Quasi-Experimental Evaluations with Multi-Item Outcomes: Potential Problems and Solutions

This dissertation seeks to investigate measurement problems related to using multi-item measures in two classes of quasi-experimental designs. It features three chapters with each chapter functioning as an independent paper. The first paper focuses on the Difference-in-Difference design and uses simulation to outline the possible effects of sum scoring and measurement invariance when using survey scale like measures. The second paper shows how this works empirically by rescoring an already completed paper with different Item Response Theory methods and showcasing the resultant differences. Finally, the third paper investigates the viability of using a measurement model to integrate information from multiple raters in Regression Discontinuity designs. Together the three papers help improve the rigor of quasi-experimental evaluations by integrating advances in measurement with those in causal inference.