My research program is primarily directed towards providing methods for identifying and incorporating latent variables in structural models. These methods aim to characterize the social genesis and evolution of attitudes, dispositions, skills, etc. at the individual level, and to examine their effects on behavior and outcomes. Examples include the effect of cultural and economic attitudes on the vote, cognitive and non-cognitive skills on life outcomes, life events and societal position on subjective happiness, and ethnic identity and locus of control on the social integration of immigrants. Claims of ceteris paribus effects are difficult to sustain in such contexts, even when the explanatory variables are directly observable. This work develops quantitative assessments that recognize the ordinal and stochastic nature of concept measurement in these areas and the (usually) discrete nature of the outcomes; it also addresses endogeneity issues and employs novel semiparametric methods.