Social Scientific Analysis of Nuclear Weapons Past Scholarly Successes, Contemporary Challenges, and Future Research Opportunities

Is the academic research agenda on nuclear security and proliferation a victim of its own success? Since their invention in 1945, nuclear weapons have occupied a central role in international politics and in international relations theory. During the Cold War and immediate post–Cold War period, scholars carefully scrutinized the meaning of the nuclear revolution for international affairs (e.g., Brodie et al. 1946; Jervis 1989), the nature and requirements of nuclear deterrence and escalation (e.g., Kahn 1960), why countries pursue nuclear weapons (e.g., Sagan 1996), and the effect of nuclear weapons proliferation on international stability (e.g., de Mesquita and Riker 1982).

In the 1990s and early 2000s, research on nuclear weapons slowed to a trickle. In the absence of great power geopolitical competition, nuclear weapons receded into the background in international politics and scholars interested in international security turned their attention to other issues, such as civil conflict (e.g., Fearon and Laitin 2003) and humanitarian intervention (Kuperman 2004). Indeed, it was during this period that we and some in our cohort were told not to study nuclear security because nuclear weapons and related policy issues (deterrence, North Atlantic Treaty Organization [NATO]) were obsolete or because the obvious questions and their answers had already been addressed by extensive scholarship during the Cold War.

By the late 2000s and early 2010s, however, a new wave of academic research on nuclear weapons began to take shape (e.g., Gartzke and Kroenig 2009). Much of this research had been inspired by a reawakening to the importance of security issues following 9/11, but it took time to conceptualize, conduct, and publish rigorous scholarship. Worried about the possibility of proliferation to rogue states and terrorist networks and, for some, the steps great powers might take to prevent proliferation, scholars reexamined questions of nuclear proliferation (e.g., Singh and Way 2004), nonproliferation (Fuhrmann and Berejikian 2012; Kroenig 2014b; Coe and Vaynman 2015; Fuhrmann and Lupu 2016; Bleek and Lorber 2014), counterproliferation (Fuhrmann and Kreps 2010); and how nuclear weapons affect the frequency (e.g., Jo and Gartzke 2007), timing (e.g., Horowitz 2009), intensity (Rauchhaus 2009; Bell and Miller 2015), and outcome (e.g., Beardsley and Asal 2009) of international conflict; and how nuclear weapons affect international influence more broadly (e.g., Gartzke and Jo 2009). Scholarship also examined the effect of nuclear status on alliances and other third parties (Palmer and David 1999; Fuhrmann and Sechser 2014a). Unlike the first wave of Cold War research, much of this scholarship incorporated more modern social scientific methods, most notably regression analysis of observational data (Gartzke and Kroenig 2016). Rather than simply recapitulate Cold War findings, this wave also made notable conceptual and empirical advances, including conceiving of nuclear proliferation as a game between smaller potential proliferators and great power counterproliferators (Kroenig 2009a; Monteiro and Debs 2014; Miller 2014), uncovering the importance of supply-side factors (Kroenig 2009b; Fuhrmann 2009a; Brown and Kaplow 2014), such as industrial capacity and nuclear assistance, for successful nuclear proliferation, and identifying possible additional benefits of nuclear possession beyond deterrence including improved crisis outcomes and diplomatic influence (Beardsley and Asal 2009; Gartzke and Jo 2009; Kroenig 2013).

This second wave of nuclear weapons research has been so successful that ambitious young scholars and their mentors have once again begun to question (like in the 1990s and early 2000s) whether the viable research questions and arguments have not already been covered. As we will argue below, this is not the case. New nuclear challenges continue to arise and rigorous social scientific analysis will continue to be needed to better understand and grapple with them. Even processes that dominated during the Cold War have reemerged, while “knowledge” in many cases is largely driven by anecdote and organizational consensus. Concepts like mutual assured destruction (MAD) and the key components of a secure second strike have not been fully supported with basic tools of social science, such as multivariate statistics and hypothesis tests.

Another major challenge to this research comes in the form of questioning the utility of commonly used social scientific methods for the subject matter at hand. Most notably, Bell (2016) examined thirty-one variables suggested by the extant literature as correlates of nuclear proliferation. He found support for very few, leading Bell to conclude that the “existing quantitative literature on proliferation produces more tentative findings than scholars typically understand” (2016, 520). Work by other researchers has produced a similar delimitation of the relationships about which scholars can have confidence (Douglass and Lanoszka 2016).

Although much more thorough than past critiques, the idea that there are too few examples of states possessing nuclear weapons for large-N statistical analysis is not new. The criticism was raised frequently during the Cold War and returned with the second wave of post–9/11 nuclear scholarship (Montgomery and Sagan 2009; F. Gavin 2014). But this did not stop the flourishing of research that contributed much to our understanding of these issues. Moreover, it should be noted that the more recent, sophisticated criticisms have focused almost exclusively on the value of statistics for studying nuclear weapons proliferation but not on questions of how nuclear weapons possession affects international political behavior. Furthermore, the most appropriate methods for studying nuclear weapons remain open to debate with several other prominent scholars arguing strongly in favor of the utility of statistical analysis for studying nuclear weapons (Fuhrmann, Kroenig, and Sechser 2014; Gartzke 2014). One of the unique contributions of rigorous research after all is the ability to quantify confidence in one’s findings. It is not clear how abandoning rigor improves our ability to infer from small samples. Significance thresholds, while critical as an objective way to assess the validity of one’s inferences, are arbitrary. In effect, failing to make the most of the available data is comparable to a type II error, rejecting claims about the world that are true. Moreover, although perhaps regrettable for other reasons, the number of observations grows with the passage of time. With each new year, we have more evidence of how nuclear and nonnuclear states behave in international affairs. Scholars today have more data than those in the Cold War and those in the future will have even more still. Although the challenges to pristine research designs in this field are real, they can be addressed. Indeed, in a section to follow, we provide suggestions to help future scholars overcome potential research design obstacles in this field.

Finally, and perhaps most importantly, there may be questions for which a lack of adequate data would be reason for scholars to throw up their hands and move on to other issues, but doing so for nuclear weapons would be irresponsible. Policy makers must make decisions on which (it is not an exaggeration to say) the continued existence of humanity may depend. These consequential decisions about nuclear strategy and nonproliferation policy are necessarily being made at present with insights and information that are far from complete. Although one should not exaggerate the influence of scholarship on policy-making, it is one input into the process (Byman and Kroenig 2016). And if scholars do not conduct rigorous social scientific analysis of the fundamental cause and effect relationships in nuclear security, no one else will. This is not the comparative advantage of think-tank scholars and the policy community. If scholars fail to contribute their best insights about the nature of nuclear politics to this debate, then policy makers will be less informed about the likely effects of their decisions and the earth will be at least somewhat less secure than otherwise.

We strongly believe, therefore, that there remains room for rigorous social scientific analysis of nuclear issues. In the past, much research has tended to respond to and inform real-world events. What then are today’s nuclear security challenges confronting statesmen that will become tomorrow’s research questions?

As with many other substantive areas in political science, the research agenda on nuclear security has tended to follow real-world events. During the Cold War, scholars focused on superpower deterrence (e.g., Glaser 1990) and in the post–9/11 era, they analyzed proliferation (Singh and Way 2004), counterproliferation (Fuhrmann and Kreps 2010), and the behavior of regional nuclear powers (Narang 2014). As in the past, the international environment is currently generating new and difficult security challenges that should inspire future scholarly research.

The most pressing contemporary international nuclear issue is North Korea’s advancing nuclear program. At the time of writing in 2017, North Korea is estimated to have enough nuclear material for thirty nuclear weapons. In addition, North Korea is rapidly advancing its missile capability (Albright 2017). Pyongyang already has the ability to range US forces and allies in the region with its current operational missiles and is working on expanding these capabilities with a variety of improvements to its forces: solid-fueled missiles, mobile missiles, submarine-launched ballistic missiles, and intercontinental ballistic missiles. Unless current trends are arrested, North Korea will become only the third US nonally (after Russia and China) to develop the ability to hold the US homeland at risk with the threat of nuclear attack. This not only endangers the United States directly but also calls into question the credibility of US-extended deterrence guarantees in Asia as well as the broader stability of the region.

There is a bipartisan consensus in Washington that the United States and the international community must once again confront and engage North Korea. The new Donald J. Trump administration just completed an internal policy review thought to contain two major elements. The first element is diplomatic. Building off the perceived success of the nuclear negotiations with Iran, Washington, and its partners will attempt to build pressure on Pyongyang with the hope of forcing Kim Jong Un to the negotiating table to place hard limits on his nuclear and missile program. The second element is military and defensive in nature. So long as North Korea possesses these capabilities, the United States must be able to deter a North Korean attack, defend itself and its allies, and in the event of war, defeat North Korea. This element of the strategy includes bolstering regional and homeland missile defenses and increasing US intelligence, surveillance, and reconnaissance and strike capabilities. In addition, there are reports that the United States is employing cyber and electronic warfare in a “left-of-launch” strategy to sabotage North Korea’s missile development program and, if necessary, blunt any future North Korean nuclear attack (Broad and Sanger 2017).

The other long-running threat to the nuclear nonproliferation regime remains Iran’s advanced nuclear program. The disastrous choice between an Iran with the bomb or bombing Iran was at least temporarily averted in October 2015 with the Joint Comprehensive Plan of Action (JCPOA), better known as the Iran nuclear deal (Kroenig 2014a). The deal allowed Iran to keep a uranium enrichment capability and lifted sanctions on the Iranian economy in exchange for verifiable constraints on Iran’s enrichment program verified with an intrusive inspections regime.

Although the deal certainly bought time, it did not fundamentally resolve the issue of Iran’s nuclear status in the longer term. Many believe that the deal will break down for one reason or another in the coming years. Even if the deal remains in place, however, it does not guarantee a nonnuclear Iran. The negotiated limits on Iran’s enrichment program begin to expire at the ten year mark and phase out completely after year fifteen. At that point, as even President Obama acknowledged, the time it would take Iran to produce at least one functioning nuclear device “shrinks almost down to zero” (Obama 2015). Unless the Iranian government fundamentally changes its foreign policy priorities in the next decade, the international community may face a new crisis with Iran as it approaches the deadlines set forth in the deal’s sunset clauses. In order for the deal to continue to function, therefore, many experts see the need for follow-on negotiations to address the sunset clauses and to extend their continued application further into the future. Building the leverage to renegotiate the terms of a hard-fought deal while living up to the strict terms of the JCPOA will be no easy task. In sum, whether the JCPOA ultimately succeeds in stopping Iran from building nuclear weapons is a story still being written.

In addition, there are broader concerns that the JCPOA may have set a dangerous precedent and undermined an important nonproliferation norm. The vast majority of countries in the world with peaceful nuclear programs do not make their own nuclear fuel. Rather, they operate nuclear reactors for the production of energy, medical isotopes, or research, and the fuel is provided by another more advanced nuclear power, such as Russia, France, or the United States. Once a country has the ability to make fuel for a nuclear reactor, it also possesses the ability to make fuel for nuclear weapons. Since the early nuclear era, therefore, Washington has attempted to draw a clear line between truly peaceful nuclear technologies and sensitive fuel-cycle technologies: enrichment and reprocessing. The United States has even imposed this distinction on its own allies, forcing Taiwan and South Korea to shut down reprocessing facilities in the 1970s. One Taiwanese scientist remarked, “when the Americans got through with us, I’m surprised we can still teach physics here in Taiwan” (Kroenig 2010, 106).

Now, however, the United States and the international community have made an exception to this long-standing nonproliferation principle. It has granted Iran a de facto right to enrich and enshrined that right in a United Nations Security Council Resolution. Moreover, the exception was not made for just any random country, but rather for a country that has routinely violated international norms and its legal nonproliferation commitments. Already, officials from states such as the United Arab Emirates and South Korea are demanding that if Iran is allowed to engage in sensitive fuel-cycle activities, then they should have similar rights. As one South Korean expert asked, “how can the Americans say they trust Iran with enrichment, but they do not trust their ally, South Korea, with reprocessing?”⁴ Officials charged with nonproliferation are attempting to maintain traditional standards, but the argument is harder to make in the wake of the Iran nuclear agreement.

Policy makers are not only concerned with current proliferation challenges but also those that appear to loom in the future. Is the spread of nuclear weapons capped? Or will the world continue to see a slow trickle of nuclear capabilities to new states? If the latter, then which states are the next nuclear proliferators? Many fear that if North Korean and Chinese capabilities continue to grow and the United States is unable or unwilling to continue to provide a credible nuclear umbrella, then US allies, such as Japan or South Korea, might be the next states that attempt to join the nuclear club (Murdock and Karako 2016). Chinese experts have expressed concern about Vietnam’s intentions after that country signed a peaceful nuclear cooperation agreement with the United States in 2014. Innovations in technology create other challenges as well. Scholars have demonstrated the importance of supply-side capabilities for enabling nuclear proliferation (Kroenig 2009b; Fuhrmann 2009b; Brown and Kaplow 2014). New additive manufacturing technologies (better known as 3-D printing) could make it much easier for states to build nuclear weapons and to transfer nuclear know-how, while making it much harder for the international community to detect and stop proliferators in time (Kroenig and Volpe 2015).

Beyond proliferation, the international community faces a number of new or renewed challenges related to nuclear deterrence and strategic stability. Nuclear weapons are the ultimate measure of national power. As such, they remain important tools for powerful actors interested in international political competition. Nuclear rivalries and related modes of contestation were muted in the twenty-five years following the end of the Cold War, but they have reemerged with Russia’s invasion of Ukraine in 2014, intervention in Syria, and continued threats to the rest of Europe, and as China takes contested space from US allies and partners in the South China Sea.

The most notable issue is a concern that the threshold for nuclear use is being lowered (Gibbons and Kroenig 2016). Several states have adopted doctrines relying on threats of nuclear use early in a conflict in order to deter or dissuade conventionally superior rivals. During the Cold War, the Soviet Union possessed a formal no first use policy. However, Moscow abandoned no first use in the early 1990s and has come to rely heavily on nuclear weapons in its military strategy and doctrine. In particular, Russian strategists have gravitated toward a concept of “escalate to de-escalate” or “de-escalatory nuclear strikes.” Rather than lose a war on its borders with a conventionally superior NATO, Russia could conduct a limited number of nuclear strikes in a bid to force the West to sue for peace on terms favorable to Moscow. Such a move would be designed to force Western leaders to choose between (as Henry Kissinger famously put it decades ago) “suicide or surrender.” US strategists are just now beginning to grapple with this challenge and several prominent analysts have argued that the United States and NATO, like during the Cold War, must once again develop a strategy and capabilities for limited nuclear war in Europe.

Russia is not the only state relying on nuclear forces to offset conventional disadvantages. Pakistan also plans to use tactical nuclear weapons on its own territory against invading Indian conventional forces in the event of a war on the subcontinent. In addition, although North Korean strategy is difficult to decipher, Western analysts believe that Pyongyang would use nuclear weapons early in a conflict with South Korea and the United States. The purpose would be to deter the United States from further escalating a conflict and to guarantee regime survival.

Others fear that the greatest threat to nuclear security may come from America’s nuclear forces. The United States has begun a program to modernize all three legs of its nuclear triad over the next thirty years at an estimated cost of US$1 trillion. The plans have bipartisan support and proponents argue that as long as nuclear weapons exist, the United States needs to possess a “safe, secure, and effective” nuclear force. They argue that US nuclear forces were built at the end of the Cold War, have outlived their expected service lives, and need to be replaced. Critics, on the other hand, charge that these plans are dangerous. They worry that US modernization efforts may fuel a new round of nuclear arms races with Russia and China. They also charge that renewing America’s nuclear forces is at odds with the US pledge in Article VI of the Treaty on the Nonproliferation of Nuclear Weapons (NPT) to “pursue negotiations in good faith” on nuclear disarmament. They further worry that some specific capabilities included in the modernization plans, such as the new air-launched cruise missile (ALCM), could be destabilizing. Enemies would not know whether the cruise missile was equipped with a nuclear or conventional warhead and, therefore, might mistakenly launch a nuclear attack in response to a US conventionally armed cruise missile.

There are strong counterarguments to each of these claims. Russia and China are also modernizing their nuclear forces and are further along in their cycles than the United States, raising questions about how US plans could be fueling a race. Article VI does not require disarmament, only that states “pursue negotiations in good faith” and there is no specific time period for when disarmament will occur. And Russia and the United States have both possessed nuclear-armed ALCMs for decades, raising questions about why it is only US plans for a new ALCM that suddenly threaten to undermine stability. Needless to say, whether and how the United States modernizes its nuclear forces will continue to be an area of heated political debate.

The final major challenge to strategic stability on the horizon may come from new forms of warfare, rather than from improvements in existing military technology. During the Cold War, nuclear weapons were the only “strategic” weapon. Over the past two decades, however, many new technologies have been deployed or envisioned which could have strategic effects. These new strategic technologies include missile defenses, cyber, space, conventional prompt global strike, hypersonic glide vehicles, and others. For example, if China were to conduct a massive cyberattack on the power grid of the Eastern United States that caused significant economic dislocation and even death, it would almost certainly be viewed as a strategic attack, even if kinetic weapons were never employed (Lindsay and Gartzke 2017). Similarly, some Chinese analysts fear that the United States may be developing a nonnuclear first strike capability to deny China’s nuclear deterrent. They offer a scenario in which the United States conducts a massive first strike on China’s small nuclear force using cyber weapons and conventionally armed missiles to significantly blunt China’s nuclear capabilities. Washington could then use its missile defenses to mop up China’s ragged second strike. In this way, the United States could conduct a “splendid” first strike all while keeping its nuclear forces in reserve for future threats or attacks. A third possibility is that a nation’s nuclear weapons or command and control could be “hijacked” by a cyberattack, nullifying deterrence (Gartzke and Lindsay 2017). “Cross-domain deterrence,” or how new technologies in a variety of domains affect the fundamentals of nuclear deterrence and stability as developed over the past seventy years, has become a buzzword with much resonance in Washington and in ministries of defense around the world (Lindsay and Gartzke 2016).

These and other nuclear-related challenges exist in the present or loom over the horizon, posing difficult challenges to foreign policy makers. As in the past, however, real-world threats also call for scholars intent on contributing to global strategic stability and countering proliferation to develop new research agendas to inform policy, the public, and other scholars.

Current security challenges provide a new set of important theoretical and empirical questions to which the next generation of scholars can apply increasingly sophisticated social scientific tools. Issues surrounding North Korean and possible future Iranian proliferation, for example, raise a range of questions about nuclear nonproliferation diplomacy. Scholars have examined the strategic interaction between proliferators and counterproliferators (Kroenig 2010; Monteiro and Debs 2014; Miller 2014), but these dynamics can be explored in greater detail. Did Tehran agree to limitations on its program solely to avoid increasing economic and military pressure? Or did a global nonproliferation norm also contribute to this outcome? (Rublee 2009). If the latter, does the signing of the Iran nuclear deal and the enshrining of Iran’s enrichment capability in international law undermine nonproliferation norms? If proliferation becomes a normal good over which states bargain, will other states be tempted in the future to launch their own nuclear programs? What are the conditions under which nonproliferation bargains break down? The 1994 Agreed Framework with North Korea proved to be a failure, but many have higher hopes for the 2015 JCPOA. Theoretically, why might states agree to limits at some point in time (x − 1) only to reverse course at time x? Indeed, if the limits were only agreed to in the harsh light of intense international pressure, will not the proliferator have a strong incentive to reverse course once the pressure of international attention is lifted under the terms of the deal? More broadly, why do states that pursue nuclear weapons programs decide to give them up (Mehta 2017a)? And what combination of inducements are most likely to lead to nuclear reversal (Mehta 2017b)?

Concerns about the future of nuclear weapons proliferation also raise a series of critical analytical questions. What are the conditions under which states protected by a patron’s nuclear umbrella decide to risk the alliance relationship and pursue an independent nuclear capability (Fuhrmann and Sechser 2014b; Kim 2016)? How will the gradual rise in the number of nuclear or near-nuclear powers affect incentives for nonnuclear states to seek to proliferate? What will the effect be of relaxing restrictions on the domestic enrichment and reprocessing of nuclear fuel? Can scholars train their models forward and offer predictions about the future of proliferation? Bell (2016) expresses skepticism about the value of this exercise, but intelligence analysts will be offering these predictions, with or without the help of social scientific methods and techniques. The result of any analysis will no doubt be improved by the application of rigorous study and methods. At the very least, social scientific forecasting can bound the limits of plausible proliferation cycles and identify variables with the largest substantive effects (Kaplow and Gartzke 2016).

It is too early to tell how 3-D printing will affect proliferation, but finer-grained study of the supply-side of proliferation may shed further light. What aspects of industrial capacity or international assistance have been most and least helpful for past attempts at proliferation? Fuhrmann (2009b) has staked out a somewhat bold claim that all nuclear-related assistance matters, but some elements, such as sensitive nuclear assistance, may be more important than others. If the rapid spread of additive manufacturing technology quickly transforms many states into near-nuclear states, then the world could benefit from a better understanding of nuclear latency (Mehta and Whitlark Forthcoming). Latency is defined as a situation in which a state has the technical capability to produce nuclear weapons, including a sensitive fuel-making capability, but, for whatever reason has not yet chosen to join the nuclear club (Fuhrmann and Tkach 2015). This issue has become increasingly salient in the wake of the Iran nuclear deal with analysts asking why Iran would go through the trouble of building an enrichment capability with little commercial application and yet agree to stop short of building the bomb. Was Tehran forced to stop in the face of international pressure, or, contrariwise, do Iran’s leaders believe they have something to gain in terms of prestige or security solely from possessing a near-nuclear capability? Why do states pursue nuclear latency and what are the effects of latency on international politics?

The threat of Russian “de-escalation” strikes and the possibility of early nuclear use from Pakistan and North Korea return questions of limited nuclear war to the forefront of the policy agenda. During the Cold War, the Kennedy Administration famously moved from a policy of “massive retaliation” to “flexible response” to offset conventional advantages of the Soviet Union.⁵This line of thinking would remain in US strategy throughout the end of the Cold War in the form of Carter’s “countervailing” strategy and Regan’s theory of victory (Freedman 1989). Kissinger (1957) made his reputation advocating for limited nuclear war and Kahn (1986) carefully analyzed each and every possible rung on the nuclear escalation ladder. But limited nuclear war received its most sophisticated theoretical treatment only in the Cold War’s twilight. Robert Powell (1989)explained how limited nuclear strikes are indeed a rational solution to the credibility problem at the heart of nuclear deterrence. But with the threat of limited nuclear war only extremely remote for over a quarter century, the trail laid down by Powell was not picked up by other scholars. It is time for social scientists to revisit this research agenda. Is limited nuclear war a rational strategy in equilibrium? How can one best deter an adversary that is employing a limited nuclear war strategy? Is there any hope that a limited nuclear war will actually stay limited?

Concerns that US modernization plans may instigate a new round of arms races returns attention to questions about the causes and consequences of arms races. As with other subjects covered in this review, arms races were the focus of a large scholarly literature during the Cold War (Richardson 1960; Wallace 1979) but have received extremely limited attention over the past twenty-five years (e.g., Glaser 2000). What are the conditions under which one state’s build up or modernization prompts a reaction in a rival state? During the Cold War, many scholars were driven by normative concerns about a hypercompetitive arms race resulting in nuclear overkill (Rosenberg 1983), leading to a proliferation of many hypothesized international and domestic causes of arms races (e.g., Evangelista 1988). Empirically, however, scholars have found that arms races are actually quite rare, raising the important question of what are the conditions under which arms races do not occur (Rider 2009)? What are the international and domestic obstacles to action–reaction cycles most commonly associated with arms racing? Opponents of US attempts at modernization of its nuclear force structure charge that it will result in an arms race, but perhaps a sustained contemporary arms competition with Russia or China is not in the cards. Similar questions are raised about US modernization and the health of the NPT. Will modernization result in a weakening of the NPT and to decisions by other countries to build nuclear weapons? Contrary to claims often made in the policy sphere, recent research suggests that there is not a link between US nuclear weapons and proliferation and nonproliferation behavior in other states (Kroenig 2016), but this is an area ripe for further research.

Finally, how will new technology affect strategic stability? Will emerging technologies call into question states’ secure, second-strike capabilities undermining stability (Long and Green 2015; Lieber and Press 2017; Gartzke, Kaplow, and Mehta 2017), or might new forms of warfare serve as substitutes for more kinetic violence that risks nuclear escalation (Lindsay and Gartzke 2016)? There could be more promise for scholars to analyze these issues by reframing the question for a narrow issue about nuclear second strike into a broader discussion of how new military technologies may influence the likelihood of conflict. Moving from nuclear-specific to general international relations (IR) theory debates, will these new technologies shift the offense–defense balance? Scholars argue that nuclear weapons are defense dominant and have contributed to seventy years of great power stability, but do any of these new technologies promise to return the advantage to the offense? Indeed, has Russia already found a new model of “hybrid” or “gray zone” warfare that allows for tactical offensive gains in an overall defense dominant environment (Kroenig 2015)? Or will these new technologies lead to rapid shifts in the balance of power that are destabilizing in and of themselves (Powell 2006)? If the latter is possible, then the important questions concern where these technologies are most likely to spread and which states are most likely to effectively incorporate them into their military structures (Horowitz 2010; Gartzke, Kaplow, and Mehta 2014). Just as scholars have spent decades studying the causes and consequences of nuclear weapons proliferation, future research agendas can examine the causes and consequences of the spread of cyber capabilities, ballistic missile defenses, space and antispace weapons, conventional strike, hypersonic glide vehicles, additive manufacturing, artificial intelligence, autonomous weapons systems, and many other as yet unanticipated weapons of war.

Despite recent challenges, there are rigorous methodologies and research designs scholars can employ to study these issues. As noted above, there is currently broad agreement among much of the scholarly community that statistical analysis of observational data is useful for studying the effects of nuclear weapons on international political behavior.

For other types of questions, such as nuclear weapons acquisition, where the small-N challenge is admittedly more severe, it is helpful to disentangle the reasons why a small number of observations might pose a problem. One critique is that it is hard to find statistically significant relationships. But, this fact of life provides research advantages as well as disadvantages. If one finds robust and significant relationships even in small samples, then one can have more confidence in one’s findings. Indeed, Bell (2016) acknowledges that while most variables are not strongly correlated with nuclear proliferation, others such as industrial capacity, the receipt of sensitive nuclear assistance, and threat environment are.

Further, evidence of statistical insignificance is informative and can be used by scholars and decision makers to assess their confidence in hypotheses or available policy alternatives. Scholars and journal editors should, therefore, be encouraged to publish the results of substantively important nonfindings, such as Kroenig’s (2016) finding that the size of the US nuclear arsenal does not appear to influence the proliferation behavior of other states.

In addition, difficulties in finding significant relationships in small samples essentially means an overabundance of type II errors. In other words, there may be true social relationships that exist in the world, but that we cannot presently confirm with our studies. This is always an issue in inference and science accepts a bias that imposes this reality. The passage of time and more data may help to address this problem. Another alternative is to increase the number of observations through computer simulations or interrupted time series analysis. In addition, scholars can ask which relationships in the study of nuclear security are substantively most important (Kaplow and Gartzke 2016). Relationships may still be worthy of attention, even if they are not statistically significant. Another option is to broaden the frame of reference by looking at “near-nuclear” and other categories to create larger samples of data. Scholars can look at the proliferation of high-end military technology broadly, for example, to glean insights about nuclear weapons proliferation specifically. Or scholars can study the causes of consequences of latent conventional military power (Gartzke and Kroenig 2016) to better understand nuclear latency.

A second critique resulting from small sample size is that many if not all relationships that can be found can also be broken. But this is true in most of social science. Most IR scholars accept the basic democratic peace finding even though it can be broken. The problem, therefore, is not that we cannot find relationships, but that we entirely cannot rule out the possibility that they are spurious. Correlation does not equal causation, but this does not suggest that we abandon statistics. Indeed, unlike other methods, statistics provides us a way of quantifying confidence in our findings. The critique does call for more extensive robustness checks and supplementing quantitative analysis with other methodologies.

Beyond statistics, there are many other methodologies that scholars can employ to study the full range of nuclear security issues. Multimethod designs can combine statistics and qualitative case studies performed with careful process tracing to improve causal inference. Qualitative-focused comparisons can compare matched and mismatched cases (Paul 1995; Solingen 2007). And, for issues with little or no data, like nuclear war, scholars can continue to employ game theory and experiments (Press, Sagan, and Valentino 2013; Quek 2016). Indeed, scholars could seek to exploit the defense and think-tank community’s extensive use of tabletop exercises and war games to design controlled experiments to better understand casual dynamics around issues such as limited nuclear war, deterrence, arms races, and nuclear proliferation.