What Do Think Tanks Think? Proximity to Power and Foreign Policy Preferences


 Through the use of survey methods, the study presents the first systematic comparison of America-based international relations professors to think tank employees (TTEs) in terms of their preferred conduct of the United States in international affairs. The difference between the two groups in their support for military intervention is stark. TTEs are 0.47 standard deviations more hawkish than professors based on a standard measure of militant internationalism (MI). Controlling for self-described ideology mitigates this effect although it remains statistically significant. Beyond quantifying their relative foreign policy preferences, this study helps to resolve why TTEs tend to assume more hawkish policies. The authors find evidence that hawkishness is associated with proximity to power. Professors who have worked for the federal government score higher on MI, as do TTEs based at institutions located closer to Capitol Hill. In general, the results point to a self-selection mechanism whereby those who favor interventionist policies are more likely to pursue positions to increase their policy influence, perhaps because they know that powerful institutions are more likely to hire hawks. Alternative explanations for differences, such as levels or kinds of foreign policy expertise, have weaker empirical support.

In September 2002 a group of foreign policy experts took out an advertisement in the New York Times warning against the Iraq War. The letter was signed by 36 scholars in the field, representing a wide range of the political spectrum. Later research would reveal that these individuals were not unrepresentative of their field, as the vast majority of IR scholars did oppose the Iraq War. Yet op-eds published in the New York Times, the Wall Street Journal, and the Washington Post in the run up to the conflict overwhelmingly favored intervention (Long et al. 2013), and support for the Iraq War reached highs of over 70% among the general public during that time period (Rosentiel 2008). Despite theories of public opinion that stress the importance of political and intellectual elites (Zaller 1992), it appears that views of IR scholars had little impact on what was perhaps the most important American foreign policy decision since the end of the Cold War. As a result of the Iraq War and other recent experiences, many scholars have self-consciously sought to bring their voices into the policy making arena (Jentleson and Ratner 2011).
Despite such efforts, most work done by academics specializing in IR continues to have little relevance for policy makers (Goldgeier and Jentleson 2015). Yet just because government officials tend not to consider the views of academics, it does not mean they do not rely on any form of expertise. Think tanks serve as an "important link" in the "long haul from academic theories to actual public policies." (Self 1993:155) Yet while there have been studies looking at the opinions of those who work in universities and even some of government officials, this class of experts has remained neglected by those studying foreign policy elites. Unlike professors, who devote most of their professional lives to teaching and scholarship, think tank researchers work full-time on influencing public knowledge-based. In other words, TTEs might be more hawkish because people with more hawkish views self-select into such jobs, because such institutions either socialize their employees or select people with certain views and not others, or because the nature of the research being done pushes individuals in a more interventionist direction.
Although our research design does not allow us to assess with certainty the relative impact of each of these factors, some of our results provide certain clues. Except in the case of views towards Iran, regional expertise is not associated with more or less hawkishness, counting against a knowledge-based explanation. In contrast, we do find some evidence consistent with the theories of self-selection and institution-selection, although more research needs to be done to determine which model more accurately describes the world.
This study hopes to inspire more research into not only the role of think tanks, but the ways in which research on international relations translates into consequences for American policy. Through a combination of tracing the ways in which influential networks, donors, lobbyists, experts, and pundits interact with powerful officials, and quantitative research into preferences and policies, we can hope to gain a more complete understanding of what determines how the United States perceives and addresses problems abroad. Few questions can be as important for scholars of international relations.

I. The "Missing Link" between Expertise and Policy
Scholars of American foreign policy have for decades been interested in the preferences of elites, recognizing the important role they play in shaping public opinion and making policy. Rather than being autonomous, a wide range of literature finds that public opinion can be influenced by the mass media in the form of agenda-setting, priming, and framing effects (Baum 2005: 32-52;Iyengar and Simon 1993;Baum and Potter 2008). Elites can also be defined at the level of decision-makers themselves. Thus, because of their influence on mass opinion and ability to implement policies directly, what this group actually thinks about the issues of the day is fundamental for understanding how the US interacts with the rest of the world.
Pioneering works by Holsti andRosenau (1988, 1990) showed that in the 1970s and 80s, opinion leaders could be classified based on their levels of support for militant internationalism and cooperative internationalism, forming a 2x2 matrix that corresponded to categories that they called hard-liners, internationalists, isolationists, and accommodationists. Militant internationalists see a conflictual world in which the United States should be prepared to use military force in order to defend itself and protect its interests (Holsti 1999:17-20). Cooperative internationalism, in contrast, emphasizes cooperation, participation in international institutions, and engagement towards solving problems afflicting the developing world such as human rights and poverty. Huntington (2004) used survey data on issues related to trade, immigration, and foreign policy preferences and found across a wide range of issues a contrast between an internationalist elite and a nationalistic public, arguing that the former is out of touch with the latter. In more recent years, much of the research on foreign policy elites has been conducted by the Teaching, Research, and International Policy (TRIP) project at the College of William & Mary, which surveys university professors in the field of international relations on a regular basis (Green & Hale 2017;Zarakol 2017). In 2014, TRIP expanded its faculty survey to cover thirty-two countries, leading to insights into elite opinion and perceptions of the field outside the United States (Wemheuer-Vogelaar 2016).
Despite previous research on professors and government officials, there has yet to be a study of the foreign policy opinions of think tanks. This is surprising, given the prominence these organizations have in the American system. The history of some of the most influential contemporary think tanks can be traced to the first decades of the twentieth century. During the Progressive Era, a small number of philanthropists provided the money to create the "Russel Sage Foundation (1907) (Abelson 2018:22-23) The influence of these institutions took off as the US created a permanent security establishment to confront the Soviet Union in the decades after the Second World War.
The RAND Corporation "played a central part in the formulation of US security orthodoxy of the 1950s and 1960s," with a relationship with the Air Force that gave it direct bureaucratic influence (Higgott and Stone 1994:25). Other important think tanks have been partisan or ideological in nature, tending to affiliate with one of the two major parties. In 1960, the Brookings Institution provided a library and meeting rooms for the Kennedy transition team (Smith 1993:129-31), and the Economist would refer to Brookings researchers as "experts on tap" for the administration. Jimmy Carter reached out to Brookings for policy advice in 1975, forming close relationships with many of its fellows (Abelson 1996:10). President Nixon's relationship with Brookings was as hostile as Kennedy's was warm; he put the organization on his famous enemies list and suggested that its building be fire bombed in order to steal files (Troy 2017). On the other side of the aisle, by one count nearly 200 employees of the top conservative think tanks served as government officials or consultants in the Reagan administration (Abelson 1996:15). The 1980 Heritage document "'Mandate for Leadership I' became a bible of sorts for many in the Reagan White House," with the organization at one point bragging that 60 percent of the proposals put forth were adopted or in the process of being adopted (Shapiro 1984).
The power of think tanks in the American system has been credited to the divided nature of our political system, with federal authority shared between the presidency, Congress, and government agencies, along with the weakness of the political parties as cohesive units able to control their members (Smith 1989:80;Weaver 1989:570;Polsby 1983:58;Weiss 1992:6-8). Think tanks have as a result been called "the intellectual epicenter of the policy planning process," as institutions that "provide the research and early policy proposals that eventually find a home on Capitol Hill." (Keskingören and Halpern 2005:104) It is now completely unremarkable for important foreign policy figures to be employed by a think tank in between stints in the government, creating a "revolving door" between the expert and policy worlds (Fuelner 1985:24;Haas 2002:7;Higgott and Stone 1994:33;McGann 2016:114 (Stone 2013:207).
Relative to work that investigates the executive, the legislature, public opinion, and interest groups, think tanks have been understudied given the prominent role they play in the American political system (Abelson 1996:ix, 1, 119). According to Rich that to a large extent shapes that shapes public opinion, lobbies American officials, and even staffs the government (Acharya 2016;Haas 2002:8).

II. The Ideologies of Think Tanks
There is widespread acknowledgment that think tanks employees and professors specializing in international relations differ in their focus, purposes, and methodology.
Scholars have thought about these differences, often focusing on the perceived shortcomings of the research of one side or the other. It is common to fault universities for their irrelevance to modern issues, or, alternatively, to argue that think tanks are fundamentally political entities that use methods that are less than optimized for finding truth (Kristof 2014;Gallucci 2014). While there may be nothing problematic about different kinds of institutions seeking different goals, survey results show a widespread desire among the academy for the research of professors to be useful to government officials and the general public (Nye 2009). In one study, 40% of international relations professors thought that work in their field had no impact on public policy or popular discourse, the results revealing a "frustrating lack of influence." (Yester 2014) This perception is confirmed by a survey of policymakers, who generally have little use for much of the research conducted by professors and are more likely to rely on newspapers than academic work to understand foreign countries (Avey and Desch 2014).
Recognizing the desire of many professors to influence American foreign policy, the organization "Bridging the Gap" has emerged in recent years with the goal of connecting Yet for all this discussion about bringing academics into the political process, there has been little thought given to how this would actually change American foreign policy. As already noted, just because most professors play little role in policy-making or public discourse about foreign policy does not mean that there are no experts with influence. Think-tanks are in a sense already "bridging the gap" between the theoretical world of scholarship and foreign policy. Studying what think tanks actually think, compared to professors, is useful in not only helping us understand the role of a unique group of scholars with an outsized influence on public policy debates, but also to give us some idea of how American foreign policy would change if the power of think tanks to affect public discourse was reduced relative to that of professors.
Thus far, there is only anecdotal evidence speaking to differences between the opinions of TTEs and other foreign policy experts, most of which suggests that the former is relatively hawkish. Think tanks are said to be part of the "blob" that makes up the foreign policy community, a term that refers to experts with an ideological or material interest in calling for more interventionist policies abroad (Walt 2018:ch. 3;Porter 2018).
If this is true, the leftward tilt of academia must be part of the story as to why (Langbert, Quain, and Klein 2016). While most social science fields see Democrats overrepresented relative to Republicans, in Washington over the long term the two parties are nearly equal in terms of political power, meaning that we can expect the supply of conservative scholars and officials to meet the demand created by Republican politicians and administrations. This alone should make TTEs more hawkish, yet it remains an open question whether the gap remains once political ideology is accounted for.
Again, anecdotal evidence suggests that it should. During the 2016 presidential campaign 122 self-described members of the "Republican national security community" denounced the foreign policy positions of then-candidate Trump for being inconsistent with their view of a United States that is committed to its defense alliances and engaged with the rest of the world (Open Letter 2016). After Trump bombed Syria in 2017, CNN analyst Fareed Zakaria announced that as a result "Donald Trump became president of the United States," a view that was seen as representative of bipartisan praise for the attack and mocked by non-interventionists (Gillespie 2018;Swanson 2017). When Barack Obama faced criticisms from left-leaning think tanks such as Brookings during his own administration, it tended to be for taking less militarily aggressive positions, including his signing off on the Iran nuclear deal and unwillingness to fully commit to supporting the Free Syrian Army (Maloney 2015;Doran and Shaikh 2013). Despite belonging to two different parties, then, the last two American presidents both tended to face criticism from think tanks and foreign policy elites on their own side mainly for their less hawkish policies. All of this implies that even when ideology is controlled for, think tank employees should still be more hawkish than IR professors.
The discussion above does not only make predictions about the views of think tanks, but also implies that hawkishness is related to proximity to power. In addition to think tank employment status, proximity to power can be operationalized in two ways: as past or present government employment status or physical closeness to power. In this paper, we use Capitol Hill to represent the center of power, and measure logged distance from that location as our proxy for spatial distance. Being close to powerful individuals increases opportunities for face-to-face interactions, known to facilitate socialization (Kelley 2004:430;Checkel 2005:807-08). These kinds of contacts can take the form of, among other things, panel discussions, interviews with the media, and access to social, business, and networking opportunities with influential figures. Those closer to the center of power are more likely to be part of the foreign policy community (Walt 2018). We do not expect to see a relationship between distance and political preferences within the category of professors, whose job description does not necessarily involve influencing public opinion, being close to media centers, and meeting with powerful figures.
Why should proximity to power, particularly in the form of think tank employment relative to working at a university, predict greater hawkishness? We put forth three, non-mutually exclusive, potential explanations, which can be referred to as self-selection, institution-selection, and knowledge-based. First of all, people who favor more hawkish positions might be more likely to seek out positions of influence and power. Second, institutions and governments might seek out those with more hawkish views, or perhaps pressure them into supporting a more aggressive posture for the United States abroad. In the words of Walt (2018:112), "[t]he busier the U.S. government is abroad, the more jobs there will be for foreign policy experts," and institutions can either select people who will advocate for the US doing more overseas from the beginning or socialize people into supporting the same. Note that one should not simply assume that think tanks are the only institutions capable of selecting based on political views or socializing their employees, as those who favor more interventionism may be less welcome at universities or pushed in a more dovish direction once they get there. Finally, the nature of the work and the focus of their research might encourage TTEs, who put more effort into studying contemporary and policy-relevant issues (Stone 2013:15;McNutt and Marchildon 2009:222), to adopt more hawkish views. Professors and TTEs research different things, and it may be the case that what information they seek out and find is driving differences in foreign policy preferences.
Our data do not allow us to put forth reasonable estimates about the relative effects of self-selection, institution-selection, and knowledge. That can only be done by gathering panel data that looks at what kinds of jobs foreign policy experts with different views seek out or how views change over time depending on whether they end up at think tanks or universities. Nonetheless, we do ask a few questions that can provide insight into different theories and hopefully guide future research. First of all, in our survey we ask how important it is for individuals to see their research have influence in the real world.
If TTEs are higher on this measure than professors, then our data is consistent with a theory in which, among scholars seeking a career in foreign policy, there is a positive relationship between greater hawkishness and interest in changing the world, and individuals who want to change the world disproportionately seek to work in think tanks rather than universities. If we do not find TTEs higher on the desired influence measure, then our data would imply a model in which people's desire to make a difference does not influence career choice as much. Perhaps there is no relationship between foreign policy preferences and the type of employment desired on the part of aspiring scholars, but institutions themselves seek certain candidates but not others based on ideology or policy positions.
It must be noted that the finding that TTEs have more desire to change the world would also be also consistent with an institution-selection effect if we assume that there is no connection between foreign policy views and employment preferences at the beginning of people's careers, but that after being selected into one institution or another, the degree to which they desire to influence the real world is shaped in a way to justify previous life decisions, in a process analogous to what has been called "choice-supportive bias." (Mather, Shafir, and Johnson 2000;Wee, Ta, and Cheok 1995). In other worlds, TTEs and professors do not start off differing in how much they seek influence, but become different based on what jobs they end up getting. While this strikes us as a plausible model of the world, we believe that, if we find evidence that TTEs are higher on desired influence, accepting that as supporting the self-selection effect makes the simplest assumptions and is more consistent with Occam's Razor. To more conclusively show a self-selection effect, however, one would need to have panel data on the preferences of foreign policy experts with regards to political views, desire for influence, and employment hopes, from both before they go on the job market and after they begin their careers.
In order to test the knowledge-based explanation, we ask respondents what regions of the world they consider themselves experts in. If greater expertise in an area predicts more hawkishness on relevant issues, and especially if think tank status is no longer predictive once expert status is controlled for, then we have strong evidence for the knowledge-based view. Conversely, if regional expertise is not predictive of foreign policy preferences regarding a given area of the world, then it is less likely that research focus is driving many of the differences we see between TTEs and professors. This is not a perfect test of the knowledge-based view, as regional expertise is determined by selfdescription, and two people may claim to be experts in a region and nonetheless focus on different things, with, for example, one studying the distant past and another spending more time on contemporary issues. With these caveats in mind, the effects of regional expertise on foreign policy preferences can at the very least provide hints about the merits of the knowledge-based explanation for differing views between TTEs and professors.

Finally, our study also asks about moral values regarding what determines when and
under what conditions individuals support the use of force. If moral values rather than regional expertise drive perceptions of threat and willingness to use force against specific adversaries, then we have evidence against the knowledge-based explanation of the divergence in views based on employment status.

Methodology and Results
Unlike the concepts of professor or government official, the question of what makes an institution a think tank can be difficult to pin down (McGann 2016:9;Abelson 2018: 8;Medvetz 2008:9;Smith 1993:xiv;Stone 2013:9). We sidestep the definitional debate about what constitutes think tanks by restricting our sample to what have been referred to as "quintessential," premier," "prominent," or "elite" think tanks (Abelson 2018:8-17, 50). According to Simon (1993:42), a think-tank is "an independent organization engaged in multi-disciplinary research intended to influence public policy." This definition is sufficient for our purposes, as our concern is the most influential think tanks, and not smaller institutions that may arguably not qualify as think tanks under a more precise definition. These organizations share many of the same characteristics, being 501(c)3 institutions that aim at influencing public opinion through means such as appearing on radio and television, providing congressional testimony, and publishing policy relevant literature including books, reports, and op-eds.
For this project, we worked through the firm Qualtrics to contact all specialists in international relations whose e-mails could be found who were listed as professors at R1 or R2 universities or think tank fellows or policy analysts. E-mails were sent out between October 2018 and January 2019. Our list was composed of individuals from the following 20 American think tanks, chosen by the authors as institutions that are particularly influential: the American Enterprise Institute, the Atlantic Council, the Brookings Institute, the CATO Institute, the Carnegie Endowment for International Peace, the

Center for a New American Security (CNAS), the Center for Strategic and International
Studies, the Council on Foreign Relations, the Institute for the Study of War, Freedom House, the Heritage Foundation, the Hoover Institution, the Hudson Institute, Human Rights Watch, the Foundation for Defense of Democracies, the New America Foundation, the RAND Corporation, the Stimson Center, the Washington Institute for Near East Policy (WINE), and the Woodrow Wilson International Center. Follow up emails were sent to those who did not respond, with the compensation being increased if needed. Professors were offered either no money or $10, while think tank researchers were offered $10, $25, or $30. In total, we reached out to 823 think-tank analysts or fellows and 868 professors. Respondents were given 3 questions that measured their support for militant internationalism (MI), 3 that measure cooperative internationalism (CI), and 2 that measure isolationism, taken from Kertzer et al. (2014). Among other questions, they were also asked about current and past employment status, demographic information, political ideology, and their views on contemporary political questions. See the Online Appendix for exact question wording and ordering. Table 1 shows the response rate for each category as well as basic demographic information. Response rates are within the normal range for e-mail surveys (Sheehan 2001). Note that in Table 1, all numbers are calculated based on the list from which each name was obtained, while in the rest of the paper people are classified as belonging to one or another category based on self-reported data. 2 As a general rule, after Table 1  professors. The extent to which gaps in political ideology and affiliation account for any differences in foreign policy view is tested throughout this paper. The first cube in Figure   3 below shows standardized scores on militant internationalism, cooperative internationalism, and isolationism for respondents in each category. The second cube gives insight into scores for employees from the 13 think tanks from which three or more individuals responded. employment status cannot be used to calculate response rates as, by definition, we do not have self-reported data from those who did not respond.  We thus find strong support of the idea that TTEs support a more hawkish foreign policy, even when ideology is accounted for, and more engagement with the world generally. To test the idea that spatial proximity is important, we use geo-coded data based on the location of each respondent and the Haversine formula to calculate the distance from Capitol Hill of each TTE and professor filling out the survey (Picard 2012). " = 2 * ,sin 1 2 " − 5 2 6 + ( 5 )cos ( " ) sin 1 > " − 5 2 @A In this case, " is the distance between Capitol Hill and individual k when that individual completed the survey, r is the radius of the earth, and are the latitudes and longitudes of the two points being compared, with the coordinates of Capitol Hill being location l.
We exclude respondents for whom d > 750, on the grounds that the effect of distance should not hold once one is far enough from DC, and to prevent the inclusion of individuals who may be on vacation or working overseas. We find a negative relationship between logged distance to Capitol Hill and militant internationalism among TTEs (p < .05). No such relationship is found among professors. Figure 1    Another measure of proximity to power we use is experience in government, including the military. In order to test whether government employment is correlated with more hawkish view, Table 2 shows the results of four regressions, with MI as the dependent variable and government service and ideology as the independent variables for each group of experts. Professors who are or have been employed by the federal government are nearly half a standard deviation higher on militant internationalism than their counterparts who have never worked in government. The results are mostly unchanged if we control for ideology. On the other hand, there is no relationship between having worked in the government and support for intervention among TTEs, likely because this population is already quite hawkish, thus restricting the range (Sackett and Yang 2000). Sample size is also an issue, as only 71 individuals in total reported ever working for the federal government.
Turning to specific issues, Figure 4 shows support for various statements related to preferences over American foreign policy. The text in the figure is a shortened version of each question, the full wording of the questions themselves can be found in the Online Appendix. The results are shown for a t-test involving whether the individual is a TTE or professor, and also a regression in which ideology is controlled for. The lines show 95% confidence intervals for standardized differences between the two groups. A positive value indicates support for a more militarily assertive foreign policy, whether in the form directly supporting the use of force or some other action. On almost all issues, TTEs are more hawkish, with the effect tending to be reduced substantially when ideology is controlled for. Nonetheless, even in the latter case the differences remain and the effects are in the expected direction. Note that Figure 4 only tells us about differences between the two groups; it does not tell us anything about how many people support any given proposition in absolute terms. We therefore discuss some of those findings here.
TTEs are much more interventionist regarding the Middle East, and this manifests itself in support for both past wars and aggressive action today. Those from think tanks are more likely to still approve of the initial invasion of Iraq and the 2011 bombing of Libya: 13.8% and 42.5% respectively, compared to 5.3% and 25.2% for professors. They were also more likely to support funding Syrian rebels battling the Assad government, Thus, we find evidence across the board showing that, aside from a few select issues, TTEs are more hawkish. The differences in willingness to endorse abstract statements expressing more support for military action translates into support for a more aggressive posture abroad across a wide range of specific issues. This leaves the question of why. In order to gain some understanding of the mechanisms that lead to the differences described, we asked individuals how important it is for their "ideas to influence policy in the real world" on a five-point scale from "not at all important" to "very important." For professors, the average score was 2.8, while the average for TTEs was approximately 4.0, a difference of over one standard deviation (p < .001). There is no statistical relationship between ideology and desire for influence, nor is there one between desire for influence and militant internationalism when employment status is controlled for (p = 0.41). Figure 5 shows desire for influence for four groups depending on current employment status and whether they have ever worked for the federal government. Figure 5. Desire for ideas to be influential by whether TTE or professor, and whether individual ever worked in government. Horizontal line indicates average among respondents.
Note that among professors, those who have worked in government are much more interested in their research influencing the world. There is less of a difference based on government employment for TTEs, but that would be expected as the average desired influence score for that group is high regardless. We thus find some evidence of a selfselection effect, where individuals who want more influence go into think tanks and government instead of universities. As mentioned above, however, we cannot rule out a model in which individuals are selected into jobs based on other criteria and then justify their career outcomes by updating their values.
In addition to asking about specific issues, we asked respondents how important seven different factors were in determining their support for the use of force on a sixpoint Likert scale from "not at all important" (1) to "very important" (6). These questions asked about whether the action in question "is moral"(Moral); "makes the US safer" (Safer); "brings the US respect as a great power" (Great Power Respect); "makes the US wealthier"; "is consistent with international law" (International Law); "makes the US stronger than other countries in the long run" (Relative Strength), "improves the lives of people in the country being attacked" (Improve Lives); and "makes the US respected as a moral leader in the world." We conducted a regression analysis with the answer to each question as the dependent variable, with employment status and the average answer across all questions in the section as control variables. The latter variable corrects for the tendency of some individuals to use the scale more differently than others; controlling for the average response allows insight into how relatively important individuals think that each factor should be. We find that TTEs care more about safety to the US (β = 0.36 SD, p < .01), and less about whether an intervention improves the lives of people abroad (β = -0.25 SD, p < .05). The differences lose their statistical significance when ideology is controlled for, although the coefficients remain in the expected directions.
We also ask respondents whether they considered themselves experts in various regions of the world, which TTEs were much more likely to do. For example, while only 7% of professors called themselves experts in China, the number was 24% for TTEs. As TTEs are more likely to self-identify as experts, a knowledge-based theory of the difference between TTEs and professors would predict that regional expertise would predict more hawkish views. It is also possible that moral values determine these outcomes of interest more than knowledge. In order to test these competing view, we a  It is important to note that our measure of expertise is based on self-identification.
A more objective measure of expertise is language skill. 3 We therefore asked respondents professors. Going through a similar educational experience does not seem to make TTEs adopt views that are less hawkish and therefore make them more similar to their counterparts who go into academia.

Conclusion
In recent years, it has become normal to despair at the lack of influence that international relations scholars have over the policy making process (Yester 2014). Yet few can doubt the influence of think tanks, as these institutions, and the networks they facilitate, function as shapers of public opinion and launching pads for careers in government. This paper is the first to back up anecdotal claims about their foreign policy preferences relative to other experts in the field with survey data. It also fits the finding into a larger story in which proximity to power is the key variable that predicts hawkishness. Attempts to bring academics into the policy-making arena would likely reduce the influence of think tanks, and ultimately lead to less interventionist American policies abroad.
Our results generally argue against an interpretation of our results that suggests that TTEs are more hawkish because they have different kinds of knowledge about the issues in question. Regional experts are if anything less hawkish towards China and North Korea, and while Middle East experts are more hawkish towards Iran, such individuals are much less likely than experts in other regions to speak the relevant language. Thus, the results argue for some kind of selection mechanism in which people with hawkish views are more likely to find their way into think tanks. Whether most of that selection is due to the kinds of individuals who seek out such jobs or because of institutions, whether universities or think tanks, selecting based on ideology, we cannot say for sure. Future research should seek to distinguish the hypotheses of self-selection and institution-selection. The two processes are not mutually exclusive; in fact they are likely to complement one another. If think tanks are more hostile to individuals with noninterventionist views, such people might be less likely to seek jobs at such institutions.
Nonetheless, the degree to which think tanks-and universities for that matter-select for individuals with certain kinds of views and not others can and should be studied further.
More work should also be done quantifying the impact of think tanks. Individual studies have documented, for example, the relationship between Brookings and the Kennedy administration (Smith 1993); Heritage and the Reagan administration (Abelson 2018), and RAND and the military bureaucracy (Higgott and Stone 1994). These studies focus on networks involving think tanks and some group of people in power. Another way to understand influence over the federal government, however, is to divide lobbies by the issues that they advocate for. Those advocating for a certain industry or foreign government are likely to form networks that expand across think tanks and other powerful institutions (Milner and Tingley 2015;Newhouse 2009). Research on domestic politics and foreign policy should seek to understand the value of think tanks as coordination mechanisms of those seeking certain political goals.
Finally, this paper raises deep questions about the nature of expertise in the modern world. It is quite striking that, even among those with similar levels of education and who spend their time researching similar issues, moral values regarding the US and its place in the world not only drive opinions on specific issues but even perceptions of threat (see Hanania and Trager 2019). While the desire to bring more expertise into the policy making arena can be perceived as a matter of simply introducing technocratic knowledge into public discussions, at its heart the project is political in nature, as one group of experts can only gain influence at the expense of others. Whether for good or ill, increasing the power of the academy relative to that of think tanks would likely lead American foreign policy in a less interventionist direction.