Statistical Analysis

Statistical analysis is what I do every day. I have very broad experience implementing a variety of models in biostatistics, educational research, political science, and other research areas.

If you need help on formal research design section for a grant application, need a set of statistical models to be implemented from scratch, or simply would like to have a consultation to discuss best practices or determine the ideal methods to analyze your data, I would be happy to help you on your next project.

Here is a sample of the methods I have employed in the past:

Standard Econometrics and Biostatistical Methods

  • Descriptive statistics (central tendency, dispersion, clustering, correlations)
    • Used clustering statistics to quickly remove geography-specific text in order to identify major topics in an analysis of federal government reports.
  • Difference tests (chi-squared tests, t-tests, etc.)
    • I employ a standard difference test several times a week!
  • Standard regression models (ordinary least squares, logistic, generalized models, fixed-effects)
    • Implemented a non-parametric model with firm-level fixed effects and bootstrapped standard errors to analyze political determinants of foreign direct investment.

Research Design

  • Clinical trials
    • Currently lead analyst and randomization coordinator for randomized control trial exploring the effectiveness of a diabetes prevention health coaching program in ten counties in the Mississippi Delta.
  • Power analysis
    • Implemented a power analysis to determine viability of observing population-level effects of teen pregnancy prevention program that is reaching thousands of Mississippi students.

Causal Inference: Randomization and Quasi-Randomization

  • Randomized control trials
    • Currently lead analyst for randomized diabetes prevention study covering ten counties in the Mississippi Delta.
  • Matching
    • I regularly employ matching in the longitudinal analysis of electronic health records and for facilitating cross-country comparisons in international relations.
  • Regression discontinuity
    • I implemented a regression discontinuity design to determine the effectiveness of a literacy intervention in which enrollment was determined by performance on a formative assessment.

Advanced Econometrics

  • Multilevel modeling
    • Implemented two-level growth model for longitudinal assessment of student-level education data.
  • Survival analysis
    • Implemented a Cox proportional hazards model to characterize passage on an elementary summative assessment with repeated attempts.
  • Count models
    • Implemented a negative binomial model to analyze determinants of counts of diplomatic cables at the country level.
  • Network modeling
    • Implemented an exponential random graph model to explore sophisticated hypotheses regarding the network of foreign direct investment.

Text as Data

  • Topic modeling
    • Implemented a machine-learned topic model to identify latent topics of ambassadorial performance in State Department reports.
  • Sentiment analysis
    • Implemented both a machine-learned (support vector machine) and dictionary-based method to identify sentiment of political text.
  • Document similarity
    • Employed an algorithm to identify the extent to which text in diplomatic cables written by embassy staff was discoverable in public State Department reports.