If you’re hungry, how are you fighting?

The third dimension on which research differs has to do with the social scale that is being studied. By social scale, I mean the degree of coordination among individuals and organizational resources needed to undertake collective action. Environmental variables may affect an individual's propensity to resort to violence and crime (Rotton & Cohn, 2003); relatively unorganized protests and riots (Hendrix & Salehyan, 2012); rebellion by organized, armed actors (Burke et al., 2009); and international conflict between states (Tir & Stinnett, 2012). Results for one type of social phenomenon may not be commensurable with other types of behavior. Moreover, causal relationships are likely to differ; the causal chain leading from water scarcity to communal conflict between farmers and pastoralists may look very different when examining conflict between nation-states over water resources. The same variables that influence conflict between states, which are highly organized entities, may not apply to conflict between individuals and vice-versa. To illustrate how social scale matters, take for example a pair of studies by Hendrix and Salehyan (see Hendrix and Salehyan, 2012, Salehyan and Hendrix, 2014). In the first study (Hendrix & Salehyan, 2012), we show that abnormally high and abnormally low levels of rainfall, which can both disrupt normal agricultural practices, predict social conflict events such as sporadic protests and riots. These types of unrest are usually short-lived, require relatively little sustained commitment, and have limited degrees of organization. By contrast, armed conflict—which requires a long-term commitment by ‘professional’ militant organizations—declines during periods of low rainfall and resource scarcity (Salehyan & Hendrix, 2014). Such groups need food and supplies in order to maintain a rebel army, and find it more difficult to sustain their operations under conditions of environmental distress. Thus, in looking at different scales of social organization, the effect of a similar set of independent variables may differ.