The Search Continues: The Progress of Quantitative Nuclear Studies and the Bridges Yet to Cross

Mark Bell (2016) surveys the landscape of the determinants of nuclear proliferation.  Bell offers a strong and analytically-driven critique to the current approach of many large-n quantitative studies and highlights room for improvement and empirical advances.  As a contributor to this scholarship, I am sympathetic to and motivated by Bell’s concerns and appeals for more carefully constructed research designs and modeling choices.[1]  I am also similarly convinced that our field can continue to improve upon this existing work by employing nuanced data and novel design.[2]

Yet, my primary criticism of this piece stems from two potential implications of Bell’s analysis.  The first is that the onus of analytical responsibility rests on the quantitative scholars (rather than on both theoretical and statistical work alike).  Second, I examine Bell’s suggestion of the potentially more limited role that this research should play in policy-making.

The introduction of quantitative scholarship on nuclear proliferation came at a time when our theoretical understanding of the two biggest issues related to nuclear weapons (its causes and consequences on the international system) was muddled.  In part, this research agenda was attempting to make sense of the growing field of theoretical work that had yet to be tested on the universe of proliferation cases (Singh and Way 2004; Jo and Gartzke 2007; Bleek and Lorber 2014).[3] Some of this early wave of scholarship was trying to wade through competing logics that each offered contradictory implications, and provide a first cut test of which, if any, had empirical merit.  Quantitative nuclear scholarship, to its credit, has done a remarkable job in demonstrating that much of the early theoretical work on the causes and consequences of nuclear weapons is similarly tentative and has produced mixed findings (Thayer 1995; Sagan 1996/1997; Hymans 2006; Solingen 2007; Tannenwald 2007; Paul 2000).  Without this initial phase of large-n testing (with the accepted concerns on explanatory and predictive power), it is important to realize that our understanding of why states pursue nuclear weapons would be even more uncertain. 

Based on this piece, I think Bell would agree that we as scholars and analysts must return to first principles.  As he states in his study, “this literature, however, offers many more distinctive explanations for proliferation than there are cases of proliferation in the historical record.” This similarly suggests the need for more thorough and rigorous theoretical thinking about the explanatory and predictive power of various theoretical mechanisms.  As every good singular modeling choice must be theoretically-driven, so too should our broader approach to multivariate testing.  If some variables are shown to have weak (or potentially harmful) predictive power, it is incumbent upon researchers to determine whether we must move beyond these explanations in the broader scholarship.  It may also then become necessary to invest more time in clearly examining the logic and mechanisms behind these causes of proliferation, and inductively and deductively develop new logic as our understanding continues to progress.

Bell also suggests that scholars should be more careful in providing policy implications based on the findings from quantitative scholarship.  As a fervent believer in the weaknesses associated with probabilistic research, I am often hesitant to provide a clearly binary recommendation for an inherently more complex, multi-causal reality. As such, I support Bell’s appeal for being more cautious in offering suggestions for how policy-makers can better predict instances of proliferation.

Yet, one of the unfortunate side effects of social science research on on-going human phenomena is that we’re sometimes going to get it wrong.  As our understanding of intentions and motivations change (in nuclear decision-making and other arenas), so too do actors’ preferences, strategies, and behavior.  Our research (both theory and analysis) is trying to keep up with an ever-evolving world, and to limit the dissemination of our findings until we’re absolutely certain of its predictive power (and until there is no chance that we’ll get it wrong) may be imprudent.  Policymakers are also operating in an asymmetric-information environment and if our research is able to contribute even a little to ongoing debates and discussions that may prevent harmful and more disastrous outcomes, perhaps it is worth sometimes ‘getting it wrong.’

The ambition of good social research is to answer genuine and significant puzzles in our field, use strong evidence to corroborate theoretically-driven hypotheses, and potentially to amend and extend the paradigm of known phenomena.[4]  While, as Bell points out, there is still much work to be done to accomplish this aims, I take comfort in the progress we’ve made to date. 

While some may read this piece and question the utility of quantitative analysis (despite Bell’s intentions to the contrary), I believe that this study suggests that academic training should more heavily incorporate a more intuitive and complete understanding of the value—and limitations—of statistical analysis.  I take this piece (and related literature) as a call to arms to further delineate and strengthen the manner in which quantitative scholarship can improve our understanding of nuclear decision-making, including specifically the causes and consequences of nuclear proliferation and reversal, as well as other complex political phenomena.[5] 


Works Cited

Bleek, Philipp C. and Erik B. Lorber. 2014. “Security Guarantees and Allied Nuclear Proliferation.” Journal of Conflict Resolution. 59(1):74-92.

Fuhrmann, Matthew and Benjamin Tkach. 2015. “Almost Nuclear: Introducing the Nuclear Latency Dataset.” Conflict Management and Peace Science. 32(4):443-461.

Gartzke, Erik, Jeffrey M. Kaplow, and Rupal N. Mehta. 2014. “The Determinants of Nuclear Force Structure.” Journal of Conflict Resolution. 58(3): 481-508.

Gerzhoy, Gene Rupal N. Mehta, and Rachel Whitlark. 2015. “Assessing the Benefits and Burdens of Nuclear Latency.” Working Paper.

Horowitz, Michael C. and Neil Narang. 2014. “Poor Man’s Atomic Bomb? Exploring the Relationship Between “Weapons of Mass Destruction.” Journal of Conflict Resolution, 58(3):509-525.

King, Gary Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Quantitative Research. Princeton: Princeton University Press.

 Mehta, Rupal N. 2015. “Buying Off Friends and Foes: The Determinants of Nuclear Reversal.” Working Paper.

 Paul, T.V. 2000. Power versus Prudence: Why Nations Forego Nuclear Weapons. Montreal; McGill-Queen’s University Press.

 Sagan, Scott D. 1996/1997. “Why Do States Build Nuclear Weapons? Three Models in Search of a Bomb.” International Security 21(3):54-86.

 Singh, Sonali and Christoper R. Way. 2004. “The Correlates of Nuclear Proliferation: A Quantitative Test.” Journal of Conflict Resolution 48(6): 859-885.

 Solingen, Etel 2007. Nuclear Logics: Contrasting Paths in East Asia and the Middle East. Princeton: Princeton University Press.

 Tannenwald, Nina. 2007. The Nuclear Taboo: The United States and the Non-Use of Nuclear Weapons Since 1945. Cambridge, UK: Cambridge University Press.

 Thayer, Bradley A. 1995. “The Causes of Nuclear Proliferation and the Utility of the Nuclear Non-Proliferation Regime.” Security Studies 4(3):463-519.


[1] See Erik Gartzke, Jeffrey M. Kaplow, and Rupal N. Mehta. 2014. “The Determinants of Nuclear Force Structure.” Journal of Conflict Resolution. 58(3): 481-508. Also, see Gene Gerzhoy, Rupal N. Mehta, and Rachel Whitlark. 2015. “Assessing the Benefits and Burdens of Nuclear Latency.” Working Paper.

[2] See Matthew Fuhrmann and Benjamin Tkach. 2015. “Almost Nuclear: Introducing the Nuclear Latency Dataset.” Conflict Management and Peace Science. 32(4):443-461 and Michael C. Horowitz and Neil Narang. 2014. “Poor Man’s Atomic Bomb? Exploring the Relationship Between “Weapons of Mass Destruction.” Journal of Conflict Resolution, 58(3):509-525.

[3] See also Bell’s discussion on selecting the dependent variable.  See also, Gary King, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Quantitative Research. Princeton: Princeton University Press.

[4] I thank David Lake and Mathew D. McCubbins for this introduction to the philosophy of science.

[5] See Rupal N. Mehta. “Buying Off Friends and Foes: The Determinants of Nuclear Reversal.” Working Paper 2015.

Discuss this Article
There are currently no comments, be the first to post one.
Start the Discussion...
Only registered users may post comments.
ISQ On Twitter