In their reply, Professors Mearsheimer and Walt focus quite reasonably on my two main claims: that their research methods are flawed and that their evidence is weak. But they begin, tellingly, by citing a range of indirect evidence that appears to depict a powerful “Israel lobby.” Lots of knowledgeable Washington insiders, they say—policymakers, journalists, candidates for office, and the like—say and do things that seem to acknowledge the “lobby's” power. I draw attention to this opening for several reasons. First, it is not clear how much weight some of this evidence will bear. Take, for example, the National Journal survey of members of Congress that Mearsheimer and Walt cite twice in their reply (and once in their book). In this survey, conducted once, in 2005, seventy-three members of Congress (out of 535—less than 15 percent) responded to the question, “Which two interest groups do you believe are most effective on Capitol Hill?” Of these respondents, thirteen (less than 20 percent of the sample) mentioned American Israel Public Affairs Committee (AIPAC) as one of their two choices. It seems something of a leap from the observation that a dozen or so members of Congress said once that AIPAC is “effective” (which might mean any number of things) to the inference that AIPAC—or the lobby more generally—is powerful. Or take the observation that important politicians regularly address AIPAC's annual conference and make friendly speeches when they do. Surely these same politicians visit other such organizations regularly. And when they appear before, say, the AFL-CIO or the NAACP, surely they say nice things about the issues that these organizations and their conference attendees care about. Successful politicians rarely voice open disagreement with the people they are talking to. But are we to conclude from this behavior that the AFL-CIO and the NAACP, or any other groups that regularly receive such visits, are powerful? Again, this inference requires something of a logical leap.
Second, this opening suggests a starting presumption that the power of the Israel lobby over American policymaking is so obvious that everyone recognizes it without having to be told. The implication of this presumption (underscored, in case you missed it, by the title they gave their piece) is that anyone who cannot see what is so obvious to everybody else is simply blind, or deluded, or misguided. If this is so—if, as they suggest, this is the “world that experienced observers of the American political scene … recognize”—one wonders why they felt compelled to write their book at all, since they seem to feel that it states the obvious. They repeat this rhetorical gambit later in their response, as in the discussion of the extent and consequences of campaign contributions by American Jews, about which I will say more later. But if the basic proposition that the Israel lobby is powerful is so obviously true, one ought to be able to observe this power in action by describing and documenting a set of precise causal mechanisms by which this influence over policymaking operates. This is where I still believe their argument falls short.
Methods, Processes, and Causal Mechanisms
Mearsheimer and Walt's response to my methodological critique makes several points. I respond here to the most salient. First, they claim that they assess alternative explanations for American policy toward Israel, pointing to chapters 2 and 3 of their book, in which they dismiss geopolitics and ethics as sufficient explanations. Having dispensed with international system-level explanations, they turn to American domestic politics as their primary causal arena. Among domestic political explanations, the influence of organized interests is certainly a plausible one, and any serious investigation of policy ought to consider the organization and activities of relevant interests as one component of a comprehensive study. But interests comprise only one element among many plausible components of a domestic politics-based account of policy outcomes; the structure and operation of policymaking institutions, the dynamics of public opinion, the nature of electoral processes, political ideologies and national political culture, and historical policy legacies all might also be components of a causal account of policy outcomes independent of interests, and often constitute rival explanations. Focusing an account of the domestic political causes of a set of policy outcomes so doggedly and exclusively on a single explanatory factor would seem to require at least a defense of this methodological move—why interests instead of any or all of the other competing domestic factors?—and, preferably, a research design that systematically assesses an interest-based explanation against other rival domestic political hypotheses. This kind of analysis can fruitfully challenge even a widely-held conventional wisdom about the outsized influence of powerful organized interests in American politics that seemed “obvious” to many observers. For a long time, for example, the standard story behind the repeated failure of the United States to enact national health insurance in the twentieth century was that it was repeatedly blocked by the influence of physicians and the effective lobbying of their main professional organization, the American Medical Association. But more recent accounts have carefully assessed this explanation alongside other hypotheses and successfully shown that organized interests were, in fact, substantially less consequential than institutions or policy legacies.Footnote 1
Mearsheimer and Walt provide neither a defense of their approach nor a clear assessment of alternative domestic explanations for policy outcomes. The transitional section between the “international” and “domestic” sections of their book moves quite abruptly from the conclusion that “the strategic and moral case is increasingly weak” (110) to a discussion of interest groups and their influence on policymaking, without a theoretically grounded defense of the exclusive attention to interests as against other factors. They do briefly discuss (and dismiss) public opinion (108–110) and, as they point out, return to public opinion two or three other times in the book, but this is not the same thing as testing rival hypotheses side by side to challenge one and advance the other. Nor do they offer anywhere a guide to how they will assess the lobby explanation against others. What rival theories and hypotheses might account for the outcomes in which they are interested? How does their selection of cases allow them to make inferences about policymaking processes? How do they canvass evidence that allows them to adjudicate among competing explanations? They suggest that they address these issues in the book, but the passage to which they refer in support of this assertion (14–18) is merely an introductory précis of the book's remaining chapters that seems to function more as a preview of coming attractions than as a detailed description of research method and inferential logic.
Second, Mearsheimer and Walt also describe and defend their process-tracing method. As I indicated, I have no quarrel with process-tracing as a methodological tool to show causality especially in cases where outcomes result from complex chains of events and where narrative methods can fruitfully expose those causal processes. In particular, process-tracing is especially useful as a way of identifying the causal mechanisms underlying a political process and of testing hypotheses about those mechanisms. In order for process-tracing to be useful as a method for making causal inferences, however, it should be built on a clear theoretical foundation and must be deployed to test hypotheses about precisely specified causal mechanisms. As Alexander George and Andrew Bennett, perhaps the most important elucidators of process-tracing as a method, write, “theories or models of causal mechanisms must undergird each step of a hypothesized causal process for that process to constitute a historical explanation of that case.”Footnote 2 Moreover, they write, “process-tracing provides a strong basis for causal inference only if it can establish an uninterrupted causal path linking the putative causes to the observed effects, at the appropriate level(s) of analysis as specified by the theory being tested.”Footnote 3 A convincing causal argument based on process-tracing will connect cause and effect in an unbroken chain of intervening micro-level causal links. Finally, process-tracing, like methods of analysis based on covariation, should investigate alternative causal hypotheses alongside the hypothesis of particular interest. George and Bennett again: “There may be more than one hypothesized causal mechanism consistent with any given set of process-tracing evidence.”Footnote 4 The use of process-tracing, as I wrote in my original critique, does not absolve analysts of the need to state and test theoretically-grounded hypotheses about the specific causal mechanisms that produce the effect. Mearsheimer and Walt mistakenly suggest that I all but concede their argument when I acknowledge that the policy outcomes in the cases they examine are consistently pro-Israel. But the burden of a process-tracing argument, which they themselves have taken on, is not simply to show that some presumed cause is associated with the expected effect but that the hypothesized causal mechanisms are operating, and on this point I remain unconvinced.
As I elaborate in my critique, Mearsheimer and Walt invoke a great many such causal hypotheses in their account that seek to connect the activities of the lobby to policy outcomes through a set of intervening policymaking mechanisms involving many aspects of the American policymaking process, from elections and public opinion to Congress and the executive branch. Confirming these hypotheses would seem to require an empirical account not just of the lobby's activities but of the workings of the specific policymaking processes that they suspect are at work in any given episode. But the case studies that form the core of their empirical argument do not provide much evidence about processes that bear on these hypotheses. The case chapters focus, rather, on the activities of people and groups associated more or less with the lobby, on the one hand, and policy outcomes on the other, without clearly discernible statements about the intervening processes that connect the former causally to the latter. Take, for example, their account of the “lobby's” role in the decision to invade Iraq in 2003 (ch. 8). The evidence in this account catalogues a great deal of activity by Israeli government officials (233–38), neoconservative American government officials and journalists (238–40), and AIPAC and other organizations (241–43). These activities amounted, they suggest, to a sustained public relations campaign in favor of war, beginning even before Bush took office, that involved not only garden-variety punditry (243–50) but also manipulation of intelligence, which they also attribute largely to the lobby's influence (250–53).
There is, of course, no doubt that these people and organizations played a role in the drumbeat for war after September 11, 2001. But Mearsheimer and Walt's account does not focus sufficiently on the processes of policymaking that connect these activities to the policy outcome at issue. How, exactly, do they conceptualize the causal processes by which the lobby's activities influenced the presidential decision to go to war, which was taken with the explicit approbation of Congress (and which is, analytically speaking, probably overdetermined even apart from the lobby's putative influence)?Footnote 5 Beyond the occupancy of influential executive-branch positions by pro-Israel neoconservatives such as Paul Wolfowitz and Douglas Feith, they give little indication. Nor, indeed, do they provide any direct evidence about presidential (or congressional) decision-making, which would seem to be useful, if not essential, in establishing a continuous chain linking cause and effect. What role did the other causal mechanisms they invoke—involving electoral considerations, campaign financing, the composition of Congress, and the like—play in the chain of events that led to the decision? It is not clear that Mearsheimer and Walt have met the explanatory standard of process-tracing because it too often remains unclear exactly what processes they are tracing.
Evidence and Inference
Which brings me to the second major part of their response, their reply to my critique of their specific arguments about the lobby and its role in American politics and policymaking—specifically regarding Congress, campaign finance, electoral behavior, and public discourse. In each case, I argued, Mearsheimer and Walt fail in their book to specify sufficiently clear causal mechanisms and hypotheses about the causal processes that connect the Israel lobby's activities with pro-Israel policy outcomes. Consequently, I suggested, their evidence does not support their contention that the lobby's power is the principal cause of American foreign policy toward Israel. In their response, they still do not explain the causal mechanisms behind their argument. Along the way they misstate several of my claims.
They also suggest that I mistakenly ignore the literature on interest groups in American politics and that my focus on these arenas of politics and policymaking—Congress, campaign finance, and elections—is misplaced. They defend their argument as being “consistent with the extensive literature on interest groups.” If by this they mean their description of what individuals and organizations with pro-Israel policy preferences do, they are correct; much of the literature on American interest groups over the last forty years or so (since Mancur Olson's withering deconstruction of pluralism) has focused on explaining the existence, proliferation, and activities of interest groups, and the activities of AIPAC and other pro-Israel actors are consistent with the cumulative picture that this literature has drawn.Footnote 6 No one disputes this. But if they mean that their causal argument is consistent with our developing understanding of how and under what conditions interest groups exert influence over policy, they are on shakier ground. First, making this kind of argument immediately and necessarily involves precisely the policymaking institutions and political processes that are at issue here—again, Congress, the presidency and the executive branch, and the electoral forces that shape the behavior of actors in those settings—and should be built on clear foundations of theory about those institutions and processes. And second, the literature on the role of interest groups in policymaking in those settings, some of which they cite, is less than conclusive about the mechanisms by which groups can influence policy outcomes.
The uncertainty of the causal mechanisms that underlie Mearsheimer and Walt's argument is, perhaps, clearest in their discussion of Congress, which contains a number of puzzling statements. I offer several possible causal mechanisms, derived from a variety of theoretical perspectives on congressional policymaking, that might explain policy outcomes toward Israel. Naturally they seize on one—the distributive approach to committee power—that might form the basis for a sound causal explanation of policy outcomes that deviate from the median preferences of Congress as a whole, a mechanism plausibly consistent with part of their argument about the lobby. Several points are worth noting here. First, they do not dispute my rejection of (or at least skepticism about) the other mechanisms that might account for the disproportionate policymaking power of a small group of members of Congress—pivotal politics, conditional party government, or the informational perspective on committees. Second, it is not clear that policy outcomes regarding Israel do deviate from the median preferences of Congress as a whole; establishing through process-tracing that this causal mechanism is indeed operating would seem to require demonstrating this. Third, as I point out in my critique, they do not establish that the relevant committees are, as they put it, sufficiently “stocked” with members whose pro-Israel preferences are deviant in this way to make a substantive difference; this is, in any event, a dubious proposition, as I point out, even for the House Committee on Foreign Affairs.
Moreover, their characterization of members of Congress as “parts” of the Israel lobby proves to be something of a moving target.Footnote 7 Early on in their response, they state that they “judged officeholders to be parts of the lobby if their attachment to Israel preceded their entry into public service or if they devoted a substantial portion of their personal or professional lives both in-and-out of office to influencing U.S. Middle East policy in ways intended to benefit Israel” (260). (As a guide to empirical inquiry this seems like a reasonably tractable definition, but as a measurement issue it implies the need for fairly extensive biographical research on members of Congress and other officeholders to ascertain their level of personal and professional commitment to pro-Israel policy before their careers in public service, which as far as I can tell they did not conduct.) But in their discussion of congressional committees in their response, they refer not to this fairly exacting empirical definition but to committees that are “stocked” with “pro-Israel” members more simply, presumably a broader category. In their book, as I point out, they suggest that it is Jewish (and Christian Zionist) members who are the critical actors, substituting religious identity for policy preferences as an indicator. These three definitions certainly overlap, but they are not interchangeable. More important, they further obscure the critical question of causal mechanisms. Are they talking about “members” of the lobby who get themselves elected to Congress? Members of Congress who are not “part” of the lobby but who share pro-Israel policy preferences? Members of Congress who are susceptible to the lobby's influence and thus vote against their preferences? It is far from clear where the lobby's influence lies in this scenario or what causal mechanism they are invoking.
We also differ on the question of whether they can document that there is a pro-Israel litmus test for occupying relevant committee chairs in Congress. On this point they repeat from their book a quotation from Representative Henry Waxman asserting that members who do not share his views on Israel “will not be chairing committees dealing with Israel and the Middle East,” which they punctuate with the tag line, “He was telling it like it is.” They seem to interpret this statement as a threat; it seems at least as likely that it was a simple statement of fact. In any event, the accession of the relevant committee and subcommittee chairs between the 109th and 110th Congresses followed straightforward seniority norms, without Waxman's (or anyone else's) intervention. If either the House or Senate Democrats had violated seniority and leapfrogged a pro-Israel member ahead of a more senior member with different views on policy toward Israel to assume a chair, this might be evidence for Mearsheimer and Walt's interpretation. But it did not happen, at least not in 2006.
Next, they claim that I ignore their discussion of the role of interest groups such as AIPAC in providing information and expertise to members of Congress and their staffs, and particularly their role in providing a “legislative subsidy,” thereby lowering the cost of policymaking for legislators.Footnote 8 This is not, strictly speaking, true, although admittedly I do not use the phrase “legislative subsidy.” In a paragraph on the subject I both suggest that this is not exceptional behavior and question its causal importance. I add here only that the “legislative subsidy” theory of lobbying assumes shared preferences between lobbyist and legislator; it is not a model in which lobbying changes the direction in which policy will go. “The proximate political objective of this strategy,” write Richard Hall and Alan Deardorff, “is not to change legislators' minds but to assist natural allies in achieving their own, coincident objectives.”Footnote 9
Finally, they repeat from the book a quotation from former Secretary of State George Shultz about the Reagan administration's failure to dissuade Congress from appropriating additional military assistance to Israel in the wake of the invasion of Lebanon in 1982. Shultz expresses his frustration at not being able to control Congress's actions (a frustration common among executive-branch officials) and notes “Israel's leverage in our Congress” (quoted on 46). In their response, they suggest that this quotation is evidence of an instance “where individuals or groups in the lobby intervened on Capitol Hill to shape policy outcomes.” But in this passage Shultz says nothing about AIPAC, or lobbyists, or committees, or anything else that might shed light on the causal mechanisms by which he supposes that the lobby may have exerted is influence over Congress's deliberation and decision-making in this case.
On the question of campaign finance, Mearsheimer and Walt make several equally puzzling statements. First, they misinterpret my remark that campaign contributions attributable to pro-Israel donors constitute “a lot of money.” In absolute terms, $7.3 million—the average amount of pro-Israel campaign contributions to federal candidates in each election cycle between 1990 and 2006—is a lot of money. But it is not “reversing course” to say that in relative terms even this amount of money is a small drop in a very large bucket. To update the data, pro-Israel contributions to candidates for Congress and president in 2008 totaled $11.5 million (bringing the average up to $7.7 million per cycle since 1990). This impressive sum amounted to a shade more than one-third of one percent of the $3.2 billion that these candidates raised during that election cycle.Footnote 10 Pro-Israel contributions ranked forty-second among “industry” groups in total donations in 2008. On the basis of these data, it is hard to conclude that pro-Israel donors are responsible for a large share of campaign financing.
Mearsheimer and Walt, however, insist that these data—which come from the Center for Responsive Politics (CRP), generally acknowledged to be the most reliable and unbiased source of campaign finance data—understate the real amount of pro-Israel money in American politics by several orders of magnitude. They offer several reasons for this claim. First, they resort again to the rhetorical strategy of asserting that what they say is something that everybody knows, implying that any skepticism of this conventional wisdom is somehow suspect; “almost everyone agrees,” they write, that American Jews are substantial funders of campaigns. Second, they repeat their estimate that Jewish donors are responsible for anywhere between 20 and 60 percent of funding for Democratic presidential candidates, which seems to be based on several (vaguely sourced) newspaper articles and a thirty-year-old memorandum from White House chief of staff Hamilton Jordan to President Jimmy Carter. I acknowledge that the CRP's data probably understate somewhat the amount of pro-Israel money because of the difficulty of assigning a substantive motive to many individual contributors, and I say so in my critique. (The CRP, which is admirably transparent about its sources and methods, acknowledges this also; the same presumably holds true for any of its sector and industry designations.) But it also seems imprudent to attribute sources and motives to individual donors that we cannot identify and use that speculative attribution as the basis for a causal inference. I preferred in my account to stick to documentable facts, not about contributions from Jewish donors, which might have any number of motivations, but about money that could be definitively linked to pro-Israel policy preferences.
Finally, Mearsheimer and Walt make an important mistake when they say that the CRP database “reports PAC money given to political campaigns but not individual contributions.” This is simply and demonstrably false; they report both, and in fact they are able to attribute millions of dollars in individual contributions to specific interests. The CRP reports, for example, that of the $11.5 million in verifiably pro-Israel contributions made in the 2008 election cycle, $2.9 million came from PACs and $8.6 million came from individuals.Footnote 11
When they turn to the question of the position of Jews as potentially pivotal voters in presidential elections, they again misread what I wrote. I do not “end up agreeing with” their suggestion that Jewish voters are commonly decisive voters in national elections. It is true, as they point out, that I noted the arithmetical possibility that Jewish voters—or, indeed, any bloc of like-minded voters—might be pivotal under the right circumstances. But the right circumstances, as the rest of the paragraph in which that statement appears makes clear, are exceedingly rare. Either one would need a freakishly close election such as 2000—the closest national election in more than a century—or one would have to suppose that Jewish voters would start voting Republican in substantial numbers. The 2000 election was an extraordinarily unusual event, in which political happenstance put heavily Jewish South Florida at the center of a bizarre electoral foofaraw, and hardly seems a solid basis for sound inference (on the “good-cases-make-bad-law” principle). And the Democratic share of the Jewish vote has not, as Mearsheimer and Walt state, “varied significantly” in the last several decades. The estimated Democratic share of the Jewish vote in the last five presidential elections (stretching back to 1992) is 80 percent, 78 percent, 79 percent, 75 percent, and 78 percent, respectively, hardly a picture of dramatic fluctuation.Footnote 12 (Moreover, if, as Mearsheimer and Walt imply, Jewish votes reward hawkishly pro-Israel policy, one might have expected George W. Bush to do considerably better among Jewish voters, particularly in his 2004 reelection bid; but evidently Jewish voters do not uniformly choose candidates that way.)
Thus “arithmetically possible” does not mean “politically realistic.” In fact, the main conclusion to be drawn from my 2004 simulation is that, as I wrote, “it is simply mathematically impossible for Jewish voters to have been decisive in the 2004 presidential election.” A repeat of the simulation for the 2008 presidential election produces the same result; if every Jewish voter in the country had voted for John McCain, Barack Obama would still have won the election, and quite comfortably.Footnote 13
Mearsheimer and Walt also mention the nomination process as a potential pathway by which Jewish voters might exercise a pivotal influence on presidential selection. In a party's nomination contest, they correctly point out, there might be closely fought primary or caucus contests in states that are otherwise typically one-sided in general elections and where Jewish voters might play a decisive role. This is certainly a possibility worth serious systematic empirical investigation (beyond the two paragraphs on the fate of Howard Dean in 2004 that they offer in their book) (164). Here I point out only that Hillary Clinton won large majorities (between 58 percent and 67 percent) of the Jewish vote in the primaries or caucuses of six of the nine states with the largest Jewish population percentage in 2008.Footnote 14 Thus it is not clear that courting and winning Jewish votes in Democratic primaries is either necessary or sufficient to win the party's presidential nomination.
On the question of the Israel lobby's role in shaping public discourse, Mearsheimer and Walt once again fail to confront the problematic question of causal mechanisms and, consequently, misunderstand my point. In the case both of newspapers and magazines and think tanks, they simply presume a causal relationship between what appears in the press or in think-tank output and public opinion (and, presumably as a consequence, policy). They focus attention on the country's major newspapers, they say, “because they exert greater influence on public attitudes.” It is, as I point out in my critique, a contestable proposition whether the content of the media's reporting and analysis has a significant impact on public opinion. Nevertheless, they seem to take this relationship for granted and treat the media's impact as a settled question. Accordingly, when they find evidence of pro-Israel bias in publications or policy research, they conclude that it must, therefore, be influencing public opinion or policy attitudes in a pro-Israel direction—but they do not demonstrate empirically that this relationship holds. They do not present their evidence in a way that constitutes a test of the underlying causal proposition that pro-Israel bias in newspapers and magazines actually affect public attitudes or policy outcomes. Thus when they suggest that their conclusions about the lobby's effects on the media and public discourse stand because I do not challenge any of the specific facts in their book or offer counterexamples, they miss the point. The issue is not whether there are media outlets or think tanks that are less pro-Israel than the ones they cite; it is what the precise impact of the pro-Israel voices they chronicle are on attitudes and policy and, again, by what causal mechanisms this influence occurs. The lobby's impact on public discourse is a hypothesis that deserves investigation—one of the more plausible ones that they offer, in fact—but one on which their data shed little light.
Alternatives
In the final section of their response, Mearsheimer and Walt commit a final, and quite revealing, misinterpretation of my critique. They suggest that I offer several alternative explanations “that might account for the lobby's influence.” This is not what I do. What I, in fact, do at the end of my commentary is to sketch a set of plausible alternative explanations for the policy outcomes in which they are interested—pro-Israel American foreign policy—that rely on causal mechanisms other than the supposed power of the lobby. As I suggest above, the power of organized interests is only one of several approaches to understanding American public policy outcomes, but it seems that it is the only one Mearsheimer and Walt are interested in entertaining. When commenting on these alternative hypotheses, in fact, they insist on interpreting them not as potentially independent accounts of American foreign policymaking but as restatements of the lobby hypothesis. As I said at the outset of my critique, my purpose was to assess their causal argument—their research strategy, the hypotheses they offer, and the evidence they deploy to test those hypotheses. Part of that assessment involved suggesting parallel directions their inquiry might have taken in order to weigh the lobby explanation against others as a step toward both showing some of their argument's limitations and suggesting directions for further research that might, as they put it, “identify the real driving forces behind the special relationship, and U.S. Middle East policy more generally.” I agree that this should be our goal, and I look forward with anticipation to the research that will realize it.
