Abstract
The military institution enjoys consistently high levels of public confidence, making it one of the most respected entities in American society. However, this aggregate trend conceals an underlying polarization among partisans, who arrive at different conclusions over the institutional quality of the military. What explains the divide? This analysis argues a partisan bias resulting from (1) exposure to different information about the military and (2) different cognitive biases in using that information to render an opinion on institutional quality. Using observational data on news media reporting patterns and original text-as-data processed through unsupervised machine learning techniques, I find that partisans during a key phase of the Iraq War were exposed to widely different types information on military performance. Furthermore, I find through original survey experimentation that when presented with information on negative performance by the military, partisans adopt very different sensibilities in how that information is used to evaluate it.