asked a somewhat different set of questions in 11 For a somewhat more extended discussion of the historical roots and character of the American ethos, see Rothman (1996). 12 Our description of "expressive individualist" characteristics more or less follows Bellah et al. (1985) and Bell (1978).

AMERICAN LEADERSHIP IN THE 1990S TABLE 4 Factor analysis results (Mean Factor Scores) l8l Collecttvist liberalism Expressive individualism Sample Bureaucrats 99-91 99.60 Business 93-19 97-10 Judges 102.15 96.91 Lawyers 99-27 102.77 Media 102.02 105.01 Religion!O5-i3 9079 TV/Movie 101.06 105.59 Sample Bureaucrats 98.22 102.97 Business 90.58 101.47 Judges 98.02 97-2 5 Lawyers 96.84 102.54 Media 104.02 106.29 Religion 104.84 88.04 TV/Movie 106.18 106.52 than those asked in the survey. Despite the changes in questionnaire content, results of the factor analysis of the data reveal two factors that closely parallel the original expressive individualism and collectivist liberalism factors found in the earlier study.' 3 Table 4 compares the standardized factor scores for the groups in both studies.' 4 These similar results suggest that the two underlying ideological dimensions continue to explain some of the coherence of elite belief systems. Not only do the factors include many of the same component questions, the relative positions of the groups remains the same (except for religious leaders) when scores are standardized, re-enforcing our belief that our questions measured the same attitudinal variables in both studies. VOTING PATTERNS Analysis of presidential voting patterns in the and samples provides further support for the contention that the underlying ideological dispositions IJ We modified some of our questions because we believed that new questions would better capture attitudes on certain issues. While we recognized that the change might make comparison more difficult, we believed that it was worth the risk. We were right in most cases. The factor loadings on the new version of the questions we used were much higher than were the older versions and we were able to set a higher cut off point in the later study. As a result, one question from each of the two factors which we listed in is not listed in, though we have included them in appendix B. M For ease of interpretation the factor scores in Table 4 have been standardized at 100 with a standard deviation of IO. We re-estimated the factor scores to include only the seven elite groups common to both the and data sets. It is important to note, however, that comparisons of these two sets of scores are relative to the means and standard deviations of the factor scores for each individual data set, so one must use caution not to compare absolute changes in these scores over time.

182 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH TABLE 5 Presidential voting behavior 1Q72 igj6 ig88 19Q2 JND DEM GOP DEM GOP DEM GOP DEM GOP Percent Percent Percent Percent Bureaucrats Business Judges Lawyers Media Religion TV/Movie 57 22 57 55 81 62 83 43 78 43 45 19 38 17 7i 25 52 56 81 68 76 29 75 49 44 19 32 24 38 11 32 4i 76 40 70 61 88 68 51 20 59 29 Source: Study of Leadership Characteristics,. Lerner el al. 1996. of elites have changed little between the two studies. Table 5 presents the results. Voting patterns remain quite stable across the two studies. Although the exact percentages of votes cast for each party's nominee varies from candidate to candidate, the partisan patterns of voting behavior remain quite constant from election to election and provide further evidence of continued polarization. Journalists and television/movie elites, for example, are the two elite groups most likely to cast Democratic votes across elections and time, just as business leaders and, recently, judges consistently favor Republican candidates. Both samples also provide interesting data about support for third party candidates. Independent candidate John Anderson received between 5 percent and 23 percent of elite votes in, with his strongest support coming from Hollywood elites and bureaucrats. Ross Perot, in contrast, did not receive much support from any of the sampled groups. His support ranged from a high of 8 percent among bureaucrats to lows of only 1 percent of the votes of journalists and judges. INFLUENCE IN SOCIETY In addition to questions about ideology and voting behavior, both surveys asked respondents to rate the influence of different groups in society. Two questions measure perceptions of group power: first, respondents were asked to rate the groups by how much influence they currently have over American life on a scale from one to seven, where one represents 'very little influence' and seven represents 'a great deal of influence'. Secondly, they were asked to go through the list again, rating each group on how much influence they would like each group to have over American life. 65 17 42 59 9i 50 83 28 76 57 39 7 46 13 8 7 1 2 1 4 5

AMERICAN LEADERSHIP IN THE IO.Q.OS 183 We compared the seven elite samples in both studies by their ratings of the current influence and the desired influence of four groups: women, business leaders, the news media, and religious leaders. In both the and studies, respondents placed the media far to the side of the scale indicating they have a 'great deal of influence' over American life. Interestingly, all groups, including members of the media themselves, reported a preference that the media exert less influence. Although none of the elite groups sampled described business leaders as exerting as much influence as the media, respondents consistently rank business leaders as powerful.' 5 Respondents also expressed much agreement in their rankings of the perceived and desired influence of blacks, rating their current influence towards the middle of the influence scale, and all expressing a desire that blacks wield more power in society. Both media and business leaders in and ranked their respective sectors' current level of influence lower than did the other elite groups. Correspondingly, the mean score of desired influence of the news media is highest among the journalists in the samples, and the mean score of desired business influence is highest among the business leaders. The results were somewhat different for religious leaders. In the sample they believed their current influence to be higher than any other sample group rated them, but in the sample the results reversed, as all but one group (lawyers) described religious leaders as wielding more power than religious leaders attributed to themselves. One of the most interesting results when comparing the data from the and studies is the consistency in the patterns of perceived influence. All of the net influence scores for each sample group are in the same direction in each of the two study periods, and the mean influence scores themselves change little. Both of these results confirm our belief that our decision to change slightly the wording of some questions does not invalidate comparisons between the two studies. ELITE OPINIONS ON SPECIFIC ISSUES How do the opinions and beliefs of elites in the two studies compare? Tables 6-8 display responses to a variety of Likert scale questions from the survey that were either exactly replicated or closely replicated in the study, and are drawn from those which loaded on the two factors in both the and studies. In each study, respondents were read a series of statements and 15 These findings also provide support for our basic argument that traditional power elite theories provide too narrow a definition of power. Indeed, every group in our sample ranked the news media as exerting even more influence than business, suggesting that people at the highest levels of various sectors of society perceive the media as quite powerful.

184 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH TABLE 6 Changes in elite opinions economic issues Percentage who agree (percentage who strongly agree) Year Government guarantee jobs' Decrease regulation* Bureaucrats Business Judges Lawyers Media Religion TV/Movie 34(8) 52 (14) 29(6) 34(9) 48 (13) 56 (19) 43 (16) 56 (16) 48 (19) 71 (30) 55 (19) 7 (33) 41 (12) 63 (30) Source: Study of Leadership Characteristics ; Lerner et al. 1996. "The government should work to ensure that everyone has a job. b Less government regulation of business would be good for the country. 61 (14) 63 (15) 86 (57) 94 (71) 67 (26) 70 (32) 72 (26) 72 (25) 62 (16) 40(6) 49 (18) 68 (26) 58 (18) 39(6) asked to tell the interviewer if they strongly agreed, somewhat agreed, somewhat disagreed, or strongly disagreed with the statement. The summary tables present two percentage totals for each elite group the total percentage of respondents who agreed, regardless of strength of agreement, followed by the percentage of respondents who expressed strong agreement. Table 6 includes two exactly replicated questions that probed respondents' opinions on economic issues that measure liberal collectivism. Responses to the question about government job guarantees suggest that elites are more concerned about job guarantees in the later survey, for all seven groups are more likely to agree that government should work to ensure that everyone has a job. In the survey, a majority of every group except business leaders agrees with the job question, with the strongest levels of agreement from the media elites (71 percent) and religious leaders (70 percent). In contrast, a majority of only one group, religious leaders, supported job guarantees in. A second replicated question compares levels of agreement with the statement, 'less government regulation of business would be good for the country'. Responses to this question suggest that support for the market has not increased among cultural elites despite the collapse of the Soviet Union, though it has

AMERICAN LEADERSHIP IN THE 1990S 185 TABLE 7 Changes in elite opinions expressive individualism Percentage who agree (percentage who strongly agree) Year Abortion right to choose' Adultery wrong b Gay lifestyle unacceptable 0 Bureaucrats Business Judges Lawyers Media Religion TV/Movie 81 (68) 88 (71) 81 (62) 86 (62) 80 (64) 79 (53) 90 (73) 83 (63) 90 (79) 97 (84) 38 (21) 4i (25) 97 (92) 96 (87) 67 (35) 87 (53) 74 (45) 94 (61) 66 (35) 87 (61) 62 (21) 87 (55) 47 (15) 78 (45) 89 (73) 92 (83) 46 (14) 77 (41) 48 (29) 56 (19) 59 (35) 65 (35) 66 (34) 66 (36) 40 (19) 49 (19) 24(9) 27(8) 72 (57) 74 (54) 25(7) 21(8) Source: Study of Leadership Characteristics ; Lerner et al. 1996. ' It is a woman's right to decide whether or not to have an abortion. b It is wrong for a married person to have sexual relations with someone other than his or her spouse. c wording: It is wrong for adults of the same sex to have sexual relations. wording: Homosexuality is as acceptable a lifestyle as heterosexuality (coding reversed). among other leadership groups. That is, slightly larger percentages of the traditionally conservative business leaders, judges, and religious leaders support decreasing government regulation, just as the traditionally more liberal cultural elites, journalists, and creators of television and film, show a sharp shift toward more liberal views. The data in Table 7 compare the patterns of responses to the questions which comprise the expressive individualism scale. With the notable exception of religious leaders, overwhelming majorities of all groups support a woman's right to choose abortion. Given the complex ethical questions surrounding the abortion issue and the close association of many pro-life organizations and religious groups, we were not surprised that religious leaders are the least supportive of abortion rights. If anything, the percentages of pro-choice religious elites seem high 41 percent of respondents are pro-choice, with one quarter strongly agreeing with the right to choose abortion. The levels of support are only slightly lower in the study, with 38 percent of religious leaders in agreement, including 21 percent who strongly agree.

l86 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH Among the remaining six elite samples, pro-choice sentiments are very high. Percentages in agreement range from 80 percent to 97 percent in and from 79 percent to 97 percent in the study. On this issue, elite views have not significantly changed. A third replicated question asked if the respondents agree or disagree that adultery is wrong. The data reveal an interesting pattern more members of elite groups believe adultery is wrong in the study than did in. Over three-fourths of all respondents in the new sample consider adultery wrong, with levels of agreement ranging from 77 percent of the Hollywood sample to 94 percent of the business leaders. In the study, in contrast, the range of agreement is much wider, varying from 46 percent of television/movie elites to 89 percent of religious leaders. The elites in the new study are also more likely to strongly agree that adultery is wrong. A majority of all groups in the sample except television/movie and media elites strongly agree; whereas, a majority of only one group in, religious leaders, expresses strong agreement. In both studies, the same two groups, religious leaders and business leaders, are most likely to agree that adultery is wrong. Similarly, the same two groups, television/movie elites and journalists, are the least likely to agree with the statement, although their levels of agreement increased dramatically 31 points between the two studies. These results seem to reflect a national trend in a more conservative direction in this regard: the percentage of respondents to general population surveys who believe that adultery is always wrong rose from about 70 percent in 1973 to about 79 percent in 1994. 16 The results of a final question, one probing respondent's views on homosexuality, are somewhat more difficult to interpret. The two distinct questions reflect the dramatic changes in the arguments surrounding the issue of homosexuality over the past two decades. The question wording, 'It is wrong for adults of the same sex to have sexual relations', focuses specifically on the respondents' opinion of homosexual practices. The study, in contrast, includes a very different statement: 'Homosexuality is as acceptable a lifestyle as heterosexuality'. These dramatically different questions reflect the recent changes in the gay and lesbian movement that have resulted in a strategic shift from trying to gain acceptance for gays and lesbians to promoting policies that give homosexuals the same status and privileges as heterosexuals. The comparison of results between the and studies is most remarkable for the lack of dramatic shifts despite the significant change in question wording. Although there is a substantial psychological distance between 16 The general population data are from national polls conducted by the National Opinion Research Center in 1973 and 1994, as reported in the New York Times (Clines, 1997, p. 5).

AMERICAN LEADERSHIP IN THE 1990S 187 agreeing that homosexuality is not immoral and accepting it as a valid lifestyle, the average change in agreement with the homosexuality question in and within each of the seven leadership sectors is only 4 percentage points.' 7 Almost as many members of elite groups who did not regard homosexual behavior as immoral in now believe that it is an alternative lifestyle that deserves the same status as heterosexuality. We believe that this shift in orientation represents a radical change toward increased acceptance of homosexuality, even though majorities of most groups are still not willing to view homosexuality as a lifestyle as acceptable as heterosexuality. Other studies of public opinion of homosexuality support our contention that public views have begun to shift as activists recast the debate. For example, in his study of middle-class public opinion, Alan Wolfe notes the separation between what he sees as three distinct categories of opinion on homosexuality. Wolfe describes a continuum of opinion that spans from the most conservative view of condemnation on one side to toleration in the middle to positive acceptance on the other end. Although a majority of middle-class Americans choose the moderate position, he notes that 'the distribution of opinion on the ends of the bell curve is not even close to being equal: nearly three times as many respondents condemned homosexuality as accepted it' (Wolfe 1998, p. 76). The question measures opinions on the 'condemnation' side of the continuum, whereas the question effectively equates the acceptability of homosexuality and heterosexuality. That is a movement past the moderate position (tolerance) and towards the perspective of a positive attitude. The final table, Table 8, compares shifts in attitudes on questions of affirmative action in the decade between the two studies. The questions are close enough in wording, we believe, to trace opinion shifts on these issues. Respondents to the survey were asked their opinions concerning special preference for blacks. Our respondents were asked if they agreed or disagreed with a statement patterned after the November 1996 California ballot proposition: 'No one should be given special preference in jobs or college admission on the basis of gender or race'. The data reveal an increase in negative opinions on special preferences in the later sample, disagreement with special preference policies rose an average of 13 points between the and samples, with those expressing strong opinions increasing an average of 15 points.' 8 " As noted earlier, it was our interpretation of changes in elite attitudies and the strategy of homosexxual activists which led us to choose this question to replace the question used in. It was out hypothesis that the battle for equal rights for homosexuals had been largely won in most of the elite communities we were studying, and that the newer question would enable us to develop a more fruitful analysis than we could have by merely repeating the older question. 18 We acknowledge that the changing pattern of responses on both this and the next question may be partly a function of the change in wording. Once again, we were faced with the choice between using the earlier question or a new one, the wording of which we considered to be superior.