Hold on a minute. Those with a passing acquaintance with classical sociological theory might say "Joas could have a point with the likes of Durkheim and Parsons. The former through the increasing mutual dependency and integration of societies via the division of labour, the latter with the wide acceptance of the normative basis of social order, but what about conflict approaches?" What about them indeed. You could be forgiven for thinking that conflict between social groups, as per Weber-influenced conflict theory; or class struggle and the competitive pursuit of markets as with Marxism might root the causes of war in the 'external' displacement of this violence. The problem is the former does not explain why war exists. The latter, how peace is possible and why warfare is the exception, not the rule of advanced societies.
Joas 2000 book, War and Modernity looks at why contemporary social theory and sociology generally has had problems thinking about and explaining war, and this is partly dues to the lingering influence of liberalism. As a philosophy of freedom, liberalism has long-acknowledged problems recognising systemic inequalities and social conflict because, classically, it limits itself to questions of right and politics solely in relation to public life. I'm not going to elaborate further (done it already, at length, except to note I broadly agree with Joas on liberalism's impact on the social sciences. On the one hand, there is an economic 'tradition' stemming from Adam Smith, a line of argument that believes boosting trade between nations incentivises cross-border relationships, draws societies closer together and renders war increasingly obsolete. The second is the republican (constitutional) tradition stemming from Kant: the notion that as republics (and constitutional monarchies) place sovereignty in the hands of the people, they are less likely to endorse war because they are liable to pay its blood price. War, where it occurs, is a throwback; an eruption of irrationality or backward attitudes. Provided economics and republics could carry on with their civilising mission, there's no reason why war cannot be "outgrown". Joas argues that these assumptions can be discerned in Durkheim and functionalism/systems theory, but infects Marxism as well. For all its formal distancing (and rubbishing) of liberalism, the liberal assumption of order without conflict is transposed onto the socialist future.
Despite casting a long, peaceable shadow on sociological theorising Joas notes liberalism has tooled-up skeletons rattling in the historical closet. The flipside of "peaceful" trade was the theft of land and resources by gunboat and bayonet. The universalism of republican government excused civilising missions and the reduction of colonial subjects to beings unworthy of rights.
That's the heritage. But it's only been allowed to persist in sociology because of the circumstances of its founding. Sociology was institutionalised as an accepted academic discipline in the two decades prior to the First World War. Its homes in Germany, France and the USA were characterised by relatively cohesive societies with a stable state apparatus, the rule of law, and the development of relatively well-integrated national economies. All three were involved in colonial adventures and, in the USA's case, war with Spain. However their participation did not cause social dislocation nor call for a concerted national effort in the way a long-running conflict would, let alone a world war. Sociology then reflected the social reality of the day. War didn't impinge on everyday, run-of-the-mill social relationships (at least at first remove anyway) hence sociology barely touched upon it.
All that is understandable prior to the Great War, but since? Western societies have been profoundly affected by the experiences of global war. Since 1945 European colonial empires have disappeared, and, in the American and British contexts, post-colonial conflict, Vietnam, the Gulf War and the "humanitarian interventions" in Sierra Leone, Kosovo, Afghanistan, Iraq and Libya have upped the cultural and political costs of war in the absence of social dislocation at home - perhaps best illustrated regarding current attitudes toward Syria and Ukraine/Russia. Yet over this period, in each and every sociological tradition war is either ignored or regarded as a minority interest. You're more likely to find sociologists writing about virtual wars than actual conflicts. The faddy sociologies of globalisation that emerged in the late 80s were more or less retreads of Adam Smith's arguments. Postmodernism's identity politics sat very much in the Kantian republican tradition, of overcoming social conflict by the recognition of difference and valorisation of the many ways to be human. Of course, this is possible because advanced capitalist countries are largely safe from bombings and military incursions. Yet it still cannot explain why, now that war "elsewhere" fills the news feeds nearly every day, sociology turns a blind eye to it.
The problem is conceptual and theoretical. The big claim of sociology is its attempt to try and explain the world, to lay bare the processes and dynamics underpinning collective behaviour that make social life possible. It's about patterns, order, and great sweeps of social movement over time and across territories. It doesn't like, and is largely allergic to contingency. It has had a great deal of trouble reconciling the theorisation of abstract structures drawn from the observation of collective behaviour with the fact human beings are free-willed, choice-making animals. The tension between structure and agency all too-often dissolves into one side disappearing agency, the other structure. It's the tyranny of the structure vs the tyranny of structurelessness. Yet all social behaviour, all rituals are a blend of mutual conditioning, of dialectical interplay and interdependence - as ably demonstrated in their own ways by Bourdieu and Giddens. The question of war, however, is especially problematic because it is a large-scale set of social processes with structural underpinnings, and yet are radically contingent in a way that can have utterly profound implications for the societies involved. That means it's difficult to pin down a chain of causation in the outbreak of conflicts, because there is no such iron will to war.
As it's the centenary year, look at the causes of the First World War. Who's responsible? Britain declaring war on Germany? The French honouring their obligations to Russia? The Russians declaring for Serbia? The Germans backing Austro-Hungary against Serbia? Serbia for training and funding the Black Hand? Gavrilo Princip's assassination of the Archduke? Or the Archduke's driver for taking a wrong turn? Historians have argued about the efficacy of each of these, and much else besides. Each of these, while building on the preceding set of decisions, were contingent. They were conditioned by the division of Europe into two grand military alliances, and these itself were an outcome of the age of high imperialism, of the division of the world between established and upstart great powers. This context however did not cause the First World War. There was no inexorable inevitability about 1914. The decision-makers all acted rationally according to their perceptions and social knowledge, leading to outcomes that probabilistically increased the likelihood of war, but they had choice within the context of their social universe. They were not passive agents of inter-imperialist rivalries. Rather, war was an outcome of the contingencies the nature of those rivalries threw up.
Let's take a less symmetrical situation: Israel's shelling, bombing and now ground invasion of Gaza. Oppositional analysis of the nature of Israeli society variously emphasise it being a settler state, a land-grabbing state, a project to create a mono-ethnic Jewish state. As colonial in origin, the role of the military in society and willingness to deploy violence against Palestinians is a systemic, structural property of that society. However, those who subscribe to this analysis protest and demonstrate against Israel's actions in defiance of their analysis. Demanding that it stop its attacks on Gaza recognises that there is room for contingency here, that the Israeli government does have a choice. Indeed it does. Israeli society is in a perpetual state of military preparedness. Its politics are so distorted by its continuing commitment to occupying the West Bank and desire to contain Gaza that, internally, there are domestic political incentives for politicians to indulge racist and warmongering rhetoric. Probabilistically, these set of social relations are more likely to lead to war than any other advanced capitalist society, and that is without the specific (and contingent) circumstances of 'normal' Palestinian unrest, peace talks, and Hamas rockets. Decisions matter.
There is no such thing as determination, causation and necessity in sociology: only probabilities. And these probabilities are constantly shifting, moving, bending with the simultaneous weight of interests, conflicts, and contingent social actions. The problem with integrating this into sociological theorising is it's too messy, too banal and, in the case of war and the causes of war, far too close to the domain of International Relations. What's more it's an approach that cannot be spelt out in the abstract, except for some basic methodological rules of thumb. The play of pattern and contingency, of structure and agency can only be realised in close empirical study. Perhaps this unwillingness to read events as opposed to canonised texts is the real root of sociology's aversion of things war-related.