I checked Twitter for the first time in two months or so and found this interesting tweet from Jonathan Bernstein at Bloomberg View.

The responses were great. This is surprising because few things reach a consensus in political science and our practitioners love to argue. However, the returns on this tweet, summarized here and aggregated on Twitter with the #psfrustration hashtag, largely focused on American politics. These are important for the layperson to know. Citizens United may be a nuisance, but money doesn’t determine elections. We can pine for a viable third party but we will not have one under a first-past-the-post electoral rule like ours. Independents are secret partisans and partisanship is a perspective that conditions our outlook toward politics and society (and not the other way around).

Nothing I saw (at least, immediately) touched on international relations. That got me thinking about what are some things I wish I could implore on non-political scientists I know (and definitely students). The following list is incomplete, but they constitute some preliminary reactions.

Religion Isn’t a Huge Deal in IR

Ask the typical person about conflict in the world and religion is likely the first off-the-shelf explanation they have for it. This is definitely the case for my students. ISIS wants to establish a caliphate, a uniquely Islamic form of government predicated on a chief executive as a religious representative of sorts. Iran is a theocracy and this is what makes them scary to Americans. Religious extremists executed the terror attacks of September 11th, 2001. India and Pakistan fight because India is Hindu and Pakistan is Muslim. I offer the last statement to mention one religion other than Islam. These are just a sample of things students say or I overhear in public.

The truth is religion does not factor into contemporary international conflict scholarship. There’s a reason for that too. On the balance, we’re not sure it ultimately matters a whole lot. At least, most scholarship does not find, and therefore discounts, that religion is a central element to the onset of violent conflict.

This is definitely true for scholarship on conflict between sovereign states in the international system. No matter how hard some empirically-oriented political scientists tried, Samuel Huntington’s Clash of Civilizations argument just will not die in popular discourse. The problems with Huntington’s treatise are multiple, including the all-too-important problems of conceptual fuzziness and a primordial understanding of various religions. Ultimately, Huntington’s onto something the extent to which he has some evidence to support his claim that religion will be the faultline for conflicts after the Cold War. He doesn’t. Most militarized conflicts occur within the civilizations that Huntington labels (e.g. Iran and Iraq) and not between them.

On that note, the Iran-Iraq rivalry has little to do with the Sunni-Shia divide in Islam. If we grant that an increasingly conservative Iranian regime (i.e. post-revolution) tried to foment a Shia rebellion in a Sunni-dominated (but diverse) Iraq as a longstanding source of tension, we should still reiterate most of the severe fighting between the two followed three sections of an unclear border that were never completely demarcated. Iraq initiated the worst stretch of fighting in the rivalry over the Shatt al-Arab water route to the Gulf and oil-rich Khuzestan. This is a more accurate cause of war between the two rivals, but most Americans probably still think it’s a function of a Sunni-Shia divide.

Even the case of India and Pakistan, ostensibly a Hindu-Muslim conflict to the casual observer of politics, is not a conflict of religion. India is predominantly (almost 80%) Hindu, but has over 120,000,000 Muslims. The territory over which India and Pakistan have fought is predominantly Muslim, which is the basis for Pakistan’s claim on the entirety of the territory. Both sides first fought over it because Jammu and Kashmir lay in the gray area of the ill-defined and contentious Radcliff Line that served as the basis to demarcate British Raj. For its part, Jammu and Kashmir tried to remain independent (nearby Hyderabad and Junagadah were forced into India) until Pakistani irregulars infiltrated the territory to force it into Pakistan. There’s a lot more to this conflict than religion. The origins of the conflict may cluster on religion, but it’s more accurately a story of mistrust and state-building post-independence that created a long-standing rivalry over the border between them.

No scholarship of which I’m aware says much about the effect of religion on conflict between states. Conflict scholarship tends to care more about features of state institutions, the overall strategic context, or the properties of the foreign policy issues they contest. There isn’t much of a relationship between religion and conflict inside states either. Fearon and Laitin, who have possibly the most-cited civil war analysis ever, find no discernible effect of religious fractionalization on civil war onset. Collier and his colleagues find the same null finding. Montalvo and Reynal-Querol find no robust relationship either. In a recent article in Journal of Peace Research, Doug Gibler and I amend, clean, and update Fearon and Laitin’s original data from scratch. We also include newer data for key indicators and include the entire international system until 2007. We still find no effect (at least on civil wars and lower-intensity conflicts over control of government).

Even research that explicitly tackles the problem of religion in civil war finds that, among “religious civil wars,” most are only incidentally about religion. Religious civil wars may be more intractable under certain conditions. They are also contested in areas that are more religious (or where certain religions are more prominent). However, they are not a typical case of civil war.

That’s not to say religion doesn’t have its own properties or effects within our subdiscipline that are worth unpacking. Religious citizens tend to see foreign policy issues differently. Religious Americans, in particular, tend to see various foreign policy issues as crusades, which can have important effects. While these findings are interesting for academics, they do not support the perspective of the typical American that most-to-all conflicts in sensitive parts of the globe are motivated by properties of religions or religious fanatics.

I think most scholars share this perspective about religion in conflict. It’s why our analytical energies look elsewhere despite what appears to be popular discourse about the centrality of religion in international conflict. I know I try to get my students to stop thinking in this way when the issue arises in class.

The “Democratic Peace” Exists, but so What?

My entry-level students and friends who took almost no political science classes in college have a working knowledge of the democratic peace. The core of this finding is that no pair of democratic states have ever fought against each other in a war, even if democracies, at the unit-level, are no more or less war-prone than other types of states. This finding is well-known and has traveled well. It might be the most important statistical finding in the history of our field. I don’t think that’s hyperbole either. Almost thirty years ago, Jack Levy christened it as the closest thing we have to an empirical law in our discipline. Few, if any, findings in our field enjoy that status. Duverger’s law comes to mind, but you will not hear U.S. presidents like Bill Clinton and George W. Bush talk about it much like both did with the democratic peace to guide or support their foreign policy initiatives.

In the aggregate, I’m not sure we should invest much in this argument or its findings anymore. No two democracies have ever fought each other in a war. It exists; but, so what? Democratic peace is a frustrating term. For one, consider that “democratic peace theory” is the full name of the research program. However, this is a misnomer. The origin of the theory is a statistical finding by Dean Babst over 50 years ago that no two elective governments have ever fought each other in war. The finding preceded the theory and every theory that’s followed has been tailored to explain those two core findings. It’s a democratic peace fact in need of a democratic peace theory.

The theories we have offered to explain these findings have not reached the same level of consensus as the core empirical findings. Early institutional theories, like Morgan and Campbell (1991), argued that democracies are more peaceful with each other because democratically-elected leaders are beholden to democratic process. This, hypothetically, slows down the conflict process between democratic states and allows time for cooler heads to prevail. However, this implies that democracies should be more peaceful, in general. They’re not. Many have tried to vindicate the “monadic” democratic peace (i.e. democracies as more peaceful overall), but I can find an obvious fatal flaw in any of those seminal works on the topic. To say the evidence that Rousseau et al. (1996) and Benoit (1996) offer in support of the monadic democratic peace is “weak” would be an understatement. Further, the early institutional arguments even contended that democratic major powers should be more peaceful, in general. We would be hard-pressed to say this describes the foreign policy behavior of France, the United Kingdom, or the United States.

What if democratic “norms” explain the democratic peace? After all, democratic institutions vary but the democratic norm of “bounded competition,” so William Dixon argues, is common to all. There are fatal problems with this argument as well. For one, the argument is practically tautological. The norms perspective assumes that democratic leaders prefer negotiation over other means of conflict management because this is how democratic leaders operate in domestic matters.  It purports to explain the presence of negotiated settlements in jointly democratic disputes by assuming that democratic leaders prefer negotiated settlements.  Notice what happened there.  This perspective has not explained that democracies prefer negotiated settlements; it just assumed it.

Second, what is “democracy” and what are “democratic norms?” We conceptualize democratic norms as values of “contingent consent”, “live-and-let-live”, “majority rule, minority rights” and other terms like that. In practice, this tends to devolve into defining concepts with more concepts, which is not the best way to proceed with deductively evaluating a claim. If we grant that democratic norms are a Potter Stewart problem and we just have to roll with what we have, we do not get any closer to matching concept to measure. Democratic peace scholars may conceptualize democratic norms, but they almost always resort to using the Polity data to measure democracy. This is because Polity (to my knowledge) is the only data set of democracy that contains observations before 1945. While Polity has the longest reach of any data set on democracy, it is a data set that is concerned with constraints on executive decision-making, not norms. This leads to some uncomfortable findings. The U.S. enjoys the maximum Polity “democracy” score for much of its history despite widespread disenfranchisement and systemic discrimination of its minorities. At last measurement, Turkey is comfortably above the conventional threshold for democracy despite having a state leader who has an aggressive majoritarian disposition that contends the winner of an election gets to impose his will in governance. These aren’t the democratic norms for which we are looking.

The most persuasive argument for democratic peace theory right now may be selectorate theory, but that theory, the data used to support it, and the empirical analyses its proponents offered for it, have a variety of problems. I still think parts of the selectorate model are useful for political scientists, but we don’t appreciate the extent of the problems associated with that argument as much we should.

All that said, the core empirical findings still survive, barring the most particular hair-splitting about things like the Cod War (pun intended) and the ephemeral declaration of war by the Allies against a democratic Finland early into World War II. The democratic peace still exists, but the core findings aren’t what concern democratic peace scholars. Read one of the core empirical articles for democratic peace and you’ll find the analytical energies are being spent on a few coding decisions that may be only tangential to the research program. So, we split hairs further about whether the democratic peace finding about the absence of war between democracies extends to general MIDs and fatal MIDs. Does it matter if we operationalize joint democracy as a binary indicator or that “weak-link” specification that William Dixon introduced into conflict scholarship? Any more these days, the main points of empirical contention of democratic peace are less about whether it exists and more an endeavor to see what can make the absolute value of the Z-value associated with the weak-link specification of joint democracy less than 1.96 in a non-directed dyad-year research design on MID onset.

My frustration with this coding debate is going somewhere. Absent a strong theoretical foundation for what is, still, an “empirical law,” academics are beginning to place the explanatory power for the finding anywhere other than democracy. There are capitalist peace arguments (and a host of issues associated with them). Perhaps the causal arrow is reversed (as I tend to believe). Peace causes democracy. There’s also work I’ve done with Doug Gibler on auxiliary hypotheses associated with the democratic peace, which argue democratic conflict selection and conflict resolution propensities are epiphenomenal to stable borders and territorial issues. Doug Gibler’s book has gone after the whole thing, recasting democratic peace as a “territorial peace” among contiguous states. If the “democratic peace” really isn’t about democracy, then what, ultimately, does the “empirical law” decree?

The contiguity qualifier is important. Parameter estimates for that weak-link specification enjoy higher levels of precision (and, thus, greater levels of statistical significance) when the researcher uses politically irrelevant dyads as a sampling frame. A Bolivia-Paraguay democratic peace is a lot more meaningful than the Bolivia-Switzerland democratic peace, though the statistical support for the latter is stronger than the statistical support for the former. This doesn’t wash out the core empirical finding, but it does lessen the scope and contribution of the research program associated with it. “Empirical laws” shouldn’t hinge on degrees of freedom. I’m sure there’s a pun in there somewhere for a conference paper.

Temporal components may matter too. Democratic peace scholarship tends to use non-directed dyad-year models with temporal left bounds that extend to 1816. These facilitate the use of the Polity data set as the measure of democracy. However, democracies were scarce before World War II. The U.S. is the longest, continuous democracy in the data set, though my previous discussion illustrates how constraints on executive selection do not neatly square with our conceptions of democratic norms. Canada is democratic since 1888. Mexico is coded as democratic in 1846, 1847, and 1863. Honduras is democratic in 1853, 1907, and 1912. Costa Rica is democratic since 1875. Colombia was democratic from 1867 to 1885. Peru had a two-year stretch of democracy in 1881 and 1882. Bolivia had a year of democracy in 1841. China and Bulgaria had one-year stretches each before World War I. The U.K., France, Belgium, Greece, Norway, and Switzerland are democratic beginning in the mid-to-late 19th century. Spain and Portugal had a few ten-year stretches near the turn of the 20th century. Australia and New Zealand enjoyed democracy in remote Oceania. The state system had just 16 democracies in 1914.1

The list of countries and most of these durations are not particularly inspiring. Much of the support for democratic peace before World War II hinges on a non-trivial democratic peace between Belgium and Netherlands and the United Kingdom and France, even if the “democratic peace” between Australia and Costa Rica and Canada and New Zealand is clearly trivial. After World War II, Cold War politics and shared preferences may account for the abundance of peace among the post-war democracies. Personally, I don’t think democratic peace scholarship has adequately dealt with this temporal component to the phenomenon.

More importantly for the typical citizen, why should any of this matter? I don’t think democratic peace scholars want to admit it, but the democratic peace is likely a backward-looking phenomenon without a lot of practical relevance to the present or future. It says little about the issues that democracies contest, assuming only that they find a way to settle their issues peacefully. However, the issues they have among themselves largely concern minor issues of fishing and trade. These are not terribly exciting issues, nor war-prone issues. If democracies promote population growth (sort of), democratic peace needs to build in additional assumptions about resource abundance in an era of climate change that I don’t think we can justify. If the implication is we should proliferate democracy across the globe, we’ll run into the problem of “bushwhacking” (pun intended). Foreign-imposed regime change rarely works the way it is intended. What good is the “empirical law” if it does not inform any sound policy?

So, yes, democracies do not fight each other, and never in war. We’re not sure (that is: I’m not convinced) that means a whole lot and that ordinary citizens should invest much stock in it. At least, I try to say this and I don’t get that far convincing the proverbial hearts and minds.

  1. Nowadays, the number of democracies may be north of 90, at least using the last estimates in my version of the Polity data set. ↩