The recent study of academics' political ideologies by sociologists Neil Gross and Solon Simmons has been touted by the authors and by Inside Higher Education as showing that liberal dominance in academia is much less great than many believe.

However, this interpetation of the data is questionable. Gross and Simmons' Table 2 (pg. 28) shows that liberal dominance is overwhelming in the humanities and social sciences, the only two categories they list where ideology is actually likely to matter in influencing research agendas and classroom instruction:

Thus, Gross and Simmons' findings indicate that liberals outnumber conservatives by 11-1 among social scientists and 13-1 among humanities professors, with liberals forming a clear absolute majority in both fields. The somewhat less lopsided overall figure (about 4-1 liberal-conservative ratio) is reached only because of the relatively balanced nature of "health sciences," physical/bio sciences, comp sci/engineering, and business faculties. With the possible exception of business, all of the latter are fields where ideology makes little or no difference in either research or teaching. There are few meaningful professional differences between liberal computer scientists and conservative ones, or between liberal and conservative physicists. Moreover, Gross and Simmons' Table 12 (pg. 41) shows the results of a different survey question on which 24% of social scientists and 19% of humanities professors self-identified as "radical." This indicates that self-identified radicals (to say nothing of the left side of the political spectrum more generally) actually significantly outnumber conservatives in both fields.

Even in business classes, where ideology perhaps matters more than in the hard sciences, it probably matters less than in the social sciences and humanities. As I understand it (based admittedly on limited knowledge - I welcome correction from experts) a high percentage of business class instruction and business school professors' research focuses on relatively nonideological issues such as techniques for running a company, developing products, and marketing; only a minority of business courses focus on public policy issues with an ideological valence. Be that as it may, business professors surely have much less aggregate influence on both academic research and classroom instruction on politically charged issues than do humanities and social science scholars.

UPDATE: To avoid confusion, I should emphasize that this post takes Gross and Simmons' definition of "conservative," "liberal," and "moderate" as given. In the next post I challenge their analysis of moderation. There is therefore no contradiction between this post and the next one, although I admit I should have made the distinction between the two clear earlier.

UPDATE #2: Some commenters note that there is a large preponderance of liberals over conservatives in biological and physical sciences as well (almost 6-1). This is true. However, by Gross and Simmons' methodology, liberals in these fields are still outnumbered by moderates and conservatives (about 55-45). Thus, liberal dominance in these fields is not as clear as it is in the social sciences and humanities, where liberals outnumber conservatives by even larger margins and also constitute a clear absolute majority of the total.

Gross and Simmons' important new study of academic political ideology may underestimate the degree of liberal dominance because of the way it categorizes political "moderation" among academics. As discussed in my last post, the authors find that 43.5% of academics are liberal, 47% are "moderate," and 9% are conservative. This leads the authors to conclude that, while there are very few conservative academics, the overall valence of the academy is moderate rather than liberal.

One problem with this conclusion, discussed in my previous post, is that the preponderance of liberals is much greater in those fields where ideology actually matters. Another is Gross and Simmons' analysis of "moderation." As they explain, the "moderate" category in their Table 2 (reprinted in my earlier post) is actually a combination of survey respondents who described themselves as "slightly liberal" (18.1%), "middle of the road" (18.0%), or "slightly conservative" (10.5%). I wonder, however, whether these self-descriptions are based on a reference group of other academics (who are well to the left of the general population) or of the general public. Many people who do not follow survey research understandably define "moderation" relative to the orientations of the people they know. For academics, these reference groups are disproportionately likely to be other academics and nonacademics with ideological backgrounds similar to those of people in the academic world. The famous anecdote about the New York intellectual who couldn't believe that Nixon had won the 1972 election because no one he knew had voted Republican may be an exaggeration; but it does contain a kernel of truth. Thus, self-described "middle of the road" and "slightly liberal" academics - perhaps even "slightly conservative" ones - may be well to the left of center by the standards of the general population.

I cannot reliably prove or disprove this theory based on the data presented in the Gross and Simmons paper. But there are some indications that it captures an important part of what is going on. For example, Gross and Simmons found that 78% of their respondents voted for Kerry (77%) or Nader (1%) in the 2004 election, and only 21% for Bush (Bush won the popular vote by a narrow 51-48 margin in the general population). Assuming that most of the self-described conservatives (20 percent of the total sample, if you count the "slightly conservative") voted for Bush, this implies that nearly all of the self-described "slightly liberal" and "middle of the road" academics voted for Kerry. By contrast, CNN exit polls indicate that self-described "moderates" in the general population voted for Kerry by a much narrower 54-45 margin. While ideology is not the only influence on voting behavior, this result certainly suggests that self-described academic centrists are on average much further to the left than moderates in the general population.

UPDATE: I should note that while there is good reason to suspect that academic "moderates" overall are more liberal than those in the general population, it is impossible to tell from the Gross-Simmons paper how this breaks down in particular disciplines. For example, it is possible that self-described "middle of the road" academics in the hard sciences are more moderate than those in the social sciences and humanities.

UPDATE #2: It is worth pointing out that Gross and Simmons do not deny the fact that academics are more liberal than the general population. As they put it (pg. 72), "we would not contest the claim that professors are one of the most liberal occupational groups in American society, or that the professoriate is a Democratic stronghold." Their main original claims are that 1) academics are more moderate than usually assumed, and 2) there is more diversity of opinion among left of center academics than conservative critics claim. The first conclusion depends crucially on the authors' definition of moderation - the issue discussed in this post. The second may well be true. In fact, I suspect that it almost certainly is. There is likely considerable divergence between the roughly 20% of humanities and social science professors who describe themselves as "radical" (see my last post) and those who are mainstream liberals. However, this finding does not change the fact that academics are overwhelmingly on the left rather than the right. Political diversity among academics does exist, but much of it is confined within a truncated liberal to radical political spectrum.

In my last two posts, I put forward some reasons why Gross and Simmons' important new paper on academic ideology understates the prevalence of liberals in academia. It is only fair to also point out a way in which that study overstates that prevalence, or at least underestimates the proportion of non-liberal academics. It does so by collapsing academics' ideologies into three categories along a single continuum: "liberal," "conservative," and "moderate." Respondents to their ideology question had the option of describing themselves as "Very liberal," "liberal," "slightly liberal," "middle of the road," "slightly conservative," "conservative," or "very conservative."

Note that this one-dimensional ideological scale entirely ignores libertarians, who - roughly speaking - are "liberal" on social issues, and "conservative" on economic ones. Some libertarians may describe themselves as "conservative" on the Gross-Simmons scale. Others, however, might pick "liberal" or "middle of the road," or simply choose not to answer the question because they don'e see a choice they like. For example, if I average out my "liberal" positions on social issues with my "conservative" ones on economic issues, I could describe myself as "middle of the road" on average. But it's a very different kind of "moderation" from that associated with, say, DLC Democrats.

Ignoring libertarians may be defensible in studies of the general population, where they are relatively rare (although even among the general public, some evidence suggests that about 10 percent are closer to being libertarian than conservative, moderate, or liberal). It is much more problematic in a study of academics, where libertarians are a much larger fraction of the nonliberal total than in the general public. In my experience, about half of nonliberal/noncentrist law professors are in fact libertarians rather than social conservatives. Lawprofs are not included in the Gross-Simmons study. But economists and political scientists (two other groups with which I have some familiarity) are, and the libertarian-conservative ratio there does not seem to me much different than that in law. Even if we cautiously assume that libertarian academics are only half as common as conservative ones, the Gross-Simmons data imply that about 5% of academics are libertarians (vs. 9.4% conservative). And another 5% would be "slightly libertarian" (vs. 10.5% "slightly conservative").

How much does this skew Gross and Simmons' overall results? It is difficult to say. It all depends on how many libertarian academics would describe themselves as "conservative" or "very conservative" when they answered the author's one-dimensional ideology question and how many would describe themselves as "liberal," falling into one of the three categories the authors classify as "moderate," or simply refuse to answer the question. Given the deepening of the conservative-libertarian split during the Bush years, I suspect that the proportion of libertarians willing to embrace the "conservative" label has been declining; this trend is likely to be unusually strong among academics, most of whom follow politics closely. My best guess - and it's only a guess - is that about 50-70% of libertarians would refuse to embrace the two most "conservative" categories in the Gross-Simmons framework. Assuming that libertarian academics make up about 6-7% of the total (perhaps an underestimate), that implies that the true proportion of right of center academics is 12-13% rather than the 9% that the authors estimate. In some fields, such as economics and other social sciences, the proportion of libertarians among the nonliberals is likely to be significantly higher than that. If you count the putative "slightly libertarian" academics (parallels to the authors' "slightly liberal" and "slightly conservative" categories), the libertarian proportion would be about twice as high, perhaps 10-14% of the total sample.

In my judgment, properly accounting for libertarians would not overturn the conclusion that the left side of the political spectrum is overwhelmingly dominant in academia - especially when you consider the factors discussed in my previous two posts. It would, however, substantially increase the estimated proportion of academics who are neither liberal nor "moderate."

UPDATE: I was remiss in not mentioning this 2005 study of social scientists' political views by GMU economist Daniel Klein and Swedish scholar Charlotta Stern, which finds that "social scientists who deviate from left-wing views are as likely to be libertarian as conservative." This finding strengthens the case for including libertarianism as a separate category in studies of academic ideology.

Harvard economist Greg Mankiw notes the mounting evidence that conservatives are underrepresented in academia, and suggests a possible remedy (without necessarily endorsing it):

Question to think about: If right-wingers are underrepresented in universities relative to the population and discriminated against by the left-wing majority, as Larry [Summers] suggests, should there be affirmative action for right-leaning academics? It seems that, on principle, those on the left (who favor affirmative action to promote diversity and correct past injustice) should endorse such a university policy, and those on the right (who more often oppose affirmative action) would be against.

The underrepresentation of conservatives (and, I would add, libertarians) is almost certainly not all due to ideological discrimination. But evidence suggests that discrimination is probably at least a part of the story. In this excellent Econlog post, economist Bryan Caplan explained why ideological discrimination is more likely to flourish in academia than in most other employment markets. Even aside from discrimination, the ideological homogeneity of much of academia causes a variety of problems, such as reducing the diversity of ideas reflected in research, skewing teaching agendas, and generating the sorts of "groupthink" pathologies to which ideologically homogenous groups are prone.

However, whether or not the discrimination is the cause of the problem, affirmative action for conservative academics (or libertarian ones) is a poor solution. Among other things, it would require universities to define who counts as a "conservative" for affirmative action purpose, a task that they aren't likely to do well. Affirmative action for conservatives would also give job candidates an incentive to engage in deception about their views in the hopes of gaining professional advancement. Moreover, conservative professors hired on an affirmative basis despite inferior qualifications would find it difficult to get their ideas taken seriously by colleagues and students. They might therefore be unable to make a meaningful contribution to academic debate - the very reason why we want to promote ideological diversity in hiring to begin with.

A recent paper by Matthew Woessner and April Kelly-Woessner argues that much of the massive ideological imbalance in academia may be due to self-selection. Theis survey evidence shows that conservative undergraduates are less interested in doing original academic research and developing a "philosophy of life" than liberal ones, while showing greater interest than liberals in making money and raising a family. The authors claim that these differences in attitude are likely to lead liberals to self-select into academia and conservatives to self-select against it.

Woessner and Kelly-Woessner emphasize that these self-selection arguments are not incompatible with discrimination-based explanations. Indeed, one of their other interesting findings is that conservative undergraduates have, on average, weaker mentoring relationships with faculty members (who at most schools are overwhelmingly liberal) even after controlling for the students' academic records. Obviously, faculty mentoring at the undergrad level is often crucial for facilitating later efforts to get into a top grad school. Nonetheless, the authors argue that their attitudinal variables probably do account for a large portion of the ideological imbalance in academia.

I agree that self-selection probably plays an important role. It would be a serious mistake to attribute the ideological imbalance in academia solely to discrimination, or even primarily. But I am somewhat skeptical about the particular variables emphasized by the Woessners. If interest in making money were a crucial variable in steering conservatives away from academia, one would expect their representation to be much higher in high-paying academic disciplines such as law, where faculty members routinely make six figure salaries and often have extensive consulting opportunities. Yet the ideological imbalance in legal academia is very large and fairly similar to that in other academic fields.

In my view, a focus on raising a family should make academia more attractive to conservatives rather than less. Relative to other professional jobs, academic careers are actually quite family-friendly. Unlike most other professionals, professors have a high degree of control over their schedules. They can also do a much higher proportion of their work at home, which makes it easier to spend time with kids. Universities also tend to have extremely generous family leave policies for faculty. Moreover, universities often give substantial tuition discounts to children of their faculty - an important benefit for social conservatives with large families. Some schools even subsidize private secondary school tuition for faculty children.

I'm not saying that the academic life is a family idyll. But it's closer to being so than most of the available alternatives for ambitious undergrads. It's true that the interest in starting a family is negatively correlated with interest in pursuing a PhD in the authors' regression model. I suspect, however, that this is a statistical artifact stemming from the fact that those conservatives with the strongest interest in raising families are also more alienated from the dominant academic ideology than even other conservatives are (perhaps because they are more likely to be highly religious and belong to theologically traditionalist denominations).

On the other side of the ledger, I'm skeptical that wanting to develop "a meaningful philosophy of life" really has much to do with wanting to be an academic. And, in the authors' regression model (Appendix A), this indicator is only a weak (though statistically significant) predictor of interest in pursuing a PhD.

Like other studies of academic ideology, the Woessner and Kelly-Woessner paper also suffers from the failure to consider libertarians separately from conservatives. As I discuss in this post, libertarians are about 10-15 percent of the general population and are likely to be disproportionately represented among non-liberals likely to be interested in pursuing academic careers. Relative to conservatives, libertarians are about 20% more likely to be college graduates (see Table 10 in the linked paper) and threfore more likely to be potential candidates for academic jobs.

Although I'm not aware of survey evidence on this point, I strongly suspect that libertarians are closer to liberals than to conservatives in their interests in doing research, developing a philosophy of life, and raising families. Yet libertarians are almost as underrepresented in academia as conservatives are. Certainly, they are nowhere close to constituting 10 percent of faculty in any field other than economics. It is possible that libertarians are more interested in making money than liberals are; the claim is often made, though I have yet to see any systematic study that proves or disproves it. But even if this stereotype is true, it doesn't explain why they aren't better represented in law and other high-paying academic fields.

UPDATE: As I implied in the original post, I don't think that either the ideological imbalance in academia or the flaws of some of the Woessners' self-selection arguments prove that there is extensive ideological discrimination. Indeed, I think the underrepresentation of conservatives in academia is partially due to self-selection factors (though probably not the ones this paper focuses on). On the other hand, there is significant evidence that discrimination plays an important role as well. See this post for links.