This dissertation explores the theoretical finite sample and asymptotic properties of several econometric estimators. Two central themes are robustness and pragmatism. This dissertation develops theory that is focused on real-world problems faced in applied econometrics. To provide increased robustness and reliability in applications, theories and tools must accurately capture the true behavior and construction of estimators. This often means taking explicit account of procedures (e.g. variable selection in Chapter II) that traditional, nonrobust theories ignore. By capturing and studying these steps, complex and technical though they may be, it is often possible to deliver highly robust results that are also accessible and easily implemented. Although this dissertation spans a broad range of microeconometrics, this common foundation links the chapters together. That is, the goal of the theory develop is to deliver practicable methods for empirical research or put already-common practices on sound theoretical footing.