We set out an array of initial conditions for both sides of a conflict that form their levels of public support. These initial conditions result in selection effects; most wars whose initial conditions are not favorable to public support never occur and are nonevents. Miscalculations occur, and some wars whose opposition was underestimated are almost immediately unpopular. Leaders can attempt to persuade the public and shift the cost and benefit perceptions. But the selection effect is strong and as a result, support for a conflict, at least initially, is generally high. We also identified a set of factors that might vary as the war is fought. Changes in strategy, alliances (on either side), and news from the battlefield can alter the expectation of costs and the values assigned to the aims of the conflict. The war’s aims can change as well. Each of these wartime factors can shape opinion by shifting costs (ETC) and value (RP) higher or lower. We can thus identify the factors that influence individuals and consequently the aggregate shape of society’s value of a conflict ex ante within and across conflicts and that predict its impact on support given a war’s estimated costs.

The Reservation Point — RP — captures the value of achieving a set of war aims and Expected Total Costs — ETC — captures a war’s anticipated costs. Both vary across persons and conflicts and together determine the relationship between military casualties and opinion. Not achieving the aims makes costs unacceptable, endogenizing winning and losing. Variations in casualty expectations over time and space lead the political consequences of war to differ temporally and geographically. The role of casualties grows over the duration of the conflict. Marginal casualties will have a proportionally greater effect on the population’s estimates of total expected casualties (ETC) at the start and cumulative casualties over the duration of a conflict. Individuals’ values of wars and their aims fluctuate, which means that opposition will be triggered at different levels of expected and observed costs. The distribution of beliefs about the value of a conflict’s war aims and the distribution of ETC — both of which may change within a specific conflict, and clearly vary across them — are the primary (although not unique) factors that shape elite and mass approval.

People with extreme views appear to have fixed positions, but actually reflect calculations unlikely to be swayed by wartime information. The Evaluative Public, whose valuations are not in the tails, use wartime information to form their conflict approval. Evaluators are sensitive to changes in beliefs about observed casualty levels and expectations of likely future casualties. The Evaluative Public represents those who change positions in war, from support to opposition, depending on their ETC and RP. Using historical data and innovative experiments, we demonstrate that people’s personal experience with war drives their estimates of total casualties. Holding goals constant, higher costs yield higher opposition while lower costs increase support. Holding costs constant and allowing goals to change leads to opinion change. Geographically and temporally proximate casualties strongly influence estimates of a war’s total costs. Thus, we see that casualty patterns affect people’s estimates of a war’s costs and that these costs, when contextualized with the value of a conflict and expected costs, shape wartime support, even in the face of strong individual-level characteristics.

Casualties affect elections in two ways. First, wartime variables affected position formation, where higher state casualties increased the likelihood that challengers openly opposed the war. Second, casualties influence Senate elections directly. Incumbents are held responsible for the conduct of the war, and their vote share is adversely affected by higher casualty rates in their states. Although both incumbents and challengers face constraints, our findings suggest that incumbents face the greatest constraints while challenger behavior is endogenous to casualties. Candidates react strategically to the information provided to them by their state-level casualties, suggesting strategy is not reserved to the battlefield. Candidates behave strategically when formulating wartime positions, rightly perceiving that electorates respond to candidate position differences when voting. Analyses of elections during the Iraq and of Senator positions are taken during the Vietnam Wars. Even when national issues dominate headlines, advertisements, and campaigning, all politics remain local – especially wartime politics.

Looking at the Korean and Vietnam Wars, we evaluate the influence of casualties disaggregated by space/hometowns and time on mass opinion in both the Korean and Vietnam wars and on individual opinion in the Vietnam War. We find a powerful connection between US casualties and public support for a war consistent with our expectations about the importance of casualty trends, the geographic locations of casualty hometowns, and the interaction of these dynamics. Disaggregated casualties are better able to capture variation in mass public and individual wartime opinion than are logged cumulative national casualties – the standard wartime measure employed. We also find that the wartime information-opinion process operates more strongly in the ex ante identifiable early stages of a conflict, and less effectively later in a conflict when casualty expectations (and thus the value of new information) begin to harden. These results strongly support the general notion that casualty patterns act as an observable proxy for our RP/ETC process by capturing information that individuals draw on to generate ETC and formulate wartime positions, improving our ability to understand and predict wartime opinion.