The Political Attitudes of Canadian Professors M. Reza Nakhaie University of Windsor and Robert J. Brym University of Toronto Canadian Journal of Sociology 24, 3 (1999): 329-353. An earlier version of this paper was presented at the annual meetings of the Canadian Sociology and Anthropology Association (University of Ottawa: 1 June 1998). Abstract: In this paper we analyze the social determinants of political attitudes among Canadian university professors as expressed in a nationwide 1987 survey. We examine four questionnaire items. They measure attitudes towards faculty unionization, faculty militancy, and faculty salary egalitarianism; and position on a left-right political scale. The results support a class-based theory of intellectual attitudes, more with respect to unionism and leftism than militancy and egalitarianism. Furthermore, a distinct rank-ordering of Canadian professors’ political attitudes by field emerges. Academics in the humanities, social sciences, arts (fine, performing, and applied), and education score consistently to the left of those in the natural sciences, business, and engineering. Field differentiation is consistent across all four measures and persists even after control variables are introduced. We also discover ethno-religious effects (with non-members of dominant ethno-religious groups tending to be more left-wing than others) and gender effects (with women tending to fall to the left of men). Finally, we outline some social mechanisms that result in attitudinal differentiation and discuss the implications of our research for theories of intellectuals and politics in postindustrial societies. Résumé: Dans cet article nous analysons les facteurs sociaux déterminants les attitudes politiques, extraits d’un échantillon important de professeurs canadiens, exprimés dans un sondage national de 1987. Nous examinons quatre questions. Celles-ci mesurent les attitudes envers la syndicalisation des professeurs, leur militantisme, leur attitude envers l’égalisation des salaires; et leur position politique sur une échelle qui va de la gauche à la droite. Les résultats démontrent que leurs attitudes intellectuelles sont basées davantage sur la théorie des classes en ce qui concerne leur attitude envers la syndicalisation et le gauchisme qu’envers le militantisme et l’égalisation des salaires. De plus, il y a une hiérarchie distincte dans leurs attitudes politiques selon la discipline. Les professeurs dans les humanités, les sciences sociales, les arts (les beaux-arts, le théatre et la musique ainsi que les arts appliqués) et l’éducation se trouvent régulièrement à gauche des professeurs dans les sciences naturelles, la gestion et le génie. La différentiation selon les disciplines est uniforme pour les quatre questions et persiste même après que les variables de contrôle ont été introduits. Il y a aussi des différences ethno-religieuses (les professeurs qui n’appartiennent pas aux groupes ethno-religieux dominants ont tendance à être plus à gauche que les autres) ainsi que des différences selon le sexe (les femmes sont plus à gauche que les hommes). Finalement, nous dressons une liste des mécanismes sociaux qui auraient pu produire la différentiation attitudinale et nous discutons les implications de notre recherche pour les théories des intellectuels et leurs attitudes politiques dans les sociétés postmodernes. The Class Location of Intellectuals There are more than 37,000 faculty members regularly engaged in teaching and research in Canadian universities but not a single survey-based sociological analysis of their political attitudes. This imbalance is especially striking given the recent shift in government policy on higher education and the resulting financial constraints imposed on the university system. Since the 1980s, the proportion of GNP spent on Canada’s universities has declined, enrollment has increased, and real expenditure has dropped (Johnson, 1985). One may reasonably expect attitudinal and behavioural ramifications among Canadian professors. Recent events lend credibility to this expectation. Faculty strikes in several universities (Manitoba, York, Trent, and Dalhousie) suggest a possible radicalization of the professoriate in some institutions. Continued resistance to unionization in other universities (Waterloo, Toronto) suggests an entrenchment of historical attitudes elsewhere. Sociological analyses are needed to make sense of the political attitudes of the Canadian professoriate, particularly because it is apparently heterogeneous and in a state of flux. Sociologists have long been interested in the political attitudes, ideology, and behaviour of intellectuals. Often, they have focused on whether intellectuals’ attitudes are class-based or classless. On one side of the debate are the non-Marxists, Karl Mannheim foremost among them. Mannheim argued that modern intellectuals form neither a class nor part of a class. Instead, they are “members of a relatively classless stratum which is not too firmly embedded in the social order” (Mannheim, 1955: 154). Intellectuals are typically recruited from various classes. However, because they participate in a common educational milieu, their class differences, and the variations in outlook normally associated with them, tend to be suppressed. Intellectuals, wrote Mannheim, are “determined in their outlook by this intellectual medium which contains all those contradictory points of view” (Mannheim, 1955: 157). Mannheim’s conclusions are echoed by many more recent scholars. David Caute (1964: 17, 19), a leading student of the European left, writes that “the sociological approach....while of cardinal importance in analysing proletarian or peasant behaviour, is of strictly limited use when applied to intellectuals....The act of [intellectuals’] political affiliation remains one of personal conviction, personal psychology, personal choice.” Shlomo Avineri (1957: 277) contends that, as far as intellectuals are concerned, there “is no a priori determination [of attitudes’]” and that “choice is the very embodiment of the intellectual’s determined social being.” Talcott Parsons (1964: 4) asserts that intellectuals put “cultural considerations before social ones” and Martin Malia (1961: 9), a respected historian of the Russian intelligentsia, that intellectuals place “ideals before interests.” As Everett C. Ladd and Seymour Martin Lipset (1975: 132–133) conclude, the capacity of intellectuals for social criticism, creativity, innovation, and attention to facts enables them to overcome their class socialization — and, for that matter, the attitudinal influence of many of the other groups and communities to which they belong. The opposing viewpoint is that intellectuals are embedded in the class structure and their attitudes do reflect their structural location. This argument is indebted to one or another variant of Marxism. Some early Marxists asserted unequivocally that all intellectuals are members of the proletariat. Others claimed with equal conviction that they are part of the petite bourgeoisie. These nineteenth-century assertions were updated in the late 1970s by American neo-Marxists Barbara and John Ehrenreich (1979), Alvin Gouldner (1979; 1985), and Erik Olin Wright (1979). The middle of the class structure had expanded enormously over the preceding century. Many of the middle ranks were filled by people with university degrees. Reflecting on this change, Gouldner held that the intellectuals constitute a new “emancipatory class.” Similarly, the Ehrenreichs maintained that intellectuals form part of a new “professional-managerial class” whose “objective class interest” lies in the overthrow of the capitalist class, although not in proletarian revolution. Wright, less sanguine about the intellectuals’ revolutionary potential and unwilling to view them as a class or part of a class, argued that they occupy a “contradictory class location” which makes them “junior partners” in the “system of exploitation.” In all three cases, attitudinal heterogeneity is briefly noted, summarily dismissed as an uninteresting side issue, and left inadequately explained; what stands out in the neo-Marxist account is the supposed uniformity of attitudes that derive from intellectuals’ common class location.0 However, the most striking thing about all the assertions briefly reviewed above, neo-Marxist and non-Marxist, is how poorly grounded they are in research. Each theorist depends partly on a few carefully chosen examples, but mostly on sheer speculation, to make his or her case. What makes it possible to argue that class is important or unimportant in shaping intellectuals’ attitudes is a lack of data on the subject. The absence of systematically collected evidence also makes it easier to argue that intellectuals tend to share — or tend not to share — a uniform viewpoint (whether emancipatory or exploitative). There is some non-impressionistic, survey-based research on the subject of intellectuals’ attitudes. What does the North American research show?1 American Research Findings American survey research on intellectuals’ political attitudes seems to lead inexorably to two conclusions. Both appear to support the non-Marxist more than the neo-Marxist position. First, intellectuals’ political attitudes are heterogeneous. For example, Stephen Brint (1984) combined data from a series of General Social Surveys. He showed that American professionals in the social, educational, and cultural fields tend to be the political left while professionals in science and business tend to be on the political right. Second, class explains little of the variation in intellectuals’ political attitudes. For instance, in their analysis of several nationwide surveys, Ladd and Lipset (1975) examined the sources of variation in liberalism, attitude towards civil liberties, and support for faculty unionism among American academics. If the class theory were valid, Ladd and Lipset would have found less liberal opinion, less positive attitudes towards civil liberties, and less support for faculty unionism among academics who received high salaries, were of high rank, published extensively, and worked in elite research-based universities. Such high achievers, it might be suggested, would have much to lose from any significant change either in the structure of the university or the wider society and might plausibly be expected to support the status quo. Class origins (measured by father’s education and/or occupation) would also be related to political attitudes if the class theory were valid: early lower-class socialization experiences would be associated with more left-liberal opinion. However, Ladd and Lipset found the following: Their four measures of current class position explained little or no variation in liberalism. Well-published professors in elite schools were the most liberal of all. Well-published professors of high rank were most likely to be pro-civil libertarian. Professors from lower class origins were not more liberal than those from upper class origins. On the basis of these and related findings, Ladd and Lipset (1975: 87) concluded that applying the “class theory of politics to academe is often ill advised.” More than that: the “academy turns the class theory of politics on its head” insofar as, among intellectuals, high material rewards are associated with a critical bent of mind, not with acceptance of the status quo (Ladd and Lipset, 1975: 148). Ladd and Lipset did note significant effects of religion and discipline on political attitudes. In their samples, minority religious status was associated with left-liberal attitudes. Specifically, Protestants were slightly more conservative than Catholics, and Jews were significantly more liberal than non-Jews. Moreover, professors in the liberal arts, especially the social sciences, tended to be liberal and left-leaning. In contrast, academics in applied and technical-scientific fields were more conservative.2 “Social scientists,” they wrote, “are uniquely drawn by their expertise to cast a critical eye on the social norms and political practices of societies” (Ladd and Lipset, 1975; cf. Ladd and Lipset, 1972a; 1972b; Lipset and Ladd, 1972, 1974; Lipset, 1970: 93–99; 1972). Ladd and Lipset did not look for gender effects, undoubtedly because of the small proportion of women in university faculties 30 years ago and the fact that gender issues were not salient at that time. With respect to faculty unionism, a contradictory finding emerged. Consistent with the class theory, Ladd and Lipset found that lower-status academics — those with fewer publications and financial resources, lower rank, inferior economic benefits, and employment in non-elite universities — were much more likely than others to favour organized collective action (Ladd and Lipset, 1975: 250–264). However, they provided no explanation for this anomaly. Canadian Research Findings There are no survey-based studies of the sources of variation in the political attitudes of Canadian professors. However, survey data concerning the effects of higher education on political attitudes in the general population have an indirect bearing on the subject. Curtis and Lambert (1976) showed that as educational attainment increases, tolerance for agents of social change decreases. Similarly, Baer and Lambert (1982) demonstrated that the higher the level of education, the greater the support for dominant ideology as measured by respondents’ negative responses to such statements as the following: “The gap between rich and poor is too much.” “The rich should pay more taxes than they pay now.” “Government’s responsibility is to provide jobs for the unemployed.” “Government should put much effort into health and medical care.” Although the association between higher education and support for dominant ideology was subsequently replicated (Baer and Lambert 1990; 1995), Guimond, Palmer, and Begin (1989a) argued that there are in fact several subgroups among the highly educated who do not support the dominant ideology. Following Ladd and Lipset, their field socialization model posits that students are relatively undifferentiated in terms of political attitudes early in their academic careers. However, as they move through the system of higher education, ideological differences between fields cause social science students to move to the left. Thus, among the 675 Quebec City-area students they interviewed, field of study had significant effects on political attitudes. Social science students showed more positive attitudes towards unions, socialists, and immigrants, while commerce students showed more positive attitudes towards the military and conservatives. As well, education in social science, regardless of one’s employment status or class origin, was associated with less of a tendency to attribute poverty to personal failure (Guimond, Palmer, and Begin 1989b: 137). These findings added weight to Guimond and Palmer’s earlier six-month longitudinal investigation of a sample of students from an Ontario university. They showed that there were no field differences at the beginning of the academic year. However, by year’s end, engineering students had become significantly more positive, and social science students significantly more negative, towards the military. Baer and Lambert (1990) questioned the generalizability of evidence from Quebec to the rest of Canada and concluded that “respondents who studied business and the professions were more likely to express dominant ideological beliefs than those who studied the social sciences” (Baer and Lambert, 1990: 498). We think this conclusion is either consistent with Guimond, Palmer and Begin’s (1989a) finding or suffers from a technical deficiency. Baer and Lambert found that the regression coefficients for social science graduates did not differ significantly from those of the reference category, people who had not attended university (Baer and Lambert, 1990: 494). “In Table 1, the only positive coefficients for social science graduates appear for social service expenditure and importance of strikers achieving their objectives. In neither case were the coefficients for these items statistically significant.” They continued: “[G]iven the peculiar properties of the social sciences that are said to engender ‘radicalism,’ the failure of the expected pattern of differences among disciplinary areas to emerge must also be regarded as damaging to the strong version of the hypothesis” (Baer and Lambert, 1990: 494). We remind the reader that Baer and Lambert (1990) did not test directly for differences between field. They only tested whether fields are significantly different form the reference category (those who had not attended university). However, despite the failure to find the “expected pattern of differences” and the lack of an appropriate direct statistical test of differences between the social sciences and other fields, they concluded that there are field differences between (a) business and the professions and (b) the social sciences (Baer and Lambert, 1990: 498). If they have in fact tested that the differences between fields are significant, then their findings can not be construed as a critique of Guimond, Palmer and Begin.3 On the other hand, if field differences are not statistically significant, they should have not concluded that field differences are significant. In sum, North American survey research does not provide a wholly clear and consistent picture of the sources of variation in professors’ political attitudes. American research suggests that class origin and current class position have no effect on liberalism and civil-libertarianism, but they do have an effect on attitude towards faculty unionism. Discipline and ethno-religious effects are observed in the American surveys, and ethno-religious effects are also evident in Canadian research on the relationship between higher education and political attitudes. However, field effects are still a matter of controversy in the Canadian research, and class is positively related to support for the dominant ideology, not negatively as in the American research. For these reasons, and also because the Canadian research deals with the general population, and not specifically with professors, there is a need for analyses that clarify the sources of variation in academics’ political attitudes in this country. In the remainder of this paper, we undertake just that task. We first analyze the sources of variation in four dimensions of political attitudes among a large sample of Canadian professors surveyed in 1987. We then discuss some of the theoretical implications of our analysis for our understanding of the political roles of intellectuals in postindustrial societies. Data and Measurement Our data source is the 1987 survey of the Academic Profession in Canada. The survey was administered by Jos Lennard to a stratified random sample of 10,212 full-time faculty members, 5,217 (51%) of whom responded. Our measures of political attitudes are derived from four questions. For two of the questions, respondents were asked to rank from 1 to 7 their agreement or disagreement with each of the following statements: “Faculty members need to be militant in order to defend their legitimate interests” (militancy). “In a period of scarce resources, faculty members should be prepared to give up individual merit pay in favour of improving general salary levels for all” (collectivism). The third question asked respondents: “Are you personally in favour of faculty unionization at your university?” (unionism). The response options were: 1 = yes, without reservation; 2 = yes, with some reservation; 3 = I am uncertain about this; 4 = somewhat opposed to the idea; and 5 = strongly opposed to the idea. Finally, respondents were told that “Political attitudes have often been described as being on the ‘left’ or on the ‘right.’ Does a left-right continuum make sense in locating your political views?” If the respondent answered “yes,” he or she was asked to locate his or her own political orientation on a 7-point scale, 1 being left, 4 being centre and 7 being right.4 These questions were all recoded so that lower and/or negative scores indicate conservative attitudes while higher and/or positive scores indicate more left-leaning responses. Among these four measures of political attitudes, liberalism is a conventional measure of political ideology which identifies left-right tendencies among respondents. The other three measures, on the other hand, are self-referential in that faculty members are asked to express their attitudes about issues pertaining to their own institutions. For example, the item dealing with union support does not represents attitudes towards the labour unions in general. It ascertains attitudes towards faculty unions. Respondents were asked to report their father’s and mother’s occupation when respondents were 16 years of age as well as both parents’ highest level of education obtained and whether their father was self-employed or supervised the work of others. We tested various measures of class background such as father’s and mother’s education and occupation, as well as a neo-Marxian conceptualization in which large employers, small employers, managers and experts are distinguished from workers (Wright, 1985: 86–94). Among the class background measures, only father’s occupation was significantly related to political attitudes. We therefore excluded other measures from further analysis. We recoded father’s occupation in six categories: professionals (professionals, self-employed professionals, and managers), semi-professionals (semi-professionals, middle-managers, supervisors foremen/forewomen), skilled workers, semi-skilled workers, unskilled workers, and farmers. Academic rank was classified as full professor, associate professor, assistant professor, sessional instructor, and other. Respondents were asked to report their basic full-time academic salary to the nearest thousand dollars for the 1987 academic year. However, because the income measure was strongly correlated with academic rank it was excluded from the analysis.5 We also used a union certification measure for the university that respondents were working in. Where respondents were unsure of certification, we cross-checked their responses with information from the Canadian Association of University Teachers. As well, we used the well-known Maclean’s magazine classification to distinguish professional/doctoral, comprehensive, and mainly undergraduate universities. The survey asked the lifetime types and number of publications of each professor. We constructed a new measure which applied the following weights for type of publication: non-refereed publication = 2, published report = 2, edited book = 3; refereed article = 3, joint book = 9, single-authored book = 18. We summed the weight of all publications and recoded this composite measure in eight categories ranging from zero to 181 and over.6 Field of specialization was measured by a question that asked respondents about the department and faculty in which their main teaching appointment is held. We recoded the variable as 1 = social sciences, 2 = humanities, 3 = arts (fine, applied, and performing), 4 = education, 5 = sciences, 6 = engineering, 7 = business. Gender was coded as 1 = female and 0 = male. Age was recoded into five categories: 34 and under (reference category), 35–44, 45–54, 55 plus. Seven hundred thirty-five respondents did not report their age. They were included as missing cases for the age variable. We used a dummy variable to distinguish Francophones in Quebec (Québécois) from others (Québécois =1, others = 0). Religious affiliation was coded in five categories: Protestant, Catholic, Jewish, other religions, and non-religious. It would be useful to know whether the respondents are truly representative of the Canadian university professoriate as a whole, especially because of the modest 51% response rate. While we cannot know this with certainty, Table 1 shows there are close similarities between the sample and the population on several important dimensions. The distribution of the sample by sex, field, and rank is virtually identical to that of the population. From an earlier analysis of this data set by Ornstein and Stewart (1996: 468), we also know that the gender income gap is extremely close to Guppy’s (1989: 748) population estimate based on administrative records for all Canadian faculty obtained from Statistics Canada. These similarities increase our confidence in the representativeness of our sample. Table 1 Frequency Distribution and Number of Cases for Field, Rank and Gender, Population and Sample Our analysis showed that over 34% of the professoriate’s fathers were in professional and managerial positions and nearly 37% had fathers in semi-professional and skilled jobs. About 20% of the professors’ fathers had semi-skilled and unskilled occupations and almost 10% of the professors’ fathers were farmers. Over 51% of the professors were in professional/doctoral universities, nearly 23% in comprehensive universities, and over 25% in primarily undergraduate universities. Over 67% of respondents were unionized. Analysis Table 2 presents standardized and unstandardized regression coefficients of liberalism, union support, militancy and collectivism on predictors. It casts doubt on Ladd and Lipset’s views concerning class theory as applied to Canadian academics.7 Professors in elite Canadian universities and professors with more publications are not more liberal than others. Contrary to what one would predict on the basis of Ladd and Lipset’s American research, assistant and associate professors, and academics with low income, are significantly more liberal than full professors and academics with high income.8 Women are also more liberal than men. Consistent with Ladd and Lipset’s findings, non-dominant religions and non-religious groups are significantly further to the left on the liberalism scale than Catholics and Protestants. In addition, Canadian field differences are much the same as in the USA. Professors in the humanities, arts, social sciences, education, and the natural sciences are significantly more liberal than those in business and engineering. Table 2 Unstandardized and Standardized Coefficients for the Regression of Political Attitudes on Predictors Table 2 also shows the relationship between union support and predictors. The results offer stronger support for the class theory than was the case with the liberalism scale. First, professors with working class fathers (skilled, semi-skilled, and unskilled) are significantly more likely to support faculty unions than those with professional and managerial fathers. The latter are not significantly different from professors with semi-professional or farming fathers. Second, associate, assistant and sessional professors are significantly more supportive of faculty unionization than full professors. Third, professors with more publications and income are less supportive of unions. Fourth, women, Québécois, and professors in primarily comprehensive and undergraduate universities are significantly more supportive of faculty unions than males, non-Québécois, and those in professional/doctoral universities. Fifth, the strongest predictor of support for unionism is whether the university in which the faculty member is employed has a certified faculty union. The effect of each of the next strongest predictors (income and being in the social sciences or the humanities) on union support is less than half that of union certification. Finally, there is a clear difference in support for unions between professors in the liberal arts and those in the applied sciences. The latter are less supportive of faculty unionism. The results for militancy and collectivism are also generally supportive of the class theory. Assistant and associate professors as well as Québécois and professors who are in unionized universities are more militant than others. Moreover, associate, assistant and sessional professors, those teaching primarily in undergraduate universities, older professors and those with lower rates of publication are more collectivists than others. Again, professors in the social sciences, humanities, arts, and education report more militancy and collectivism than professors in other fields. We now turn to an analysis of interaction effects. Research suggests that gender interacts with academic status and discipline. There are fewer female than male faculty members in universities, and women are likely to be concentrated in the lower ranks (Muszynski, 1995). Moreover, women in the social sciences and humanities tend to be more feminist in orientation than those in natural sciences, business, and engineering. As well, research suggests that Québécois in the social sciences are more left-leaning than their English counterparts (Baer and Lambert, 1990). Finally, since Québécois professors are more unionized than their counterparts in English Canada, we expect that lower rank Québécois professors are more left-leaning than their English counterparts. Using protected block tests to exclude non-significant interactions,9 we tested each of the above possible interactions. An analysis of F-tests revealed only four significant interactions: discipline interacted with both gender and with language for the collectivism measure and rank interacted with language for both the union support and militancy measures. That is, the product terms for these four sets of interactions were greater than zero (see Tables 3 and 4). The R2 changes reported in Tables 3 and 4 indicate the strength of the interaction effects. The nature of these interactions are reported as b coefficients. As can be seen, Québécois professors in the social sciences are more collectivist than Anglophones when compared to Anglophone business professors. This interaction was significant at the .07 level, which is acceptable in a one-tailed-test. No other significant relationship emerged. The only significant interaction between discipline and gender was found for professors in arts. Here, female professors were less collectivist than male professors when compared to female professors in business. Table 4 reports the interactions between rank and language for measures of union support and militancy. We thought that Québécois assistant and associate professors might be more supportive of unions and more militant than Anglophones in these ranks when compared to Québécois full professors. However, these interactions are not significant. The only significant interactions are for “others” with respect to union support and for sessionals with respect to militancy. For example, Québécois sessional professors are significantly less militant than their Anglophone counterparts when compared to Québécois full professors. Table 3 Interactions between Field and Language and Gender Table 4 Interactions between Rank and Language Accounting for Discrepancies Our analysis of Canadian data on professors’ political attitudes highlights class, field, gender, and ethno-religious effects. The most left-leaning professors in Canada tend to be those who originated in lower classes and now occupy lower-class positions in the academic hierarchy. They tend to be professors who specialize in fields other than applied science and business. They tend to be Québécois and women. They tend not to identify with the dominant religious groups in Canadian society. Bearing these findings in mind, how can we account for major discrepancies between our findings and those of Ladd and Lipset (1975) and Baer and Lambert (1990)? Two main discrepancies stand out between our study and this earlier research. First, we think there is little difference between Guimond, Palmer and Begin’s findings and those of Baer and Lambert. Both showed that the relationship between education and intergroup attitudes is not a simply linear relationship and that it varies by disciplines. Baer and Lambert showed that the linear relationship and support for dominant ideology is not statistically significant for social sciences, arts and science students but is significantly negative for business and professional students and for all students taking together. Guimond, Palmer and Begin similarly showed that social science students have more positive attitudes towards socialists and immigrants while commerce students have more positive attitudes towards the military and conservatives. Moreover, they showed that, while the overall linear relationship between level of education and intergroup attitudes is not significant for four of the seven target groups (socialists, anglophones, the military and conservatives), it is positive for attitudes toward francophones and immigrants and negative for attitudes toward unions (Guimond, Palmer and Begin, 1989: 201). The second major discrepancy is between the class effect we discovered (higher positions in the academic hierarchy produced more right-wing attitudes) and the inverse class effect found by Ladd and Lipset (1975) (higher positions in the academic hierarchy produced more left-wing attitudes). It should first of all be noted in this connection that Ladd and Lipset’s findings applied only to liberalism and attitude towards civil liberties. They found the same class effect we did on a measure of support for faculty unionism. Moreover, they failed to conduct a proper multivariate analysis. We cannot be sure that their findings would have held if they had controlled for potentially confounding factors. Finally, their survey was conducted in 1969, at the height of an unprecedented economic boom, when the North American system of higher education was flush with cash and expanding at record rates. Class effects in academia may well have been muted during this period since virtually everyone was doing well. In contrast, the survey on which we based our analysis was conducted in 1987, during a deep economic recession and at a time when government austerity measures, including educational cutbacks, had far-reaching negative effects on the well-being of Canadian university professors, particularly junior faculty members in more vulnerable, untenured positions at smaller universities. Under such conditions, one could reasonably expect to find a heightened association between class position in the academic hierarchy and leftist attitudes. Some Social Mechanisms of Political Differentiation The North American survey research reviewed above suggests that we need to abandon the key premises of Mannheimian and much of Marxist thinking about intellectuals. Assuming that intellectuals are either (relatively) classless or (more or less) firmly embedded in a particular class location encourages analysts to paper over attitudinal heterogeneity among them. Such assumptions also make it difficult for the rest of us to account in social-structural terms for variations in intellectuals’ political outlook. Intellectuals are embedded in the class structure, but in order to assess the impact of this social fact on their political attitudes it is necessary to analyse how economic and political opportunity structures affect their mobility patterns. Nor are economic and political opportunities governed only by class structures: ethnic, gender, and religious structures also create and constrain mobility. More concretely, the political allegiances of intellectuals depend on their social origins and on the structure of opportunities for education, employment, and political involvement that they face in the course of their careers. These opportunity structures are, in turn, shaped by the relative power of major classes and other groups. Thus, in order adequately to explain intellectuals’ political attitudes, one must trace their paths of social mobility as they are shaped by the capacity of classes and other groups to expand the institutional milieux through which intellectuals pass in the course of their careers. To the degree that these milieux are imprinted with the interests of the classes and other groups that control them, they circumscribe the class and other group interests reflected in intellectuals’ political attitudes. Our argument is based on Antonio Gramsci’s (1957; 1971) important writings on intellectuals. Although fully developing and illustrating Gramsci’s ideas lies well beyond the scope of this article,10 we conclude by offering a few observations that illustrate how our argument helps make better sense of the survey data reviewed above: • Social origins. Ideological predispositions are not randomly distributed in the population. To varying degrees, they are associated with certain population categories. For example, left-wing ideologies are generally associated with lower classes and right-wing ideologies with upper classes. However, the strength of that association varies over time and between countries and regions. In Canada, the association is comparatively weak, for reasons we have discussed elsewhere (Brym, Gillespie, and Lenton 1989; Nakhaie, 1992; Nakhaie and Arnold, 1996). As a consequence, in our analysis of Canadian professors’ political attitudes, we did not find a significant association between class of origin, on the one hand, and liberalism, militancy, and collectivism, on the other. The strength of the association between social origin and political attitudes is also affected by the degree to which the professoriate is open to the recruitment of people from diverse population categories. Greater diversity increases the chance of finding strong associations, less diversity decreases the likelihood. Contributing to the absence of a significant association between class origin and three of the four political attitudes we examined is the fact that, historically, a large majority of Canadian professors have been narrowly recruited from middle- and upper-class origins. For example, 71 percent of the professors in our sample have had fathers among the managerial, professional and/or semi-professional occupations. What is generally true for lower classes also usually holds for other historically disadvantaged minorities, such as Québécois, women, and non-members of dominant religious groups: historical disadvantage is generally associated with the crystallization of liberal-left attitudes.11 If the drive to achieve upward mobility on the basis of individual merit is thwarted as a result of discrimination, people seek collective solutions to the problem — collective solutions that demand the elimination of existing, discriminatory practice (Breton, 1978). Thus, Québécois professors have been more activist and left-leaning than non-Québécois professors since they were first stimulated by the independence project during the Quiet Revolution of the 1960s (Breton, 1989; Brooks and Gagnon, 1987; Brym and Myles, 1989; Brym and Saint-Pierre, 1997). Similarly, the growing feminist movement in both English and French Canada consistently focussed on women’s struggle for freedom from domination since the early 1970s (Barret and Hamilton, 1986). As a result, women in the general population (Kopinak, 1987) and in the professoriate tend to be more left-liberal than men. • Intellectual milieux. Disciplines are characterized by distinctive ideological subcultures, outlooks, and political styles. To be sure, self-selection plays a role in the recruitment of individuals to disciplines with distinctive ideological outlooks insofar as academically-inclined individuals will to some degree be attracted by the disciplines that resonate with their ideological predispositions. But the crystallization of distinctive disciplinary subcultures and styles of thought is a social force in its own right. Each discipline is a social organization that increases the frequency and intimacy of interaction among its members through professional associations, journals, and conferences. As a result, professors are likely to know members of their profession in other parts of the country or the world far better than they know professors in cognate disciplines in their own universities (Ladd and Lipset, 1975: 56). This is significant because each discipline has a distinctive approach to its subject matter, an orientation that tends to be transmitted to members of the discipline. For example, in post-World War II North America and Western Europe, liberal arts disciplines have, on the whole, developed a more critical view of their subject matter than applied disciplines (Ladd and Lipset, 1975: 68). That is partly because their subject matter requires the analysis and critique of human affairs. The “culture of critical discourse” (Gouldner, 1979) that characterizes the liberal arts is far different from the technocratic and conformist orientation of the natural scientific and applied disciplines, which do not even require or, in most cases, even suggest opinions about human affairs. One must be careful, however, to recognize the historical specificity of disciplinary political cultures. The humanities and social sciences are not universally left-wing disciplines, nor are science and engineering universally on the right. For example, in North America and Europe before World War II, technocracy and radicalism converged. The political left was committed to science and progress, which the reactionary right often mocked. Therefore, as Eric Hobsbawm (1994: 544) observed: The typical British scientist of the 1930s was a member of the (Left-wing) Cambridge Scientists’ Anti- War Group....The typical young American physicist of the 1930s was more than likely to be in political trouble in the post-war years of the Cold War for his pre-war or continuing radical sympathies....The typical French scientist was a sympathiser with the Popular Front of the 1930s and an active supporter of the Resistance during the war....The typical refugee scientist from central Europe could hardly not be hostile to fascism....Scientists who stayed in or were prevented from leaving fascist countries or the USSR could not avoid their government’s politics either.... If fascism pushed many scientists to the left, its defeat in 1945 removed one of the major causes of political radicalism among scientists. At the same time, big science came much more firmly under the control of big government and big business. Governments and corporations were happy to forge a closer union with science because of the enormous military and financial benefits it could foster. For their part, scientists who wanted access to the enormously expensive resources needed to conduct modern research were obliged to approve the marriage. • Market control of intellectual subcultures. Various markets demand intellectual products, but that demand is always accompanied by pressure on suppliers to conform ideologically to the interests of those who control the markets. Insofar as markets for different intellectual products are controlled by different classes and other groups, the professoriate is pulled in a variety of ideological directions. Governments and corporations are the main markets for the intellectual products of engineers, natural scientists, and business professors. Along with new accounting principles, bridge designs, and nickel-smelting technologies, the people who control these markets expect a high degree of ideological conformity to their corporate and government interests. In contrast, the main markets for the intellectual products of liberal arts professors are controlled by other academics, the educated public, the mass media, and various grassroots organizations, including trade unions and citizens’ groups. Those markets are more likely to allow and even encourage a left-wing outlook. In short, the ideological contrast between different segments of the professoriate is due in part to their different patterns of association with, and service for, dominant versus subordinate groups and organizations. On a broader historical canvas, therefore, the shape of the class structure and its political expression have a major bearing on the political ideologies of the professoriate. The extent to which professors develop leftist ideologies depends on such factors as the degree to which subordinate classes are organized and influential, business classes are weak and divided, regimes are moderately repressive, and political structures facilitate intellectual employment outside state bureaucracies (Brym, 1980; Brym and Myles, 1989; Karabel, 1996). In short, by investigating the structure of demand for intellectual services, one can go a long way towards understanding the social roots of intellectuals’ ideologies. The above considerations emphasize the historical variability and structural determination of intellectuals’ political attitudes. Our analysis should therefore increase skepticism about the transhistorical claims made by many Mannheimians and neo-Marxists. Intellectuals are not “free-floating.” Nor are they destined to follow a preordained political script because they are all members of a particular class or part of a class. Intellectuals are members of many groups, including classes. As the case of Canadian professors suggests, an adequate explanation of their political attitudes requires that we assess the cumulative lifetime impact of the institutional milieux through which they pass as these milieux are shaped by larger class and other group forces in particular historical contexts. Appendix: Variations by Discipline Above, we treated each field as if it is characterized by a uniform ideological outlook. However, from the American literature, we know that there is considerable ideological heterogeneity within academic fields (Ladd and Lipset, 1975). Table 5 addresses this issue by showing a detailed distribution of our four measures of political attitudes by discipline. Table 5 shows the following: Table 5 Four Dimensions of Political Attitudes by Field and Discipline • Overall, professors scored to the left of centre on all four of our measures, although less so with respect to militancy and significantly less so with regard to collectivism. The mean scores for liberalism, militancy and collectivism are 4.64, 3.93 and 3.63, respectively. These scales range form 1 to 7. The mean score for unionism is 3.30. Union support ranges from 1 to 5. • The political attitudes of social science, humanity, arts, and education professors cluster together, as do the political attitudes of professors in technical and applied disciplines. • As we move from the social sciences to the humanities to education to arts, and then to the sciences, business, and engineering, the leftist tendencies of the professoriate decrease and their rightist tendencies increase. • As we move from liberalism to unionism and then from militancy to collectivism, the leftist tendencies of professors in all disciplines decrease and their rightist tendencies increase. (Keep in mind that unionism is measured on a five-point scale, while the other three measures are on a seven-point scale). Thus, Canadian professors are more likely to express liberalism, support faculty unionism, and wish to act militantly than to express support for the economic well-being of their colleagues. • Within each field, the Canadian professoriate is differentiated. For example, there are 0.86-point and 1.02-point differences between sociologists and geographers in liberalism and militancy respectively. Also note the gaps between economists and those in accounting. Said differently, Canadian professors are differentiated ideologically by field and subfield. Notes 0 “The deepest rift... [is] between the managers, administrators, and engineers, on the one hand, and those in the liberal arts and professions, on the other.... But we should not overestimate the significance of this division. The PMC [professional-managerial class] at mid-century still constituted a single, coherent class” (Ehrenreich and Ehrenreich, 1979: 28). “The search for an agent [of historical change] is a search for power by those who feel that they already have knowledge....Which intellectuals choose which agents, whether high or low, depends in part on their values and ideals and, in part, on their structured opportunities [neither of which is examined by Gouldner -- R.N. and R. B.]....Yet both are probably shopping with similar motives and with the same idea of what constitutes a desirable agent....Marxism [for example] is....a synthesis of the vanguard ideologies of intellectuals and technical intelligentsia, facilitating their political unity” (Gouldner, 1985: 26, 25, 41). “Different categories of intellectuals....will be closer to the proletariat or the petty bourgeoisie...There will inevitably occur class locations which are not only contradictory, but ambiguous, i.e., it is impossible to provide an abstract, a priori classification of such positions. What this indicates is the inevitable limitation of a purely structural analysis of class and class relations” [due to which, Wright never tries to explain ideological heterogeneity in social-structural terms -- R. N. and R. B.] (Wright, 1979: 206, fn. 13). back to main text 1 We restrict our overview to the North American literature because we recognize the difficulty of making transhistorical and transcontextual generalizations on the subject of intellectuals’ attitudes (see below). back to main text 2 Throughout our discussion, “field” refers to a cluster of cognate disciplines (social science, business, etc.), while “discipline” refers to a narrow area of study (sociology, accounting, etc.). back to main text 3 Guimond, Palmer and Begin (1989:204) have argued that “[s]ocial science students at the university level are more positive towards ‘socialists’ while commerce students are more negative...” They (1989:205) reiterated that “the results of the present study regarding ‘unions’ replicated the findings of Curtis and Lambert (1976) within the commerce and science samples but not in the social science sample” (for a continuation of this debate, see Guimond and Palmer, 1994; Baer and Lambert, 1995). back to main text 4 Although the non-response rate for the liberalism item is high (38.4%), it is not problematic because we are interested in group differences and we found that the rate of non-response was uniform across various groups. One should also bear in mind that there is nothing unusual about our procedure. Liberalism is typically measured in the way we measure it here and non-response rates are typically high. back to main text 5 When income was included in the model containing all other variables it was significantly related to Liberalism and union support in that the higher the income the lower liberalism and union support. However, inclusion of this measure systematically reduced the differences between associate professors (and eliminated that of assistant professors and sessional category) with full professors. Other results were essentially unchanged. Because of such a multicollinearity, we decided to exclude income from our analysis. back to main text 6 Obviously, our measure is far from perfect. For example, we are unable to account for differences in quality of publications, and our ad hoc weighting procedure may overestimate the importance of books. However, based on our combined experience as faculty members at five different universities in Atlantic Canada and Ontario, and our general knowledge of other universities in the country, we believe that this categorization adequately captures the weight accorded to different types of publications across various Departments. Furthermore, we examined the effect of each type of publication separately. Results showed that among various types of publications, the number of refereed articles and chapters in a life time were significantly related to the measure of union support (b = –.004; sig = .0002). For our measure of collectivism, two indicators of publication was shown to be significant: number of articles and chapters in a life time (b = –.009; sig. = .0000) and number of books edited in a life time ( b = –.084; sig. = .0000). Other types of publications were not statistically related to any of the political attitudes. back to main text 7 We use the term “class” in much the same was as Ladd and Lipset (1975), to capture hierarchal inequalities in universities. While this conceptualization obfuscates the intrinsically relational, exploitative and antagonistic relations between classes that are rooted in the social division of labour, we found little support for the class effect, whether measured by parent’s educational background or in neo-Marxist terms. back to main text 8 Rank and income are highly correlated. Due to multicollinearity, we did not enter both independent variables simultaneously. Therefore, in the multiple regression analysis reported in Table 2, we show only the association of liberalism with rank. back to main text 9 For each of these interactions we estimated a model. Thus, for example, the estimated model for the interaction of gender with discipline becomes: Y = a + b1*female + b2* education + b3* arts + ... + b7 * engineering + b8 (education * female) + b9 (arts * female) + ... +b13 (engineering * female) (with business as the reference category) + (all other variables). back to main text 10 For this, see Brym (1978; 1980; 1987; 1995); Brym and Myles (1989). back to main text 11 We emphasize “historical” because, as the case of North American Jews illustrates, it may be generations before upward mobility erases a culturally ingrained sense of disadvantage. back to main text References Avineri, S. 1957 “Marx and the intellectuals.” Journal of the History of Ideas 28: 269–278. Baer, E. and R. Lambert 1982 “Education and support for dominant ideology.” Canadian Review of Sociology and Anthropology 19: 173–195. 1990 “Socialization into dominant vs. counter ideology among university educated Canadians.” Canadian Review of Sociology and Anthropology 27(4): 487–504. 1995 “The politics of Canadian social scientists: A reply to Guimond and Palmer.” Canadian Review of Sociology and Anthropology 32(2): 243–246. Barret, M. and R. Hamilton 1986 The Politics of Diversity. London: Verso. Breton, R. 1978 “Stratification and conflict between ethnolinguistic communities with different social structures.” Canadian Review of Sociology and Anthropology 15: 148–157. 1989 “Quebec sociology: Agendas from society or from sociologists?” Canadian Review of Sociology and Anthropology 26: 557–570. Brint, S. 1984 “‘New class’ and cumulative trend explanations of the liberal political attitudes of professionals.” American Journal of Sociology 90: 30–71. Brooks, S. and A. Gagnon 1987 “Social scientists and politics in Canada.” In A. Gagnon, ed. Intellectuals in Liberal Democracies, pp. 19–42. New York: Praeger. Brym, R. 1978 The Jewish Intelligentsia and Russian Marxism: A Sociological Study of Intellectual Radicalism and Ideological Divergence. London: Macmillan. 1995 “Intellectuals.” 1995. Encyclopaedia of Democracy, 4 vols., S. M. Lipset, ed. Washington: Congressional Quarterly Books, 2: 613–616. 1980 Intellectuals and Politics. London: George Allen & Unwin. 1987 “The political sociology of intellectuals: A critique and a proposal.” In A. Gagnon, ed. Intellectuals in Liberal Democracies, pp. 199–210. New York: Praeger. Brym, R., M. Gillespie, and R. Lenton 1989 “Class power, class mobilization, and class voting: The Canadian case.” Canadian Journal of Sociology 14: 25–44. Brym, R. and J. Myles 1989 “Social science intellectuals and public issues in English Canada.” University of Toronto Quarterly 58(4): 442–451. Brym, R. and C. Saint-Pierre 1997 “Canada.” Contemporary Sociology 26: 543–546. Caute, D. 1966 The Left in Europe Since 1789. New York: McGraw-Hill. Curtis, J. and R. Lambert 1976 “Educational status and reactions to social and political heterogeneity.” Canadian Review of Sociology and Anthropology 13: 189–203. Ehrenreich, B. and J. Ehrenreich 1979 “The professional-managerial class.” In Pat Walker, ed. Between Labor and Capital, pp. 5–45. Montreal: Black Rose Books. Gouldner, A. 1979 The Future of Intellectuals and the Rise of the New Class. New York: Macmillan. 1985 Against Fragmentation: The Origins of Marxism and the Sociology of Intellectuals. New York: Oxford University Press. Gramsci, A. 1957 “The southern question.” In The Modern Prince and Other Writings, pp. 28–51, L. Marks, trans. New York: International Publishers. 1971 Selections from the Prison Notebooks, Q. Hoare and G. Smith, eds. London: Lawrence and Wishart. Guimond, S. and D. Palmer 1990 “Type of academic training and causal attributions for social problems.” European Journal of Social Psychology 20: 61–75. 1994 “The politics of Canadian social scientists.” Canadian Review of Sociology and Anthropology 31(2): 184–195. Guimond, S., D. Palmer and G. Begin 1989a “Education, academic program and intergroup attitudes.” Canadian Review of Sociology and Anthropology 26(2): 193–216. 1989b “Education and causal attributions: The development of ‘person-blame’ and ‘system-blame’ ideology.” Social Psychology Quarterly 52: 126–140. Guppy, N. 1989 “Pay equity in Canadian universities, 1972-73 and 1985-86.” Canadian Review of Sociology and Anthropology 26: 743-758 Hobsbawm, E. 1994 Age of Extremes: The Short Twentieth Century, 1914–1991. London: Abacus. Karabel, J. 1996 “Towards a theory of intellectuals and politics.” Theory and Society 25: 205–233. Kopinak, K. 1987 “Gender differences in political ideology in Canada.” Canadian Review of Sociology and Anthropology 24: 23–38. Johnson, A. 1985 Giving Greater Point and Purpose to the Federal Financing of Post-Secondary Education in Canada. Ottawa: Office of the Secretary of State. Ladd, E. and S. Lipset 1972a “The politics of American political scientists.” PS 4: 135–144. 1972b “Politics of academic natural scientists and engineers.” Science 176: 1091–1100. 1975 The Divided Academy. New York: McGraw-Hill. Lambert, R. and J. Curtis 1979a “Educational status and subscription to dominant ideology.” In J. E. Curtis and W. G. Scott, eds. Social Inequality in Canada, pp. 412–423. Scarborough: Prentice-Hall of Canada. Lipset, S. 1970 “The politics of academia.” In D. C. Nichols, ed. Perspectives on Campus Tensions, pp. 85–118. Washington: American Council on Education. 1972 “Academia and politics in America.” In T. Nossiter, A. Hanson, and S. Rokkan, eds. Imagination and Precision in the Social Sciences, pp. 211–290. London: Faber and Faber. Lipset, S. and R. Dobson 1972 “The intellectual as critic and rebel: With special reference to the United States and the Soviet Union.” Daedalus. 101(3): 137–198. Lipset, S. and E. Ladd 1972 “The politics of American sociologists.” American Journal of Sociology 78: 67–104. 1974 “The myth of the ‘conservative’ professor: A reply to Michael Faia.” Sociology of Education 47: 203–213. Malia, M. 1961 “What is the intelligentsia?” In R. Pipes, ed. The Russian Intelligentsia, pp. 1–18. New York: Columbia University Press. Mannheim, K. 1955 Ideology and Utopia. L. Wirth and E. Shils, trans. New York: Harvest. Nakhaie, M. 1992 “Class and voting consistency in Canada: Analyses bearing on the mobilization thesis.” Canadian Journal of Sociology 17: 275–299. Nakhaie, M. and R. Arnold 1996 “Class position, class ideology and class voting: Mobilization of support for the New Democratic Party in the Canadian election of 1984.” Canadian Review of Sociology and Anthropology 33: 181–213. Ornstein, M. and P. Stewart 1996 “Gender and faculty pay in Canada.” Canadian Journal of Sociology 21: 451-481. Parsons, T. 1969 “The intellectual: A social role category.” In P. Rieff, ed. On Intellectuals, pp. 3–24. Garden City NJ: Anchor. Wright, E. 1979 “Intellectuals and the class structure of capitalist society.” In Pat Walker, ed. Between Labor and Capital, pp. 191–211. Montreal: Black Rose Books. 1985 Classes. London: Verso.