These two useful books represent important contributions to the literature on violent ethnic conflicts. Monica Duffy Toft proposes a useful theory and adduces convincing evidence on some of the key determinants of severe ethnic violence. Stephen Saideman provides an even stronger theory and strongly suggestive evidence for explaining states' decisions to intervene in ethnic conflicts over secession. Both books provide a mix of quantitative analysis and case studies to support their claims more or less convincingly. Both conclusions deserve prominent attention in the literature.

As Toft summarizes her central theme in The Geography of Ethnic Violence, different actors view the same territory in different ways. The outbreak of ethnic war, in her view, is explained by the theory of indivisible territory: If both sides in a conflict see control over a disputed territory as indivisible, they are likely to fight over it (pp. 1–2). More specifically, ethnic groups are most likely to demand sovereignty over their territory if they are a concentrated majority, representing the majority of the population in what they see as their homeland. States are most likely to resist such demands if, as is usually the case, they are multiethnic and fear setting a precedent for other potentially secessionist groups.

Toft's reasoning is sensible: Concentrated majority groups, because they are the majority, have a claim to democratic legitimacy for their demands, and at the same time are most likely to have the capability to rebel. Her statistical data provide some eye-catching support for these contentions. In particular, she finds that concentrated minority groups, representing 48% of all minorities at risk, account for 78% of ethnic rebellions, and that 63% of concentrated majorities engage in at least some ethnic violence. Most of the rest of the ethnic rebellions are accounted for by concentrated minorities.

While Toft's theory makes sense, the value of the book is weakened by the straw-man nature of the alternative arguments she considers: the materialist explanation (a simplistic conflictual modernization model); the nonmaterialist explanation (a simplistic security dilemma model); and the elite manipulation explanation. She would have done better to test more sophisticated versions of both of the first two approaches that are put forward in different parts of Donald Horowitz's 1985 classic Ethnic Groups in Conflict: the psychological dynamic model he proposes early in the book, and his advanced-backward distinction among groups and regions. This would have added nuance both to the theory testing and to Toft's use of her own theory.

The strength of Toft's theoretical approach is that it joins constructivist logic about the definition of group identity with rationalist analysis of the interaction of the consequent group preferences. However, virtually all of the analytical leverage comes from the constructivist logic: If both sides in a dispute inflexibly demand full control over all of a disputed piece of territory, it hardly needs a formal model to show that they will come into conflict. This fact turns up the theoretical weakness of the work: The theory gives too little weight to identity construction, leading Toft to overemphasize materialist-rationalist factors, especially demography, in her case analyses, while the evidence points to the importance of identity construction.

In the case of Georgia's conflict with secessionist Abkhazia, for example, Toft argues that because they were a local minority, the Abkhazians did not at first define their demands as indivisible, and that this changed only after the war began. Her own evidence, however, shows that Abkhazia was inflexibly demanding complete independence before the war. The constructivist part of her theory explains this behavior perfectly: The Abkhazians' notion of homeland driven largely by fears of Georgian domination motivated them to make this demand not in spite of but because of their demographic weakness. But this is not the argument Toft makes. Similarly, in the Tatarstan case, she overemphasizes the role played by the Tatars' not-quite-majority demographic status (48.5% in Tatarstan), while underrating the fact that nearly half of all families in Tatarstan were ethnically mixed. Here was the real demographic factor for ethnic moderation in Tatarstan (driven, in turn, by the high degree of ethnic toleration that allowed for such high rates of exogamy in the first place).

Toft's statistical case is actually in some ways stronger than she makes it. Being a concentrated group is, as she notes, very nearly a necessary condition for an ethnic rebellion. The only exceptions are Lebanon's Sunnis and Rwanda's Tutsis, however, these exceptions prove the rule, as Lebanon's Sunnis were dragged into a war started by others, and Rwanda's Tutsis started with a base of support in neighboring Uganda. What Toft neglects to clarify is that the demographic factor is not a sufficient condition for violence: Only 25% of concentrated majorities rebel. The rest of the variance is accounted for by other variables.

The strengths of The Ties That Divide are the converse of Toft's work, as the book is characterized by excellent theoretical argumentation and case studies but weaker statistical data. Saideman's theoretical argument is simple but sound: State leaders tend to back foreign ethnic groups with which their own constituents sympathize, especially when the leaders feel their position is threatened (and so feel a need to bolster their base of support).

The alternative theories Saideman considers are the prominent ones in the field, and they perform surprisingly badly. He drives another nail in the coffin of the old conventional wisdom already convincingly attacked by Alexis Heraclides' claiming that states that are vulnerable to secessionists tend not to support other's secessionists. He also convincingly debunks a realism-based model suggesting that states might tend to support secessionists in states that threaten them.

Rather, his case studies show that quite consistently, states choose their policies on ethnic or racial grounds. In the Congo-Katanga crisis of the early 1960s, for example, black African states backed the African nationalist government of Congo, while minority white regimes in Africa, along with colonial power Belgium, supported the pro-white secessionist Katangan leadership. In the case of the Nigeria-Biafra conflict, the main line of cleavage was religious, and so Nigeria's government, dominated by the Muslim Hausa-Fulani group, garnered support from Cameroon (with a strong Fulani constituency), Niger (which was 46% Hausa), and the Muslims of the Arab Middle East.

Eschewing a simpleminded clash-of-civilizations argument, however, Saideman also shows that some cases that might appear to violate ethnic logic actually do not do so. Thus, in the Yugoslavia case, Orthodox Greece opposed Orthodox Macedonia because Greek nationalist ideals objected to the Macedonians' appropriation of the name Macedonia. Some of his arguments are a bit strained: Christian Ethiopia's backing of Muslim Nigeria to promote civic nationalism, for example, sounds more like the vulnerability argument than the ethnic one. Still, all in all, his argument holds up far better than the alternatives.

There are two objections, significant but not decisive, that could be raised about the research design. First, the Congo case is not as crucial a case as Saideman claims. Because the key cleavage was racial, it was less precedent-setting than it might have been; white racial domination was already a cause in terminal worldwide decline. Still, if not a pivotal case, it is still an important one, not least because proponents of other theories claim it as a source of supporting evidence.

More important is the design of the quantitative part of the book. The first portion of the quantitative analysis simply sums up all of the observations from the three case chapters: useful, but not, as Saideman concedes, anything like a representative sample. The second portion, while using the large Minorities at Risk (MAR) data set, focuses on the wrong unit of analysis. Although Saideman's logic, as he concedes, is dyadic, the statistical analysis asks what factors lead groups to gain support in general, or states to offer it in general, rather than looking for ethnic links between patrons and clients. He pleads overload, as there are tens of thousands of potential state-group dyads derivable from the MAR data set. This is a fair point, but a smaller data set consisting of a sample of politically relevant dyads would still have allowed him to generate a more convincing set of tests. As it is, the value of his statistical tests is (as he concedes) mostly to provide more evidence to undercut the rival hypotheses, rather than to support his own.

These caveats aside, both books are valuable and make important contributions to the literature. Both authors sustain their main cases. Toft is clearly right that demographic patterns have an important effect on the likelihood of ethnic violence, as does the way nationalist ideology defines a group's relationship to its homeland. Saideman is clearly right that ethnic ties strongly influence states' policies toward ethnic separatism. Anyone interested in the causes of, or international relations of, ethnic civil wars should read both books.

Both also have notable policy implications, but primarily negative ones. Toft's offers further evidence for why ethnic civil wars are so intractable. Saideman's points out that while prescriptive arguments about international intervention tend to assume that the international community can and will act in concert, states rarely do so because their domestic politics often drives them to back different sides. These cautionary notes are already well known to policymakers, but they are important for any theorists who might want to venture into the realm of policy prescription.