Systemic effects, statehood, and dyadic research designs

These three articles (Cranmer & Desmarais 2016, Diehl & Wright 2016, and Poast 2016) productively extend an emerging debate about the theoretical and empirical utility of dyadic research designs in International Relations (IR) scholarship.  While they obviously reach different conclusions, together they reflect a deeper reorientation in the field. I support that reorientation, which manifests as varying levels of discomfort with the relative inattention to systemic factors in IR research after the Cold War.  This metatheoretical orientation toward domestic and dyadic levels of analysis potentially interacts in pernicious ways with empirical research using dyads. In this brief post, I underscore these risks and highlight another that is not specific to the dyadic approach but further complicates empirical conclusions reached through it.

I begin at the same place as Diehl and Wright (2016)  do—with the inherent path dependence in research processes. The widespread adoption of a dyadic analysis coincided with the end of the Cold War and a related frustration with the inability of prominent systemic theories to anticipate or explain this monumental change in world politics. In response, the field embraced domestic explanations of international outcomes (Oatley 2011), perhaps most prominently in democratic-peace research. The adoption of an empirical strategy capable of examining how a comparison of the domestic attributes of two states could shape interactions between them facilitated this broader theoretical orientation. As part of this mutually reinforcing process, the modeling assumptions associated with a dyadic design reached the status of standard operating procedures. They shaped the presentation and interpretation of potential new contributions to this debate. I have seen this as both an author and a reviewer. For example, manuscripts that try to evaluate some claim related to the democratic peace within a monadic or state-year research design are often asked to confirm those findings with dyads for the sake of comparability. 

This path dependence in research practices and the underlying assumptions associated with dyadic research identified in Cranmer and Desmarais (2016) and Poast (2016)  increases the risk that these reinforcing theoretical and empirical orientations foreclose the search for alternative explanations at the systemic level for important outcomes like military conflict. My own research on the democratic peace (McDonald 2015) illustrates how these theoretical and empirical orientations helped mistakenly reaffirm the robustness of the democratic peace. It rests significantly on the premise that the regime type for any single state in the international system (and thus for dyads as well) reflects an equilibrium that is jointly set by internal coalitional attributes and a larger set of systemic factors.  Consequently, shocks associated with great power politics influenced these internal regime outcomes, induced waves of democratization (Gunitsky 2014) more broadly in the system, and also shaped conflict patterns among these democratic states. 

This research (in process) has uncovered a deeper set of theoretical and empirical modeling challenges that stem from the near universal adoption of states as the principle organizational form in international politics. These challenges are often magnified when using dyads, because they match states with other states to create the sample. 

For example, the existence of hierarchical relationships (e.g. Lake 2009) challenges a reliance on legal recognitions of sovereignty to identify both states and dyads. Reflective of the hyperdyadic dependence discussed in Cranmer and Desmarais (2016), membership in a hierarchical order generally means that a state does not have the foreign policy independence that is implied by standard monadic and dyadic research designs. 

More broadly, systemic waves of statehood in the twentieth century (Reus-Smit 2013) have shocked the population of states upward, influencing empirical conclusions reached through dyadic analysis in multiple ways. The growth of states alters the relative balance of observations from any single year in a standard time-series cross sectional data matrix. Years with more states will comprise a larger proportion of the sample. Given state growth in the twentieth century, the temporal aggregation of observations into a single sample creates recency effects in which coefficient estimates rest more heavily on the correlations among variables in later years.  This can obscure changing relationships over time. The explosion of sample size generated by moving from a monadic to a dyadic design compounds these effects.  This tendency has helped to obscure important changes in the relationship between regime type and peace over time.

The decision to rest units of analysis on statehood creates a second problem that opens larger questions about endogeneity in the study of conflict. Statehood often emerges as part of a peace settlement among multiple states.  In the most prominent waves of statehood and democratization that followed the ends of World War I and the Cold War, statehood, democracy, and peace for many political units were all effectively mutually constituted. Consequently, the absence of conflict for many of these newly independent states—and their resulting dyads in the ensuing period—may simply reflect the terms of the larger international settlement that recognized them. Apart from altering the temporal composition of the sample, these systemic shocks point to a series of omitted variables that could challenge existing conclusions. 

Finally, the values of prominent control variables in dyadic studies of conflict are influenced by the number of states that exist in the international system (McCormack and McDonald in process). Two states are less likely to be contiguous when there are more states in the system.  A dyad is less likely to contain a great power when there are more states in the system. The time counter measuring the duration since a conflict was observed in a cross-sectional unit is also reset when a state gains entry to the international system. Moreover, these statehood-induced changes in the levels of these control variables are all magnified when moving from a monadic to a dyadic research design. 

Thus, I read this debate over the utility of dyadic research designs as part of a larger call to think more broadly about how systemic factors shape prominent questions in the study of international relations. Despite substantial progress, prevailing research procedures centered on the use of dyads have subtly pushed the field away from such perspectives. This conversation helps reintegrate them in a productive way.


Gunitsky, Seva. 2014. “From Shocks to Waves: Hegemonic Transitions and Democratization in the Twentieth Century.” International Organization 68(3): 561–97.

Lake, David A. 2009. Hierarchy in International Politics Ithaca: Cornell University Press.

McCormack, Daniel and Patrick J. McDonald. n.d. Hierarchy and the Unit of Analysis Problem in International Relations. University of Texas and University of Pennsylvania.

McDonald, Patrick J. 2015. “Great Powers, Hierarchy, and Endogenous Regimes: Rethinking the Domestic Causes of Peace.” International Organization 69(3):557-588.

Reus-Smit, Christian. 2013. Individual Rights and the Making of the International System Cambridge, UK: Cambridge University Press.


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