Dissertation submitted in partial fulfillment for a PhD in Political Science at the University of Illinois at Urbana-Champaign, 2018
R Code available in word cloud directory.
Rather hilariously, despite my interest in agent-based modeling and machine learning (GitHub repositories for both available for review), the research design for my dissertation ended up consisting of a series of survey experiments.
The survey sample was a convenience sample of bilingual Chinese university students. The sample was convenient because I had limited access to a subset of university departments that would allow for the survey to be conducted. Within each department, however, the treatment was randomly assigned.
The combination of these factors effectively meant that my analysis consisted of a series of between-group analyses (also known as A/B testing). Though not methodologically sexy, I presented careful analyses of my experimental data and present convincing visualizations to complement my narrative.
Techniques used in my dissertation included data visualization with ggplot; lots and lots of two-way ANOVA, Chi-square tests, Fisher's Exact Test (and other fun parametric and non-parametric tests); ordered logit regression; ordinary least squares regression; Factor Analysis and Principle Component Analysis (both Chapter 4).
Dissertation is available available at UIUC Ideal.
Data is currently embargoed pending further cleaning of identifying information. Codebook available upon request after further cleaning as well.