Editorial Board:
Editor:
George H. Conklin,
 North Carolina
 Central University

Board:
Rebecca Adams,
 UNC-Greensboro

Bob Davis,
 North Carolina
 Agricultural and
 Technical State
 University

Catherine Harris,
 Wake Forest
 University

Ella Keller,
 Fayetteville
 State University

Ken Land,
 Duke University

Miles Simpson,
 North Carolina
 Central University

Ron Wimberley,
 N.C. State University

Robert Wortham,
 North Carolina
 Central University

Editorial Assistants

John W.M. Russell,
 Technical
 Consultant

Austin W. Ashe,
 Duke University

® 

Volume 8, Number 2

Fall/Winter 2010
 

Political Ideology and Perceptions of Bias 
Among University Faculty

by

Susan Bullers,

Melissa Reece,

and 

Christy Skinner

University of North Carolina Wilmington

Background

    Political bias in U.S. Colleges and Universities has been a recurring topic of interest and empirical study for decades. In the 1960's and 1970's researchers were interested in the negative career effects of liberal political activism. More recent debates over the preponderance of liberal professors are motivated by concerns that students are not getting a balanced perspective in their course content and/or that faculty career trajectories are negatively affected by personal political beliefs that differ from the institutional norm.

    Findings from the 1960's to the present have found the following trends; in aggregate, U.S. College and University faculty are more likely to tend toward democratic/liberal identification than republican/conservative identification and have higher rates of liberal/democratic identification than the general population (Bartlett 2005; Eitzen and Maranell 1968; Hamilton and Hargens 1993; Klein and Stern 2004; Lipset and Ladd 1974; and  Rothman, Lichter, and Nevitte 2005).  Faculty political views vary significantly by field of study with the social sciences tending toward liberal, the natural sciences tending toward "moderate," and business, engineering, and agriculture tending toward conservative (Eitzen and Maranell 1968; Faia 1974; Hamilton and Hargens 1993; Klein and Stern 2004; and Lipset and Ladd 1974); women and minorities tend to be more liberal than male and non-minority faculty (Faia 1974; Hamilton and Hargens 1993; and Lindholm, Astin, Sax, and Korn,2002); faculty at elite schools tend to be more liberal than those at non-elite schools (Eitzen and Maranell 1968; Hamilton and Hargens 1993; and Lipset and Ladd 1974); rates of publications and grants are higher among liberal faculty than among conservative faculty (Hamilton and Hargens, 1993, Lipset and Ladd 1974); and younger faculty tend to be more liberal than older faculty (Eitzen and Maranell, 1968; Faia, 1974; Klein and Stern 2004; and Lipset and Ladd 1974).  Findings also show that US southern faculty are more likely to be democrats than non-southern faculty, although democratic party identification in the U.S. south does not necessarily translate into a liberal political ideology (Eitzen and Maranell 1968).

    Although there are several differences in scope and methodology among these studies, the most fundamental question of this line of inquiry lies in the conceptualization of "bias."   Individuals' political opinions and beliefs vary considerably in an engaged intellectual environment (see Lipset and Ladd 1974). This is an integral part of scholarly inquiry and is the very topic of research among many social scientists.  Holding an opinion, either minority or majority, is generally upheld as an individual right in academia.  The more pertinent issue underlying these debates is whether or not institutional influence is exerted on the grounds of those opinions.  It is the perception of such influences that we are interested in exploring here. 

    Several divergent explanations have been offered for the higher rates of liberal faculty, as well as higher publication, grant, and elite school appointments among liberal faculty.  Conservatives argue that institutional biases block the career trajectories of faculty with conservative beliefs (Brooks 2003; Rothman, Lichter, and Nevitte 2005), while liberals argue that the knowledge gained in higher education leads one to a more liberal ideology (Bartlett 2005; Lipset and Ladd 1974). 

    Previous research has addressed this question by analyzing shifts in political views among faculty over time, with consideration given to the effects of long-term influence of institutions' normative political beliefs, aging of faculty, self-selection into academia of those with normative views, and discrimination against those with minority views in both promotion and publication opportunities. Although the findings regarding aggregate political shifts over time are conflicting, Hamilton and Hargens (1993) found a small shift toward conservatism between 1969 and 1984 and Rothman, Lichter and Nevitte (2005) found a shift to the left between 1984 and 1999.  Hamilton and Hargens' (1993) detailed study of political shifts indicates that between 1969 and 1984 shifts involved movement to adjacent categories in a five point scale (left, liberal, middle of the road, conservative, right) and that the majority of the movement has been away from the "middle of the road" category in both left and right directions.  The aggregate shift was toward the right but the degree of shift was negligible.

    Looking at retrospective changes in individual faculty political views, Faia (1974) found that political beliefs among faculty were formed in their undergraduate years and were relatively stable thereafter.  Faia suggests that continued institutional norms, beyond undergraduate education, had a minimal effect in changing personal political ideology among faculty. 

    Another facet of perceived bias concerns the social psychological aspects of in-group/out-group perceptions.  Perceptions regarding the attributions of motivations, size, tactics, power and influence of  one's own group as well as one's opposition are influenced by individuals' subjective, selective, and often skewed, perceptions.  The majority of recent literature regarding political bias in higher education emanates from agendas proclaiming that those with a particular political ideology are disenfranchised (Brooks 2003; Klein and Ster 2004; and Rothman, Lichter and Nevitte 2005).  However, most of the evidence presented does not speak directly to disenfranchisement but rather to unequal distributions of political ideologies among university faculty.  Although predominance of a particular view can lead to the disenfranchisement of those with minority views, any analysis of disenfranchisement needs to analyze the specific negative effects of holding minority views rather than establishing the fact that there is a majority ideology. 

    The current study uses data from a faculty survey to examine personal perceptions of political bias in a university.   Perceptions of dominant political ideologies and personal political party identification, as well as perceptions of the negative effects of one's political ideology on one's career trajectory will be examined. 

    The survey (Appendix A) was mailed via campus mail to the faculty at a Comprehensive University 1 level State University in the southeastern U.S. in the spring of 2004.  The population included 551 full-or part-time faculty members with teaching duties. Faculty members were asked to fill out the questionnaire and return it through campus mail to a dedicated mailbox.  Questionnaires included ID numbers that corresponded to names on a master list. The master list was kept in a password–protected file in the PI's office, and was destroyed when data collection and matching was complete, in accordance with the University's IRB protocol.  Two data files were constructed; one with all variables and one with identifying variables of age, race, and sex removed. These precautions assured confidentiality in that only the principle investigator had access to the data file that contained identifying variables.  This process allowed for case matching with additional public record information and prevented duplication of responses.  Responses were entered into an SPSS data file using only assigned ID numbers.  A website containing a brief description of the study, research methods, and confidentiality measures was posted and referred to in the cover letter.

    Data regarding tenure status, department, discipline, and salary were acquired from public records and entered into the data file by matching ID codes. 

    226 faculty members responded, resulting in a response rate of 37.4%.  This response rate is low but at or above response rates for surveys with similar content.   There are several cautionary notes associated with such a low response rate.  Because this is a controversial topic, there is no reason to believe that the non-responders are randomly distributed across the political spectrum.  Despite confidentiality measures, the matching process and linkages with public information may have created a hesitance for some to respond.  Although individuals may have been motivated to respond, or not, to increase or decrease counts for particular political views, the ID system used prevented multiple submissions.  Given these limitations and the purpose of the study, we will focus on the interrelationships of the variables rather than the distribution of the characteristics in the population.

    Exact variable wording can be found in Appendix A.  Political party affiliation was measured on a seven-point scale ranging from strong democrat to strong republican and a write-in "other" category, consistent with the General Social Surveys item format (NORC, 972-2004).  This variable was collapsed into a three-category variable, "political party" in which "strong democrat," and "not very strong democrat" were recoded as "democrat"; "independent, close to democrat," "independent," and "independent, close to republican" were coded as "independent;" and "not very strong republican" and "strong republican" were coded as "republican."  The "other" category was retained as in the original because there were few cases and many did not indicate any specific beliefs.

    Political ideology was measured on a seven-point scale ranging from "extremely liberal" to "extremely conservative," again, consistent with the General Social Surveys item format.  This variable was collapsed into a three-category variable, "political ideology" in which "extremely liberal," "liberal," and "slightly liberal" were recoded as liberal; "moderate" was retained as "moderate;" "slightly conservative," "conservative" and "extremely conservative" were recoded as "conservative." 

    Personal bias perception items included a checklist item, "negative effects," asking if one's political beliefs had negatively affected their career at any of the following levels; department, college, university, and field.  Another item, "attacks," asked if respondents they had ever suffered personal attacks or harassment from colleagues because of their political beliefs. Responses to this item were yes/no checks and a write-in response space was provided for number of incidences.  Productivity was measured with a write-in count of academic publications, or presentations if in the Fine Arts.   One item, "conceal" asked respondents about any felt need to conceal their political beliefs at any of the following levels; department, college, university, and field.  The "promotion" checklist item asked if respondents felt that their political beliefs had negatively affected their career during any of the following decision processes; hiring, third year review, tenure, promotion, contract renewal, distribution of resources, and acceptance of professional works. 

    General bias items included a checklist item, "faculty bias" asking if respondents felt there was a general bias among faculty at their institution against; conservatives, liberals, or a write-in category for "others." Respondents were also given a checklist item, "dominant faculty ideology," asking if they felt the predominant political ideology among faculty at their institution was conservative, liberal or a write-in category for "other."  Another item, "dominant student ideology," asked if respondents felt that the predominant political ideology among students at their institution was conservative, liberal, or a write-in category for "other."  Respondents were also asked if they thought that their political ideology was made evident to their students during their teaching ("ideology evident in teaching"). Response categories ranged from "definitely" to "definitely not."  Survey demographics included year of birth, recoded to "age;" year of hire at their institution, recoded to "years at Institution;" "sex;" and a write-in race/ethnicity item that was recoded as "minority" or "white."  Open-ended general comments were solicited as well.

    Variables collected and matched from the public records data included "tenure" status, which was recoded as tenured/not tenured and annual "salary" in dollars, including grant salary. The "discipline" variable was recoded into broader groups to maintain confidentiality among departments with low response counts.  The department recode criteria maintains the general political affiliation trends found in previous research; "Liberal Arts" includes the social sciences, humanities, library, and fine arts; "Business" includes all departments in the business school, "Science and Math" includes physical sciences, math and computer science; "Education" includes all programs in the school of education, and "Health" includes nursing, health, and physical education.  Note that this institution does not have a medical school; traditionally moderate in political views, or an engineering school; traditionally conservative in political views.

Preliminary analyses and response rates

    Because previous findings report a weaker correlation between Democratic Party identification and liberalism in the US south than in the rest of the country we conducted a preliminary analysis to determine the correspondence between political party identification and political ideology for our sample.   This analysis will also inform questions regarding the political party identification of those who self-identify as moderates on the political ideology scale.

    Next, response rates by tenure and discipline will be presented. A Chi-square analysis was performed on the response by tenure table.  Low cell counts in the response by discipline tables precluded meaningful statistical testing.  Response rates can only be compared by variables included in the public information data file so we are are not able to compute response rates by any of the political opinion variables. 

Political ideology factors

    Political ideology distribution by discipline, tenure and gender were analyzed with cross-tabulations and Chi-Square test of significance. Due to low cell counts the Chi-square test of significance was not valid for the political ideology by discipline table.  In order to compare responses regarding political ideology dominance to perceptions of political ideology bias, distributions of reported bias against, and dominance of, liberals and conservatives are both presented in cross-tabulations, by political ideology.  Reports of the likelihood that one's political ideology is evident in one's teaching were analyzed by discipline and political ideology.  Again, these analyses included cross-tabulations but had low cell counts that prevented statistical testing. 

Negative effects of political ideology

Reported negative effects of political views at various university levels, negative influences of political views at various career decision points, the felt need to conceal political beliefs, and attacks or harassment for political beliefs were all analyzed in cross-tabulation by political ideology.   Again, low cell counts precluded Chi-square analyses here. 

Multivariate Analyses

    Reported negative effects of political views at various university levels, negative influences of political views at various career decision points, the felt need to conceal political beliefs, and attacks or harassment for political beliefs were all analyzed in cross-tabulation by political ideology.   Again, low cell counts precluded Chi-square analyses here. 

    Previous research has shown that Democratic Party affiliation in the US south is less strongly associated with liberal political ideology than elsewhere.  However, in this sample 97% of Democrats identify as "liberal."  Party affiliation distributions show that the majority of liberals are democrats and the majority of moderates and independent but a slight majority of Conservatives are Independents  (Table 1). 

    The overall response rate was 37.4% with somewhat higher response rates among non-tenured faculty than tenured faculty (Table 2).  Full professors had the lowest response rates at 30.6%.  Response rates among the five disciplines ranges from 22% in education to about 40% for both science/math and liberal arts.

    The overall political ideology of responding faculty shows a liberal preponderance;  70.9% (Table 3). Political ideology generally followed previously reported trends in differences by discipline. Business tended toward conservative, liberal arts tended toward liberal. Health was split with a small preponderance of liberals. Education and science/math were more strongly liberal in this sample than in previously reported studies at other universities. This may be due, in part, to the fact that this university does not have a medical school or an engineering school, which typically report a more conservative political ideology.  Again, the distribution of self-report characteristics may be compromised by the response rate for this sample.

    Descriptive statistics show a lower percentage of liberals, than moderates or conservatives, without tenure, and a higher percent with tenure (Table 4). However, conservatives have a higher percent of respondents with full professorship than either of the other two ideology categories. 

A Chi-Square test of these data shows that the relationship between tenure and political ideology is not statistically significant (Chi square = 4.03, 4df, p=.402).  Political Ideology does differ significantly by gender, with women reporting higher rates of liberal ideology than do men (Chi square = 8.214, 2df, p=.016).

    Findings suggest that respondents did distinguish between political dominance and political bias.  Reports of a dominant liberal political ideology at the University were fairly consistent across all groups (55%-66.7%), with the higher rate reported by conservatives (Table 6).  Rates of political bias were much lower and differed substantially between conservatives and liberals.  Only 5.3% of respondents indicated a bias against liberals whereas 18.9% of respondents indicated a bias against conservatives.  Bias against liberals was reported by liberals (6.9%) and moderates (5%).  Bias against conservatives was reported by all groups with 48.7% of conservatives, 25% of moderates and 10.4% of liberals reporting such a bias.

    Trends in these findings suggest that Liberals are more likely to report that their ideology is evident in their teaching than are moderates and conservatives (Table 7).   Combining responses of "definitely" and "somewhat likely," liberals rank highest (56.3%) while moderates and conservatives are substantially less likely (27.8% and 25.7% respectively).    Evidence of political ideology in teaching also differs by discipline (Table 8).  Again combining the "definitely" and "somewhat likely" categories, ranking of positive response rates were as follows, liberal arts (60.4%), science/math (42.3%), health (38.5%), education (25%) and business (21%). 

    Responses regarding the overall negative effects of political views show a similar percent of reported effects across all groups but at different levels for each group (Table 9). The negative effects for liberals are more likely to be reported at all levels; department, college, university, and field, whereas moderates report negative effects most often at the field and to a lesser extent, at the department level. Conservatives report negative effects most often at the Department and College level, and to a lesser extent at the field level. Overall, 10.2% of faculty reported any negative effect of political ideology at any level. 

    Responses regarding perceptions of negative influences of one's political ideology in any career decision process were around 5% for both liberals and moderates but over 15% for conservatives (Table 10).   For liberals, these effects were most often reported at the "promotion" and "other" decision points.  Moderates reported negative effects exclusively at the "promotion" decision.  Conservatives reported these effects at almost all decision points but most strongly at the "other" and "resource distribution" decision points.  Although salary is affected by several factors, a post-hoc ANOVA test revealed no statistical difference in salary (including grant salary) among the three groups (F= .170, df=2,200, p=.844, ns). 

    33 percent of Conservatives reported a need to conceal their political beliefs at any level, while 20% of moderates and 18.8% of liberals reported a need to conceal their political ideology at any level (Table 11).  For liberals this need was reported most often at the department level, for moderates it was reported most often at the department and field levels, and conservatives reported the need to conceal their political ideology at all levels, especially at the university, department and college levels. 

    Rates of reported harassment/attack were relatively consistent across all groups (Table 12). Moderates reported the highest rate of harassment/attacks at (10%), followed by liberals (9.1%) and conservatives (7.9%). 

    Multivariate logistic regression analyses (Tables 14, 15 and 16) show that minority status is the only significant predictor of bias perceptions, for all three models.  Factors which approached significance include non-tenured faculty being more likely to report negative effects from their political views and full professors being less likely to feel the need to conceal their political views.  Post hoc analysis of minority status by political views (Table 17) showed that minorities were more likely that non-minority individuals to report conservative political views.  This finding is inconsistent with previous studies.