(see his full essay below).

Interdisciplinary Ph.D. Program in Social Psychology

RUNNING HEAD: Academic success and political orientation

What's in a grade?:
Academic success and political orientation

Markus Kemmelmeier Cherry Danielson
University of Nevada Wabash College

Jay Basten
University of Michigan

Published version:
Kemmelmeier, M., Danielson, C., & Basten, J. (2005). What's in a grade?: Academic success and political orientation. Personality and Social Psychology Bulletin, 31. 1386-1399.

Corresponding author:

Markus Kemmelmeier
Interdisciplinary Ph.D. Program in Social Psychology
Department of Sociology/300
University of Nevada
Reno, NV 89557
Phone: (775) 784-1287
Fax: (425) 790-6838
markusk@unr.edu
What's in a grade?:
Academic success and political orientation

Abstract
Expanding the literature on person-environment fit, we argue that political orientation is an important factor in shaping academic success in college. Based on social dominance theory (Sidanius & Pratto, 1999), it was expected that academic disciplines which are more likely to provide students with future access to social and economic power tend to favor individuals who hold attitudes that strengthen the existing societal order. In a longitudinal sample of undergraduate students at a major American university (n = 3890), we demonstrated that student grades in these disciplines, but not in other disciplines, are positively related to a pre-college measure of conservatism. This association between conservatism is consistent over time and subgroups, thus implicating higher education in the reproduction of social hierarchy. The discussion examines the causal processes underlying the relationship between political orientation and academic success in college.
(138 words)

Key words: PERSON-ENVIRONMENT FIT
POLITICAL ORIENTATION
CONSERVATISM
ACADEMIC SUCCESS
SOCIAL DOMINANCE THEORY
MULTILEVEL ANALYSIS
What's in a grade?:
Academic success and political orientation
People are more successful in environments where they "fit." This idea of person-environment fit is not only intuitively appealing, but it has received much empirical support in the social and behavioral sciences. For instance, employees whose values match those of their workplace environment are more satisfied and more successful than those whose values are more discrepant, often resulting in a longer tenure with the organization (e.g., Chatman, 1991; O'Reilly, Chatman, & Caldwell, 1991). Similarly, college students whose values approximated those of their instructors tend to receive higher grades (Abrami & Mizener, 1985) and are more satisfied with their college experience (Gottfredson & Holland, 1990; Mount & Muchinsky, 1978; Nafziger, Holland, & Gottfredson, 1975). Hence, there appears to be little doubt that individual performance and outcomes are at least in part a result of the fit between individual and contextual characteristics (see also, e.g., Goodman & Svyantek, 1999; Holland, 1966, 1985; Walsh, Craik & Price, 2000).
In the present research we examine the impact of socio-political beliefs as a determinant of "fit" within the context of higher education. Although few academic environments are openly committed to specific political ideologies, academic disciplines vary with regard to the kinds of socio-political messages they communicate to their students. For instance, as students become socialized into their respective fields, social science students are more likely to blame social problems such as poverty and unemployment as residing in the political system, whereas business students are more likely to blame the poor and unemployed themselves (Guimond & Palmer, 1990; see also Guimond, Begin & Palmer, 1989; Guimond & Palmer, 1996; Guimond, 1999). To the extent that certain political worldviews are dominant in different academic environments, it must be expected that the fit between an individual's views and those prevalent in the academic field helps shape the individual's performance in that field. Specifically, in the present paper we examine to what extent students' political orientation shape their academic success in different academic environments.
Social Dominance Theory as a framework for studying political orientation in education
Social dominance theory (SDT) provides a comprehensive framework for studying the interactive effect of individual political orientation and various educational environments (Sidanius, Pratto, Martin, & Stallworth, 1991; Sidanius & Pratto, 1999). Drawing on social identity theory (Tajfel & Turner, 1986), elite theories (Dahrendorf, 1959, Michels, 1962; Mosca, 1939; Pareto, 1979), and modern stratification theory (Van den Berghe, 1978), SDT is predicated on the assumption that societies are inherently hierarchical, with some groups having greater access to resources and power than other groups. Because dominant groups are likely to embrace this hierarchical arrangement, they are motivated to maintain their favorable social position in various ways. For instance, SDT predicts that members of dominant groups discriminate against members of subordinate groups in important societal domains, either in interpersonal contexts or through societal institutions that produce differential outcomes for dominant and subordinate groups (see Sidanius & Pratto, 1999 for a review).
The maintenance of societal hierarchy is facilitated by beliefs that legitimize the superior positions of dominants and help justify the subordinate position of others. Hierarchy-enhancing (HE) beliefs are ideas and attitudes that provide an intellectual or moral justification for unequal relationships between groups. Prime examples include racism, sexism, classism, elitism and conservatism. In contrast, hierarchy-attenuating (HA) beliefs refer to views that are antithetical to the idea of social hierarchy, such as a belief in various universal rights of humans, egalitarianism and socialism (see Sidanius & Pratto, 1999, for a comprehensive discussion). HA beliefs tend to be associated with the support of or opposition to a variety of policies aimed at reducing social inequality, e.g., affirmative action or women's rights, whereas HE beliefs tend to be associated with supporting policies that maintain group hierarchy (e.g., Federico & Sidanius, 2002; Pratto, Sidanius, Stallworth, & Malle, 1994).
Because HE ideology helps to justify the superior position of the dominant group, the distribution of HA and HE beliefs in society follows group position. For example, groups that occupy dominant positions in the societal hierarchy tend to endorse HE beliefs more strongly than those who are in subordinate positions. Conversely, because HA beliefs challenge societal inequalities, subordinate groups are more likely to hold HA beliefs than dominant groups. For example, women and members of racial-ethnic minority groups are more likely to hold HA beliefs and less likely to hold HE beliefs (e.g., Pratto, Stallworth, & Sidanius, 1997; Sidanius, Pratto & Rabinowitz, 1994; Sidanius, Pratto & Brief, 1995)-a pattern that is cross-culturally invariant (Pratto, Liu, Levin, Sidanius, Shih, Bachrach & Hegarty, 2000).
Similar to classifying beliefs, SDT allows the classification of environments according to whether they support and promote either a HE ideology or a HA ideology (Van Laar, Sidanius, Rabinowitz, & Sinclair, 1999). HE environments refer to institutions and organizations that hold a substantial degree of social, economic and political power, or which provide access to these types of power. Typically, HE environments are committed to the existing social, economic, and political order, which maintains the group-based system of social hierarchy. Some HE institutions and organizations can be linked to the disproportionate allocation of benefits to members of dominant groups (e.g., money, opportunity, quality education, good health), and the disproportionate infliction of harm on those of non-dominant groups (e.g., low income, capital punishment, inferior education, poor health). Examples of HE environments include criminal prosecutors, the penal system, police and security forces and, often, corporate firms and their executives (e.g., Van Laar et al., 1999). Examples of educational institutions that allow individuals to assume positions in HE environments are typically those that prepare individuals to assume leadership positions in business, economics, and politics. With regard to higher education within a capitalist system, this applies mainly to business schools and economics departments.1
In contrast, HA institutions and organizations tend to be committed to principles of equality and social justice, i.e. reducing the disparity between dominant and subordinate groups. Typically, HA institutions command only comparatively low levels of social or economic power. Examples include human rights organizations, the public defender's office, labor organizers, and social welfare organizations (Van Laar et al., 1999). Similar to HE institutions, there are educational institutions that actively promote social equality, such as social work schools or departments of sociology on many university campuses.
Matching up persons and environments
There is good reason to believe that individuals with HE and HA beliefs (henceforth: HE and HA individuals) are most likely to end up in corresponding HE and HA environments. For instance, Sidanius, Liu, Pratto and Shaw (1994) showed that police officers are more likely to endorse HE beliefs than public defenders who, in turn, are more likely to hold HA beliefs. Van Laar et al. (1999) proposed four different processes by which congruence between person and environment is created: (a) institutional selection; (b) institutional socialization; (c) self-selection; and (d) differential success. First, institutional selection refers to the process by which HE or HA institutions selectively admit individuals that "fit in", e.g., based on egalitarian or elitist predispositions of job candidates (Pratto, Stallworth, Sidanius & Siers, 1997). Second, institutional socialization is the process by which organizations instill their values and beliefs into newly recruited members. For instance, Carlson and Sutton (1974) and Teahan (1975) found that, over the course of their training, police recruits tended to become more authoritarian and more prejudiced. Guimond (2000) reported similar findings for military officers.
Third, through the process of self-selection, HE and HA individuals are assumed to choose to be in environments that correspond to their beliefs. Indeed, individuals with HE beliefs are more likely to aspire to careers that are associated with social and economic power (Pratto et al., 1997; Sidanius, Pratto, Sinclair & van Laar, 1996; Sidanius, Van Laar, Levin & Sinclair, 2003). Lastly, there is reason to suspect that institutions differentially reward those whose orientation matches their organizational culture, even when controlling for other factors. A study by Leitner and Sedlacek (1976) on performance ratings of campus police officers is a case in point. These authors found that even after other factors were statistically controlled, more prejudiced police officers tended to receive more favorable performance evaluations. While all four processes identified by Van Laar at al. (1999) may contribute to creating a match between HE and HA persons and environments, in the present study we focus on the latter two, namely, self-selection and the issue of differential success in higher education.
Academic success and -political orientation
When students enter college, they are likely to hold socio-political beliefs that can be described in terms of the HE/HA distinction. For instance, they may describe themselves as conservative or liberal, and be more or less prejudiced against members of other groups. The endorsement of HE or HA beliefs should have implications for their academic success in HE and HA environments. To the extent that the ideologies of student and academic environment are concordant, the student can be expected to receive higher grades than students in the same academic environment whose ideology is discordant with it. In other words, grades may be in part a reflection of the ideological fit between students and their major. This leads to the prediction that HE environments should be conducive to the academic success of the HE students, whereas they are not conducive to the success of students who embrace an HA ideology.
Two recent studies have thus far studied the role of HE beliefs for student outcomes in higher education. Using a cross-sectional design, Van Laar et al. (1999) examined the implication of levels of racism, a HE belief, for students with HE or HA majors, whereas Sidanius et al. (2003) addressed the same question using a longitudinal design. Both studies showed that, consistent with the idea of a person-environment fit, college students who were classified as having an HE major were high on measures of anti-egalitarian beliefs (e.g., racism) compared to students who were classified as having an HA major. Van Laar et al. found that this difference that became more pronounced as students advanced in their educational career (see also Sidanius et al., 1991), but this effect was not replicated by Sidanius et al. (2003).2
More importantly, both studies found evidence that the relationship between HE beliefs and academic success varied as a function of HE or HA major. In HA majors (e.g., sociology, anthropology etc.), there was a substantial negative relationship between the students' level of racism and their grade point average (GPA), whereas for students in HE majors the direction of the relationship was still negative, but nonsignificant. In other words, within HA majors HE students were less successful than HA students. These data provide preliminary evidence that HE beliefs are an important factor in shaping students' academic success in HA and HE environments in higher education.
Studying person-environment effects in higher education
While their findings were consistent with expectations, the studies by Van Laar et al. (1999) and Sidanius et al. (2003) also illustrate a problem that has continuously plagued the research on person-environment fit. If an individual succeeds in a particular environment, it is unclear whether the person would have succeeded in another environment, simply because individuals choose to be in specific environments and are not randomly assigned to them. That is, although an interaction between person and environment is presumed, strictly speaking it is not demonstrated because the person's success in the environment is confounded with self-selection. In order to be able to attribute causality to the environment, one may want to use random assignment to pair up individuals with environments, e.g., students with majors, but such a procedure would be ethically unacceptable in higher education. Alternatively, one may want to demonstrate that the same person does succeed in one environment, but does not succeed in another environment. If the person is kept constant across environments, an interaction between person and environment can be inferred with great confidence.
In the present research, we avoided any possible self-selection confound by choosing this second approach, taking advantage of the unique structure of college. In contrast to other organizations, in college it is not unusual for students to expose themselves to different environments during the same time period. Whereas the majority of students has only one major, they may nevertheless attend both courses taught by HE departments and courses taught by HA departments during the same semester. This allows the researcher to conduct a "within-person" analysis in which the student is kept constant, but in which only the environment is varied. To the extent that the same person's characteristics have different implication in different environments there is strong evidence for a person-environment interaction.
A related problem in the studies by Van Laar et al. study (1999) and Sidanius et al. (2003) results from using a student's of overall GPA as measure of academic success. Overall GPA is a composite of all grades that a student has received during his or her academic career at an academic institution, including grades pertaining to the student's major as well as grades in courses that the students took outside of their declared major. Thus, it is not clear that the variability in the relationship between overall GPA and racism for students in HA majors was driven by grades received in the HA major, by grades received in HE courses or by classes that were neither taught by HE or HA departments. For example, it is possible that the negative relationship between HE beliefs and grades in HA majors found by Van Laar et al. (1999) and Sidanius et al. (2003) was due to high HE students performing poorly in non-HA classes.
To avoid this ambiguity, in our investigation we use GPA scores that are specific to performance in different types of classes. In particular, we computed a GPA score reflecting performance in classes taught by HE departments and, separately, we computed a GPA score reflect performance in classes taught by HA departments. To the extent that socio-political ideology is indeed differentially associated with HE grades and HA grades, one can be confident that any relationship between socio-political ideology and GPA is not created by third variables. Further, this approach allows one to conduct a strict test of the differential implications of HE beliefs in different academic environments, as, by analyzing the HE grades and the HA grades of the same student, the person can be kept constant across environments.
Asymmetrical vs. symmetrical person-environment effects
While rarely discussed in the literature on person-environment fit, it is plausible to assume that social environments vary in the degree to which they expect participants to "fit". Some organizations, institutions or departments expect their members to adopt or comply with a pre-defined set of norms and expectations, whereas others are more tolerant of diverging views (e.g., O'Reilly & Chatman, 1996; see also the related concept of tightness-looseness in cross-cultural psychology, e.g., Chan, Gelfand, Triandis & Tzeng, 1996; Carpenter, 2000). In fact, the data by Van Laar et al. (1999) and Sidanius et al. (2003) suggest an asymmetrical association between HE beliefs and academic success in HE and HA environments: Only in HA environments did HE beliefs predict grades, but not in HE environments, suggesting that HA environments are less tolerant of those with diverging views than HE environments. Note, however, that the interpretation of the Van Laar et al. and Sidanius et al. data is troubled by the ambiguities identified earlier.
In the present context, we argue that HE environments are more likely to reinforce HE values than are HA environments to reinforce HA values. The central argument for this asymmetry is derived from the differential implications of mismatched person-environments for the maintenance of social hierarchy. For any group-based hierarchy to continue to exist, it is important to ensure that those holding the majority of power and resources in society are also invested in the hierarchy's continued existence (cf. Mills, 1956; Sidanius & Pratto, 1999). Given the turnover that HE organizations and institutions inevitably experience, gate-keeper institutions must ensure that those next in line to assume power share the HE ideology of the current elite. If the gate-keeping processes breaks down and those transported to the top hold HA beliefs, it is likely that the hierarchy ceases to exist. In other words, not promoting HE individuals within HE institutions may have devastating consequences for the system of group-based dominance itself. By comparison, any system of differential selection occurring within HA institutions is much less consequential. Promoting HA individuals and impeding HE individuals might encourage opposition to the social hierarchy among the part of HA individuals. But because HA institutions are less likely to advance their members to assume the reigns of power than HE institutions, such processes have relatively little implication for the system as a whole. (The only exception being the case in which individuals with HE beliefs are "accidentally" promoted within HA institutions, which would undermine the effectiveness of HA institution and strengthen any existing system of group-based dominance.)
This analysis suggests that the continued existence of the social hierarchy is foremost a matter of the effectiveness of HE institutions. Given that hierarchical social systems seem to have reproduced themselves successfully in virtually every known society (Sidanius & Pratto, 1999), it must be expected that HE institutions will be more likely to produce an advantage for HE individuals than HA institutions will for HA individuals. This is consistent with sociological evidence showing that HE institutions, especially in education, tend to reproduce existing patterns of social stratification (e.g., Cookson & Persell, 1985; Domhoff, 1983, 2001; Mills, 1956).
The present study
In the present study we focus on political orientation, i.e. self-described liberalism-conservatism, as a predictor of academic achievement in college. Although the term can assume a variety of different meanings, conservatism is most commonly associated with support for capitalism and general opposition to equality of individual outcomes (Jost, Glaser, Kruglanski, & Sulloway, 2003; Sidanius & Pratto, 1999). Social dominance theorists have demonstrated that conservatism is connected to numerous HE beliefs, including racism. Moreover, the association between conservatism and other HE beliefs is fully explained by social dominance orientation (e.g., Sidanius & Pratto, 1993; Sidanius, Pratto & Bobo, 1996). In other words, conservatism shares with other HE beliefs a support for group inequality and, thus, is generally considered a HE belief itself (Pratto et al., 1994). We predict that higher levels of conservatism are systematically linked to academic success in HE environments but not in HA environments.
Using data from a four-year longitudinal study, we examined students' academic success as a function of conservatism while controlling for a range of predictors of academic success. Previous investigations by Van Laar et al. (1999) and Sidanius et al. (2003) systematically controlled for the influence of race-ethnicity and gender in their analyses.3 Regrettably, these theorists did not take into account more substantive predictors of academic performance, such as standardized test scores. This raises questions with regard to the interpretation of any correlation between grades and HE beliefs, as such an association could be a reflection of the differential academic preparation on the part of students with HE beliefs. Indeed, in order to identify non-ability related influences on grades, it is imperative that objective levels of academic preparation be partialed out (cf. Farkas, Grobe, Sheehan & Shuan, 1990; Farkas, Sheehan & Grobe, 1990). Therefore, to enhance the interpretability of our data, we included SAT scores as a covariate in all our analyses. By controlling academic aptitude, any differential implications of conservatism in different academic environments reflect genuine person-environment effects, that is, differential responses of (HE or HA) institutions to (HE and HA) individuals, and vice versa.
Method
Respondents
This study focused on one entering cohort at a major public university in the United States (n = 5534) during the late 1990s. In conjunction with the Cooperative Institutional Research program (CIRP), new entering students completed a comprehensive freshman survey, which was administered during student orientation, prior to students' first semester. The response rate to the CIRP survey was 89%. The present sample focuses on students who granted researchers permission to access their academic information during their college career (approximately 85% of CIRP respondents), who had completed the measure of political orientation (see below), and for whom information on scholastic aptitude prior to college entry (SAT or ACT equivalent) was available. Thus, the analysis sample is comprised of 3890 student (50% women). Most students described themselves as White (78%), about 7% described themselves as Black, 2% as Native American, 4% as Latino/Hispanic and 12% as Asian American. (Because of the number of multi-ethnic students, roughly 5% of the sample, self-descriptions using more than one racial-ethnic label were permitted.) Analysis of the demographic characteristics of this sample revealed that it closely matched the entering cohort as a whole.
Data and procedures
Political orientation. As part of the CIRP, students were asked to describe their political orientation as "far left" (2.7%), "liberal" (34.6%), "middle of the road" (42.0%), "conservative" (34.6%) or "far right" (1.2%).4 These responses were coded as 1 through 5, respectively. Note that in the present study, political orientation was only assessed once prior to college entry, and not, as would be desirable, repeatedly over the course of the study. However, this does not appear to be a serious limitation as Sidanius et al. (2003) demonstrated that students' HE beliefs are highly stable during college. Astin (1993) further showed that political orientation is largely stable during the college years, even though there is some evidence that students tend to become more polarized (see also Dey, 1996).
Outcome variables. From the registrar's office, we obtained detailed academic records for all students in this study. This was comprised of all courses a student had completed during the four years of the study, including the letter grade and credits received for each course. Drawing on the coding schema provided by Van Laar et al. (1999) and Sidanius et al. (2003), we coded classes based on whether they were taught in the Economics department and by the Business School, both HE environments according to SDT, or whether they were taught in HA disciplines. Specifically, classes taught by the Business School comprised the areas of accounting, marketing, finance, management, corporate strategy and organizational behavior/human resource management. Classes taught by the following departments of programs were considered HA: American culture, African American studies, cultural anthropology, education, nursing, sociology and women's studies. (Courses taught by other departments were not examined in this study.) All letter grades were converted to grade points according to the conversion scheme used by the university.
Control variables. From the CIRP survey, information was selected on aspects of students' socio-demographic background. These factors included sex, ethnicity/race (dummy coded), mother's and father's education, parental income, and whether English was the student's first language. Further, scholastic aptitude was included as additional control variable using students' SAT or converted ACT scores.
Results
Multilevel modeling
As primary analytical strategy in this research we relied on multilevel regression modeling (e.g., Bryk & Raudenbush, 1992; Heck & Thomas, 2000). This statistical approach offers a number of important advantages. First, it allowed us to deal with the fact that our data were nested and, therefore, nonindependent, which renders more conventional analytical methods such as ANOVA or OLS regression inappropriate. Second, multilevel modeling allows the simultaneous estimation of several effects of interest at multiple levels of analysis. This aspect makes it is particularly suitable for the estimation for within-person differences across different environments and time. Third, not all cases (e.g., students) need to contribute an equal numbers of observation (e.g., grades from both HA courses and HA course), as multilevel models are based on maximum likelihood estimation procedures, which renders them highly tolerant of incomplete or missing data. This latter feature was particularly important given that course-taking patterns in higher education are highly variable between students, with only few students taking the same types of courses at the same time.
All models were estimated using the linear mixed model procedure MIXED offered by SPSS 12.0.2 (e.g., Noru_is, 2003). Typically, we modeled theoretically interesting factors and covariate and compared several viable variants of the same basic model. The best-fitting model was determined based on various information criteria, including Schwarz's Bayesian Criterion (SBC), Aikike's Information Criterion (AIC) and the -2 restricted likelihood log.
Self-selection processes and person-environment fit
In a first step, we sought to replicate the finding by Van Laar et al. (1999) and Sidanius et al. (2003) that students sort themselves into HE and HA environments according to their HE beliefs. That is, we asked whether students who choose HE courses are more likely to be conservative than students who choose HA courses. In addressing this issue it is critical to consider that students are nested within particular classes (e.g., Economics 101, Economics 435, Sociology 101 etc.), and that data from individual students are not independent. For instance, two students taking Economics 101 are likely to be more similar in their political orientation than a student taking Economics 101 and a student taking Sociology 101. Therefore, in the present model we compared courses with respect to the conservatism of the students who take them. Notice that this approach rephrases the above question somewhat, as we are now asking whether HE courses (e.g., Economics 101) are more likely to attract conservative students than are HA courses (e.g., Sociology 101).
Our model took into consideration that specific courses were taught repeatedly during the four years considered in this study, and that each time a course was taught it comprised different students.5 In addition, we allowed for the possibility that the average political orientation of students taking a particular course might change over time. Specifically, in the model the data were structured such that students were nested within semesters, and semesters were nested within courses. Course type (HA vs. HE vs. neither HA nor HE) was treated as a fixed-effect predictor. Time (semesters) was assumed to have a linear effect on course takers' political orientation and was modeled as a random effect. (The inclusion of quadratic and cubic terms did not improve the model fit). Moreover, we included the interaction between course type and time (semester) to see whether the anticipated difference between HA courses and HE courses would vary over time. Lastly, we included all of our control variables as fixed-effect predictors. (Results for control variables are not reported here.)
As expected, there was a pronounced difference in the political orientation of students in HE courses compared to students in both HA courses and controls, i.e., courses that were neither classified as HA nor HE (see Table 1). Because students in HA courses did not differ reliably from those control, it appears that HE courses consistently attracted more conservative students than any other type of courses. This confirms the notion that students with HE beliefs self-select into courses and disciplines that are compatible with their mindset. It also supports the idea of an asymmetry in the effects of HA and HE environments, as students in HA courses did not differ from the bulk of courses taught at the university.
Interestingly, the interaction terms show that the differences in the political orientation of students in HA courses and those in HE courses were shrinking over time, b = -.028, SE = .011, p < .01, as was the difference between HE courses and control courses, b = -.037, SE = .009, p < .001. The simple effects in Table 1 reveal that this was primarily due to the fact that students taking the same HE courses later rather than earlier in their college careers were less conservative, but no similar change was found for HA courses. Students taking control courses tended to become slightly more conservative over time; yet the gap between control courses and HE courses never closes: When comparing estimated means for students' last semester (spring term of their fourth year), conservatism levels in control courses are still significantly lower than those in HE courses.
Differential success as a result of person-environment fit
Recall our prediction that conservatism would be related to academic success in HE environments, but not in HA environments. To test this hypothesis, we generated a multilevel model which examined the impact of conservatism. Specifically, in our model grades were nested within semesters, and semesters within students. Conservatism, a person characteristic, was included as a fixed effect predictor. Time (semesters) was used as a repeated measure and modeled via linear, quadratic and cubic terms as recent research has documented that semester GPA changes in a nonlinear fashion (Grove & Wasserman, 2004). To explore whether the relationship between grades and conservatism changed over time, all interactions between conservatism and the three terms representing the influence of time were also included (as well as our control variables). This model was used to model HA grades and, separately, HE grades.
A first glance at the results in Table 2 shows that HA grades were markedly higher than HE grades, a pattern corroborated by other analyses on the present data set, suggesting that, on average, HE courses are more rigorous than HA courses (Grove & Wasserman, 2004). More importantly, Table 2 shows no appreciable relationship between students' political orientation and their success in HA courses. Given the nonsignificant interaction with time, this null effect seemed to persist for students' entire college career. In stark contrast, we did find the predicted association between conservatism and HE grades. Although not large in size, the nonsignificant interaction terms with time suggest that the association between conservatism and higher grades in HE grades holds across the entire four years of college. These data strongly suggest that, as predicted, in HE courses academic success depends on students' levels of conservatism, whereas this is not the case for academic success in HA courses.
While the above analyses provide evidence that conservatism is linked to better academic outcomes in HE classes, from a social dominance perspective it is important to demonstrate that HE institutions produce a tangible advantage for individuals with HE beliefs that persists after graduation. Institutions of higher education can only be said to contribute to the maintenance of social hierarchy, if it is plausible that they facilitate the ascent of their HE graduates into positions of power outside of college.
Applied to grades within the context of a university setting, this results in the prediction that, at least with regard to HE fields of study, individuals with HE beliefs should graduate from the institution with a higher GPA than individuals who do not endorse HE beliefs. This has implications for students' cumulative HE grades or HE GPA. Because the differential influence of conservatism should be most apparent in students who were continuously exposed to the same academic environment, we focus here on students who took HE courses during four subsequent years as well as students who took HA courses during four subsequent years. (We chose years rather than semester as units of time because there were too few students who had taken the same type of course during every semester of their college career).
We used a multilevel model similar to the one above, except that we generated a cumulative HE GPA and a cumulative HA GPA separately for each of the four years of the students' tenure at the university. Years were treated as repeated measures factor nested within student, choosing a heterogeneous autocorrelation matrix as error structure to represent the correlation nature of the subsequent cumulative GPAs. Our predictors were time (years), represented through three random effect terms, as well as conservatism and its interactions with time. Table 3 shows again a positive relationship between conservatism and HE grades, but no association at all between conservatism and HA grades. Interestingly, the influence of conservatism is malleable over time as shown by the three significant interaction terms.. Conservatism's net influence on HE grades during the first year (b = .121) drops in the second year (b = .073), only to rise again in the third year (b = .147), after which it decreases again in the fourth year (b = .091). While we cannot offer a cogent explanation for this specific pattern, it is important to note that during every year examined, the coefficient is reliably greater than 0. However, there is no evidence of a compounding advantage for high-conservatism individuals, as this would have suggested a steady increase in the coefficient over time.
Strict test of person-environment effects
A strict test of person-environment effects requires it to be shown that the same person characteristic plays out differently in different environments. Hence, we focused on students who took both HE and HA courses during the same academic year, thus keeping students constant across HE and HA environments. If the same students' level of conservatism can be shown to differentially influence HE grades and HA grades, it is clear that differential student outcomes are not entirely the product of self-selection, but the joint product of individual and contextual characteristics. Focusing on students who, within any given year of the study, took at least one HE course and one HA course, we generated a multilevel model in which grades were nested within years, years within courses, and courses within students. We included course type (HA vs. HE), conservatism, and time (linear, quadratic and cubic terms) as effect predictors, as well as all necessary interactions to examine whether the influence of conservatism varied as a function of course type and time. (As before, a compound symmetrical heterogeneous error structure was used to accommodate the correlated nature of observations over time).
Table 4 summarizes the results of this analysis. The central finding was a significant conservatism by course type interaction, which showed that the same individuals' HE grades were related to conservatism, whereas their HA grades were not. This interaction varied over time, as evidence by the significant interactions with the three time effects. We followed up by computing the simple effects for HA and HE grades within each of the four years of the study. As seen at the bottom of Table 4, except for Year 2 conservatives received significantly higher HE grades than liberal students, whereas conservatism was never associated with academic success in HA courses. Incidentally, the number of students taking both types of courses within a year decreased from Year 1 to Year 4 (left column). This trend reflects that students sort themselves into different majors, making it less likely that a student with an HE major will take an HA course, and vice versa.
Yet, as mentioned above, from a social dominance perspective it must be shown that a fit between HE beliefs and HE environments has the potential of producing an enduring advantage for HE individuals within HE fields. Hence, we focused on individuals who took both HE courses and HA courses over a period of more than one year. Analyses of this group are highly diagnostic of the differential impact of conservatism because they allow us to keep students constant across environments. However, there is a trade-off. Given the dynamics of college, students sort themselves into majors with most of their course-taking occurring in that discipline and only a minority of students can be expected to continue taking course in HE and HA disciplines. Indeed, Table 4 (right column) shows that only a little over 100 students took both HE and HA courses during both of their first years in college, only 22 students maintained this pattern into their third year, and only 9 students took both HE and HA courses during every single year of their college career. Because a smaller sample implies a loss in statistical power, a priori the chances of finding support for our hypotheses were slim. However, if we find still find supportive evidence, one has to conclude that conservatism's effect on HE grades is quite powerful.
In the present multilevel analysis, we analyzed cumulative GPAs across a period of four years. We used the same three-level model described above, in which years were nested within courses, and courses within students. To capture the compounding nature of conservatism's influence on cumulative GPA, we restricted ourselves to the 104 students who had taken both HE and HA courses at least during the first two years of the study.
Despite the comparatively small sample, a significant course type by conservatism interaction yielded unequivocal support for the idea of a differential association between conservatism and grades (see Table 4, right column). Simple effect analyses illustrate that this differential association grew over time. (Note that, although none of the terms involving time reached significance, linear, quadratic and cubic terms for time were still included to allow for a direct comparison of parameter estimates with those of the model for non-cumulative grades, Table 4 left column.) Specifically, over the course of four years of taking both HE and HA courses, the coefficient for conservatism's association with HE grades exploded. To illustrate the magnitude of this difference, consider that at the end of college, a self-described conservative student (conservatism rating = 4) had a HE GPA that was by roughly 1.5 points or one and a half letter grades higher than that of a self-described liberal student (conservatism rating = 2). At the same time, there was no statistically reliable difference with regard to HA GPA. In sum, our data demonstrate that conservatism does play a role in HE environments, but not in HA environments, even when respondents are kept constant across environments.
Discussion
The present study provides a re-examination of the role of HE beliefs on academic success. To the extent that institutions of higher education are involved in the reproduction of social hierarchy, we predicted that HE environments within academia would serve a gatekeeper function by differentially bestowing benefits on individuals with matching HE, but producing a relative disadvantage in individuals with incongruent HA beliefs. Our results strongly supported this idea. First, we conceptually replicated pervious findings showing that students do sort themselves into academic environments that match their political orientation. Students in HE courses tended to be more conservative than students in HA course as well as in control courses. This finding corroborates previous findings on the role of self-selection in creating a congruence between individual HE/HA beliefs and HE/HA environments in a college setting (Van Laar et al., 1999; Sidanius et al., 1991, 2003).
Second, especially when exploring cumulative grades obtained in HE courses, we found a positive association between conservatism, an important HE belief, and academic success. This finding is consistent with the literature on person-environment fit, which has generally demonstrated that performance is enhanced when the characteristics of individuals are compatible with those of their environment (e.g., Chatman, 1991; O'Reilly, Chatman, & Caldwell, 1991). Moreover, this result provides support for a prediction by social dominance theorists (Van Laar et al., 1999) that HE educational institutions are involved in the reproduction of a group-based hierarchical social system by producing an advantage for those individuals who are most likely to keep that very system in place.
By contrast, there was no evidence that political orientation played any significant role in shaping academic success in HA environments. In other words, whether a student's beliefs matched or mismatched those of the HA environment had no implications for their HA grades. This finding suggests that HA environments might be more accepting of a broad range of student perspectives whereas HE environments appear to be more sensitive to whether student perspectives are compatible with those of the academic discipline.6 More importantly, this pattern conforms to our expectation that, because of the different implications for the continued existence of the hierarchy itself, HE environments should be more prone to create an advantage for HE individuals than HA environments for HA individuals.
This pattern, however, diverges from the one found by Van Laar et al. (1999) and Sidanius et al. (2003). Recall that these authors identified a reliable negative correlation between grades and HE beliefs for HA environments, but not for HE environments. In other words, these authors found that HA environments inhibited the success of HE students but not the success of HA students, with no similar dynamic occurring in HE environments. Although their data do highlight the role of HE beliefs in shaping academic success, we are somewhat skeptical with regard to the inference they allow concerning the hypothesized person-environment interaction. Specifically, students' overall GPA does not exclusively reflect academic outcomes in their major, but includes outcomes in a variety of other, non-major related classes, which may have contaminated the findings reported by Van Laar et al. (1999) and Sidanius et al. (2003). Our study avoided this ambiguity by separating HA grades from HE grades, which provided for an assessment of academic outcomes that was specific to a particular type of environment. Hence, we argue that our data more accurately capture the relationship between HE beliefs and outcomes in different academic environments.7, 8
Further, the unique structure of college, along with the availability of simultaneous outcomes specific to HE and HA environments, allowed a more detailed analysis of the role of self-selection and other causal mechanism in creating person-environment effects. Previous research has typically argued that person-environment congruency is primarily created through processes of self-selection and attrition (cf. Holland, 1966, 1985; Schneider, 1987; Schneider, Smith, & Goldstein, 2000). Very much in keeping with this conclusion, Sidanius et al. (2003) tested all four causal mechanism proposed by Van Laar et al. (1999) (i.e. institutional selection, institutional socialization, self-selection, and differential success), but only found self-selection to contribute to a match between HE/HA persons and HE/HA environments. We consider it likely, however, that this investigation did not support the differential success explanation because self-selection masked the evidence available to Sidanius et al. (2003).
To separate out existing effects of self-selection from the differential responses of HE/HA environments to individuals with HE beliefs, we focused on students who took both HE course and HA courses within the same period of time. Because none of these students differentially self-selected to HE or HA environments, they represent the strictest test case for how academic environments "respond" to students with specific characteristics. Indeed, using multilevel regression analysis, we found that a conservative political orientation benefited students when they took HE courses, but remained without consequences when they attended HA courses. This pattern unambiguously supports the notion that individuals are more successful in environments in which they fit and, thus, the differential success explanation put forth by Van Laar et al. (1999).
Admittedly, one might wonder what kind of students decide to repeatedly take both HE and HA courses. Even though they represent a theoretically interesting case, they may be somewhat atypical with regard to the remainder of the student population. For instance, because students often adjust their course-taking pattern if they are not successful in a particular discipline, one would expect self-described liberals to disproportionately avoid taking HE course, as they receive considerably lower grades than their conservative counterparts (see Table 4, right column). However, this is not supported by the data. If anything, the average level of conservatism seems to decrease among those individuals who continue taking both HE and HA courses over successive years (Year 1 M = 2.92, n = 408; Year 2 M = 2.88, n = 104; Year 3 M = 2.73, n = 22, Year 4 M = 2.44, n = 9). Thus, future research will need to explore further, why within this group no differential attrition of liberal students occurs even though they suffer a notable academic disadvantage.
What processes promote the success of conservative students in HE environments and what processes hinder the success of liberal students in the same environments? The results observed in this study are consistent with both SDT as well as the literature on the person-environment fit; yet, both literatures are somewhat agnostic with regard to the specific intra-psychic, interpersonal and organizational processes involved in creating the observed relationship between HE beliefs and success in HE environments. For instance, it is possible that HE students exert more effort in HE classes than HA students. HE students are more likely to aspire to HE careers (Pratto et al., 1997; Sidanius, et al., 1996, 2003); hence, they might be more motivated to study hard as they consider the course materials more relevant for their future than HA students. Likewise, because they feel more at ease with the materials and the style in which it is presented, it is possible that HE students are more intrinsically engaged, spending more time and effect on exams and assignment. For instance, because HE courses primarily focus on economic matter, broadly speaking, it is possible that conservative students are more comfortable with the subject because making money is more likely to be a personal goal for them than for liberal students (e.g., Hicks, 1974).
Another possibility concerns the presence of bias. Unless there is no ambiguity with regard to the standards of evaluation (e.g., through the use of a multiple choice format), grading student work often entails a subjective judgment on the part of the instructor. This subjective element of the grading processes can make grades vulnerable to biases (e.g., Lavin, 1965; Stockard, Lang & Wood, 1985). Indeed, sometimes instructors use grades to reward or punish student behavior that is not immediately related to academic performance. Farkas, Grobe et al. (1990) documented that middle-school teachers tended to assign higher grades to students whom they perceived to have good classroom behavior and good study habits-an effect that occurred even after objective levels of coursework mastery were statistically controlled (see also Farkas, Sheehan & Grobe, 1990). Thus, it seems possible that HE instructors assign lower grades to a liberal student than a conservative student given that faculty in HE departments tend to be more conservative than their counterparts in HA departments (e.g., Klein & Western, in press; Lipset 1982).
Whereas such blatantly unequal treatment is a possibility, we find it highly implausible. In order to be able to discriminate against liberals, instructors of HE courses would need to possess sufficient knowledge about the political views of individual students. Yet, the size of large college classes is often not conducive to instructors being familiar with more than a handful of their students. Further, in large classes grading procedures are often highly standardized, e.g., through the use multiple choice tests, leaving little room for evaluative biases. One could argue, however, that any interpersonal discrimination should be evident in settings in which student-instructor interaction is more intense and less impersonal. Thus, to the extent that an evaluative bias exists, it might be more apparent in smaller classes than in larger classes. When we re-analyzed our data comparing the effects of political orientation on grades obtained in smaller class (<50 students) with those obtained in classes with large enrollment („50 students), no differences in the nature or strength of effects were apparent. Therefore, it is extremely unlikely that HE instructors exhibit any conscious or unconscious grading bias against left-leaning students.
But even if grades are assigned without bias and reflect the quality of students' academic work, biases might be more subtle. Instructors might, perhaps inadvertently, treat conservative and liberal students differently by creating a classroom climate that is more encouraging of the former than the latter group, and indirectly encourage better performance (e.g., Harris & Rosenthal, 1985). Likewise, teaching methods and classroom structure might be more amenable to conservative than liberal students, e.g., by emphasizing competition over cooperation (cf. Kelley & Stahelski, 1970; Kuhlman & Marshello, 1975). While these issues are beyond the scope of the present paper, future research should address the nature of the processes by which HE students receive higher grades in HE environments.
Although the mediating mechanisms are not yet clear, the present data do show that the advantage gained by HE students in HE environments persists all the way through the four years of college. In other words, students with HE beliefs are likely to graduate with higher HE GPAs than their fellow students with HA beliefs who may have taken the same classes and who entered college with the same SAT scores. Although the effect after four years is still small (see Table 3, top panel), in a competitive market for HE jobs, even a small advantage might translate into conservative students being able to secure on average more attractive positions. In this regard it is somewhat surprising that we found little evidence of a compounding effect of academic advantage for HE students in HE courses. For students who consistently took HE course during four successive years, the correlation between political orientation and HE grades fluctuated somewhat, but remained more or less the same. If HE courses generate an advantage for HE students at every step of the way, one should have expected coefficients to gradually increase (e.g., Abelson, 1985; Martell, Lane, & Emrich, 1996; Weesie & Wippler, 1987). Note, however, that such a compounding effect was indeed found for students who consistently took both HE and HA courses at least two years, with the clearest evidence emerging for students who took did so for all four years of college. That is, even though this (small) group of students may be atypical of the student population as a whole, it most clearly illustrates that (dis)advantage adds up! But again future research is needed to explore why a compounding effect was found for one, but not the other group of students taking HE courses.
In conclusion, we believe that the present research makes an important contribution to understanding the role of political orientation in shaping success in different environments. Our study confirms a social dominance approach by extending and complementing earlier research in higher education conducted under this perspective (Van Laar et al., 1999; Sidanius et al., 2003). Specifically, we demonstrate how higher education creates a differential advantage for individuals who can be expected to maintain a system of group-based social hierarchy. In this sense, this research demonstrates that the Ivory tower is by no means distant from society, as is often alleged; rather, it is deeply entrenched in maintaining its current hierarchical structure (cf. Jackman & Muha, 1984).
Authors' note

The data used in the present article were collected under a grant by the Andrew W. Mellon Foundation to Nancy Cantor. We are indebted to Nancy Cantor, Heather Gillespie, Andy Karpinski and two reviewers for their comments on earlier drafts of this paper. Address correspondence to Markus Kemmelmeier, Interdisciplinary Ph.D. Program in Social Psychology, Department of Sociology/300, University of Nevada, Reno, Nevada 89557 USA. Electronic mail may be sent to markusk@unr.edu.
References
Abelson, R. P. (1985). A variance explanation paradox: When a little is a lot. Psychological Bulletin, 97, 129-133.
Abrami, P. C., & Mizener, D. A. (1985). Student/instructor attitude similarity, student ratings, and course performance. Journal of Educational Psychology, 77, 693-702.
Astin, A. W. (1993). What matters in college? Four critical years revisited. San Francisco, CA: Jossey-Bass.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury, CA: Sage.
Carlson, H. M., & Sutton, M. S. (1974). The development of attitudes as a function of police roles. Personality and Social Psychology Bulletin, 1, 113-115.
Carpenter, S. (2000). Effects of cultural tightness and collectivism on self-concept and causal attributions. Cross-Cultural Research, 34, 38-56.
Chan, D. K.-S., Gelfand, M. J., Triandis, H. C., & Tzeng, O. (1996). Tightness-looseness revisited: Some preliminary analyses in Japan and the United States. International Journal of Psychology, 31, 1-12.
Chatman, J. (1991). Matching people and organizations: Selection and socialization in public accounting firms. Administrative Science Quarterly, 36, 459-484.
Cookson, P., & Persell, C. (1985). Preparing for power. New York: Basic Books.
Dahrendorf, R. (1959). Class and conflict in industrial society. Stanford, CA: Stanford University Press.
Dey, E. L. (1996). Undergraduate political attitudes: An examination of peer, faculty, and social influences. Research in Higher Education, 37, 535-554.
Domhoff, G. W. (1983). Who Rules America now? A view from the Eighties. Englewood Cliffs, NJ: Prentice-Hall.
Domhoff, G. W. (2001). Who Rules America? Power and Politics (4th ed.). New York: McGraw-Hill.
Farkas, G., Grobe, R. P., Sheehan, D., & Shuan, Y. (1990). Cultural resources and school success: Gender, ethnicity, and poverty groups within an urban school district. American Sociological Review, 55, 127-142.
Farkas, G., Sheehan, D., & Grobe, R. P. (1990). Coursework mastery and school success: Gender, ethnicity, and poverty groups within an urban school district. American Educational Research Journal, 27, 807-827.
Federico, C., & Sidanius, J. (2002). Racism, ideology, and affirmative action: The antecedents and consequences of "principled objections" to affirmative action. Journal of Personality and Social Psychology, 82, 488-502.
Goodman, S. A.., & Svyantek, D. J. (1999). Person-organization fit and contextual performance: Do shared values matter? Journal of Vocational Behavior, 55, 254-275.
Gottfredson, G. D., & Holland, J. L. (1990). A longitudinal test of the influence of congruence: Job satisfaction, competency utilization, and counterproductive behavior. Journal of Counseling Psychology, 37, 389-398.
Grove, W. A., & Wasserman, T. (2004). The life-cycle pattern of collegiate GPA: Longitudinal cohort analysis and grade inflation. Journal of Economic Education, 35, 162-174.
Guimond, S. (1999). Attitude change during college: Normative or informational social influence? Social Psychology of Education, 2, 237-261.
Guimond, S., & Palmer, D. L. (1990). Type of academic training and causal attributions for social problems. European Journal of Social Psychology, 20, 61-75.
Guimond, S., & Palmer, D. L. (1996). Liberal reformers or militant radicals: What are the effects of education in the social sciences? Social Psychology of Education, 1, 95-115.
Guimond, S., Begin, G., & Palmer, D. L. (1989). Education and causal attributions: The development of "person-blame" and "system-blame" ideology. Social Psychology Quarterly, 52, 126-140.
Harris, M. J., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects: 31 meta-analyses. Psychological Bulletin, 97, 363-386.
Heck, R. H., & Thomas, S. L. (2000). An introduction to multilevel modeling techniques. Mahwah, NJ: Erlbaum.
Hicks, J. M. (1974). Conservative voting and personality. Social Behavior and Personality, 2, 43-49.
Holland, J. L. (1966). The psychology of vocational choice: A theory of personality types ad model environments. Waltham, MA: Blaisdell.
Holland, J. L. (1985). Making vocational choices: A theory of vocational personalities and work environments. Englewood Cliffs, NJ: Prentice-Hall.
Jackman, M. R., & Muha, M. J. (1984). Education and intergroup attitudes: Moral enlightenment, superficial democratic commitment, or ideological refinement? American Sociological Review, 49, 751-769.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychological Bulletin, 129, 339-375.
Kelley, H. H., & Stahelski, A. J. (1970). Social interaction basis of cooperators' and competitors' beliefs about others. Journal of Personality and Social Psychology, 16, 66-91.
Klein, D. B., & Western, A. (in press). How many Democrats per Republican at UC-Berkeley and Stanford? Voter registration data across 23 academic departments. Academic Questions.
Kuhlman, D. M., & Marshello, A. (1975). Individual differences in the game motives of own, relative, and joint gain. Journal of Research in Personality, 9, 240-251.
Lavin, D.E. (1965). The prediction of academic performance; a theoretical analysis and review of research. New York: Russell Sage Foundation.
Leitner, D. W., & Sedlacek, W. E. (1976). Characteristics of successful campus police officers. Journal of College Student Personnel, 17, 304-308.
Lipset, S. M. (1982). The academic mind at the top: The political behavior and values of faculty elites. Public Opinion Quarterly, 46, 143-168.
Martell, R. F., Lane, D. M., & Emrich, C. (1996). Male-female differences: A computer simulation. American Psychologist, 51, 157-158.
Michels, R. (1962). Political parties: A sociological study of the oligarchical tendencies of modern democracy. New York: Free Press. [Original published 1911]
Mills, C. W. (1956). The power elite. New York: Oxford University Press.
Mosca, G. (1939). The ruling class: Elements of political science. New York: McGraw-Hill. [Original published 1896]
Mount, M. K., & Muchinsky, P. M. (1978). Person-environment congruence and employee job satisfaction: A test of Holland's theory. Journal of Vocational Behavior, 13, 84-100.
Nafziger, D. H., Holland, J. L., & Gottfredson, G. D. (1975). Students-college congruence as a predictor of satisfaction. Journal of Counseling Psychology, 22, 132-139.
Noru_is, M. J. (2003). SPSS 12: Statistical procedures companion. Upper Saddle River, NJ: Prentice Hall.
O'Reilly, C., & Chatman, J. (1996). Culture as social control: Corporations, cults, and commitment. Research in Organizational Behavior, 18, 157-200.
O'Reilly, C., Chatman, J., & Caldwell, D. F. (1991). People and organizational culture: A profile comparison approach to assessing person-organization fit. Academy of Management Journal, 34, 487-516.
Pareto, V. (1979). The rise and fall of the elites. New York: Arno. [Original published 1901]
Pratto, F., Liu, J. H., Levin, S., Sidanius, S., Shih, M., Bachrach, H., & Hegarty, P. (2000). Social dominance orientation and the legitimization of inequality across cultures. Journal of Cross-Cultural Psychology, 31, 369-409.
Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social dominance orientation: A personality variable predicting social and political attitudes. Journal of Personality and Social Psychology, 67, 741-763.
Pratto, F., Stallworth, L. M., & Sidanius, J. (1997). The gender gap: Differences in political attitudes and social dominance orientation. British Journal of Social Psychology, 36, 49-68.
Pratto, F., Stallworth, L. M., Sidanius, J., & Siers, B. (1997). The gender gap in occupational role attainment: A social dominance approach. Journal of Personality and Social Psychology, 72, 37-53.
Schneider, B. (1987). The people make the place. Personnel Psychology, 40, 437-454.
Schneider, B., Smith, D. B, & Goldstein, H. W. (2000). Attraction-selection-attrition: Toward a person-environment psychology of organizations. In W. B. Walsh, K. H. Craik, & R. H. Price (Eds.), Person-environment psychology: New directions and perspectives (2nd ed., pp. 61-85). Mahwah, NJ: Erlbaum.
Sidanius, J. & Pratto, F. (1999) Social dominance: an intergroup theory of social hierarchy and oppression. New York: Cambridge.
Sidanius, J., & Pratto, F. (1993). Racism and support of free-market capitalism: A cross-cultural analysis. Political Psychology, 14, 381-401.
Sidanius, J., Liu, J. H., Pratto, F., & Shaw, J. (1994). Social dominance orientation, hierarchy-attenuators and hierarchy-enhancers: Social dominance theory and the criminal justice system. Journal of Applied Social Psychology, 24, 338-366.
Sidanius, J., Pratto, F., & Bobo, L. (1996). Racism, conservatism, affirmative action, and intellectual sophistication: A matter of principled conservatism or group dominance. Journal of Personality and Social Psychology, 70, 476-490.
Sidanius, J., Pratto, F., & Brief, D. (1995). Group dominance and the political psychology of gender: A cross-cultural comparison. Political Psychology, 16, 381-396.
Sidanius, J., Pratto, F., & Rabinowitz, J. (1994). Gender, ethnic status, ingroup attachment and social dominance orientation. Journal of Cross-Cultural Psychology, 25, 194-216.
Sidanius, J., Pratto, F., Martin, M., & Stallworth, L. M. (1991). Consensual racism and career track: Some implications of social dominance theory. Political Psychology, 12, 691-721.
Sidanius, J., Pratto, F., Sinclair, S., & van Laar, C. (1996). Mother Theresa meets Genghis Khan: The dialectics of hierarchy-enhancing and hierarchy-attenuating career choices. Social Justices Research, 9, 145-170.
Sidanius, J., van Laar, C., Levin, S., & Sinclair, S. (2003). Social hierarchy maintenance and assortment into social roles: A social dominance perspective. Group Processes and Intergroup Relations, 6, 333-352.
Stockard, J., Lang, D., & Wood, J. W. (1985). Academic merit, status variables, and students' grades. Journal of Research and Development in Education, 18, 12-20.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In W. G. Austin, & S. Worchel (Eds.), The social psychology of intergroup relations (2nd ed., pp. 33-48). Chicago: Nelson-Hall.
Teahan, J. E. (1975). A longitudinal study of attitude shifts among Black and White police officers. Journal of Social Issues, 31, 47-56.
Van den Berghe, P. (1978). Man in society: A biosocial view. New York: Elsevier North Holland.
Van Laar, C., Sidanius, J., Rabinowitz, J. L., & Sinclair, S. (1999). The three Rs of academic achievement: Reading, 'riting, and racism. Personality and Social Psychology Bulletin, 25, 139-151.
Walsh, W., B., Craik, K. H., & Price, R. H. (2000). Person-environment psychology: A summary and commentary. In W. B. Walsh, K. H. Craik, & R. H. Price (Eds.), Person-environment psychology: New directions and perspectives (pp. 297-326). Mahwah, NJ: Erlbaum.
Weesie, J., & Wippler, R. (1987). The cumulation of incompetence in organizations. Journal of Mathematical Sociology, 12, 347-382.
Endnotes

1 Some might argued that law schools should be included in this list. However, the status of law schools as hierarchy-enhancing institutions is unclear as many law graduates pursue careers within hierarchy-attenuating environments (see Sidanius et al., 1991).
2 Effect sizes reported by Sidanius et al. (2003) even suggests that there was a trend in the opposite direction, with students in HE majors and HA majors becoming more similar as educational tenure increased (see their Table 2).
3 Van Laar et al. (1999) also controlled for student rank and political conservatism. Inclusion of the latter variable reflected an attempt to separate the unique effects of one HE belief (racism) from the effects of another HE belief (conservatism).
4 As pointed out by one reviewer, the use of rather extreme labels as response scale anchors may have truncated variance on this scale, implying that the regression coefficients for conservatism reported in the present study may underestimate their true parameters.
5 "Course" refers here to the distinct course number listed in the university's course catalog, and our records reflected a student had taken course x during semester y. However, any information about different sections of the same course or different instructors was not available to us.
6 A potential alternative interpretation of this asymmetry is that HE students take a wider range of courses than HA students who are more likely to self-select into course environments with which they are compatible. This explanation may appear plausible in light of the fact that as part of their core curriculum many American universities require undergraduate students to take ethnic diversity courses or social science courses, which are frequently taught by HA departments. Such requirements may expose HE students to HA environments, but not HA students to HE environments, thus disproportionately increasing variability in course choice among the former group, while allowing narrower choices on the part of the latter group. However, this is not the case. For instance, among students who took both HA and HE courses during at least one year of the study (n = 857, see Table 4, left column), left-wing and liberal students were slightly more likely to take HE courses than HA courses (53% vs. 47%) during the course of the study. This distribution was more skewed for right-wing and conservative students (58% vs. 42%, respectively), indicating less, not more variability in the course choices of HE students.
7 Note that the difference between Van Laar et al. (1999) and Sidanius et al. (2003), on the one hand, and our study, on the other hand, may also be attributed to the use of different HE beliefs. While these earlier investigations focused on racism and generalized anti-egalitarianism, we focused on political orientation. However, from a SDT perspective, all three measures represent HE beliefs, that can be expected to affect academic outcomes in a similar way.
8 Our overall findings seem to conflict with earlier reports of a generally negative correlation between college GPA and conservatism (Sidanius et al., 1991). Such a finding may be an artifact resulting from a neglect of the fact that students are nested within courses that vary dramatically with regard to the average GPA that students receive. Given that conservative students self-select into more rigorous HE courses, their overall GPA can be expected to be lower than those of their liberal fellow students. Indeed, analyses on the present data confirmed a negative relationship between GPA and conservatism when the data were not treated as nested (b = -.030, SE = .003). However, in the context of the appropriate multilevel analysis, this association disappeared (b = .004, SE = .003).
Table 1
Multilevel regression model predicting course takers' level of conservatism

___________________________________________________________________________
Estimate SE
____________________________________________________________________________
HA courses vs. HE courses -.350 *** .054
HA courses vs. controls .016 .029
HE courses vs. controls .366 *** .047
Time (within course type)
HA courses -.005 .005
HE -.024 ** .008
Controls .007 *** .002
____________________________________________________________________________
**p < .01; ***p < .01

Note: Time refers to the eight semesters included in the present study. Parameter estimates for control variables (race/ethnicity, parents' education, parental income, scholastic aptitude, gender, and English as first language) are not reported here.
Table 2
Multilevel regression model predicting academic success for HA grades and HE grades

___________________________________________________________________________
Dependent Variable
__________________________________________
HA grades HE grades
____________________ ____________________
Estimate SE Estimate SE
____________________________________________________________________________
Intercept (average grade) 3.400 *** .024 3.104 *** .049
Time(linear) .042 + .024 -.341 *** .038
Time(quadratic) .029 *** .008 .090 *** .011
Time(cubic) -.001 *** .001 -.006 *** .001
Conservatism -.010 .026 .150 ** .050
Conservatism x Time(linear) .004 .030 -.075 .046
Conservatism x Time(quadratic) .001 .010 .020 .014
Conservatism x Time(cubic) .000 .001 -.002 .001
____________________ ____________________
No. of students in analysis n = 2761 n = 1782
____________________________________________________________________________
+p < .10; **p < .01; ***p < .001

Note: Conservatism was centered at the grand mean for each analysis. Time refers to the eight semesters included in the present study, with the first semester coded as 0. Parameter estimates for control variables are not reported here.
Table 3
Multilevel regression model predicting cumulative HA grades and HE grades for individuals who took HA courses or HE courses during four subsequent years

___________________________________________________________________________
Dependent Variable
_____________________________________________
Cumulative HA Grades Cumulative HE Grades
____________________ ____________________
Estimate SE Estimate SE
____________________________________________________________________________
Intercept (average grade) 3.484 *** .039 3.403 *** .057
Time(linear) -.087 * .036 -.208 *** .035
Time(quadratic) .092 *** .026 .072 ** .026
Time(cubic) -.018 ** .006 -.006 .006
Conservatism .007 .040 .121 ** .047
Conservatism x Time(linear) .014 .042 -.109 * .044
Conservatism x Time(quadratic) -.010 .031 .075 * .033
Conservatism x Time(cubic) .003 .007 -.014 * .007
____________________ ____________________
No. of students in analysis n = 263 n = 364
____________________________________________________________________________
**p < .01; ***p < .001

Note: Conservatism was centered at the grand mean for each analysis. Time refers to the four years included in the present study, with Year 1 coded as 0. Parameter estimates for control variables are not reported here.
Table 4

Multilevel regression models predicting cumulative academic success for students who enrolled in both HE and HA courses
______________________________________________________________________________

Grades per year Cumulative grades
___________________ _____________________
Estimates SE Estimates SE
______________________________________________________________________________

Intercept (avg. HA courses) 3.435 *** .046 3.595 *** .093

HE courses -.496 *** .039 -.440 *** .063
Conservatism -.014 .048 -.027 .086
Time (linear) -.102 .130 -.104 .146
Time (quadratic) .157 .118 .143 .146
Time (cubic) -.039 .026 -.036 .035

HE courses x Time(linear) -.084 .148 -.141 .246
HE courses x Time(quadratic) .011 .133 .065 .253
HE courses x Time(cubic) .013 .030 -.001 .062

Cons. x Time(linear) .035 .169 -.053 .207
Cons. x Time(quadratic) -.002 ** .151 .112 .217
Cons. x Time(cubic) -.003 ** .033 -.032 .053

Conservatism x HE .145 ** .050 .217 * .084

Cons. x HE x Time(linear) -.416 * .193 -.036 .354
Cons. x HE x Time(quadratic) .364 * .172 .027 .380
Cons. x HE x Time(cubic) -.075 * .038 .016 .095

No. of students in analysis n = 857 n = 104

Simple effects of Conservatism

Year 1
HA grades -.014 .050 -.027 .086
HE grades .130 ** .048 .190 + .113
n = 408 n = 104
Year 2
HA grades .017 .050 .001 .079
HE grades .035 .044 .225 * .098
n = 349 n = 104

(Table 4 continued)

Year 3
HA grades .028 .052 .061 .096
HE grades .197 *** .048 .446 ** .152
n = 229 n = 22
Year 4
HA grades .005 .061 -.035 .160
HE grades .150 ** .052 .762 * .273
n = 220 n = 9
______________________________________________________________________________
***p < .001; **p < .01; *p < .05; +p < .10

Note: HA courses served as reference category. Conservatism was centered at the grand mean for each analysis. Time refers to the four years included in the present study, with Year 1 used as reference category. Parameter estimates for control variables are not reported here. Ns next to simple effects indicate the number of students included in the analysis who took both HA courses and HE courses that year.