Colloidal microparticles, particles large and slow enough to be imaged easily using optical microscopes yet small enough to be thermalized in a solvent such as water, provide a unique window into the thermodynamic processes behind phase transitions. Specifically, microparticles can be uniquely identified and imaged throughout the field of view for the duration of an experiment, allowing visualization of both the evolution of the sample as a whole and the activities undertaken by each particle during this evolution. In addition to the high spatial and time resolution of experiments, the ability to control the interparticle interaction in these systems via modification of the particle shape, surface, and the suspending solvent make microparticles an extremely attractive system for modeling the dynamics of crystallization and melting. In this thesis, I report the results of two experiments using colloidal microparticles, both containing surprising results. Chapter 4 reports on the first recorded instance of catalysis of crystal layer growth via slight size-mismatching colloidal particles, while Chapter 5 describes the dynamics of a nonintuitive finding whereby a high-density facet of a crystal melts faster than its lower-density counterpart. Together, these experiments demonstrate the importance of considering statistical dynamics during phase changes as well as highlight the utility that colloidal systems bring to the table for investigating dynamics during phase changes.