About Dr. Seals

I am a Pensacola native and earned both a BS in mathematics and a MS in mathematical sciences here at the University of West Florida. I then moved to Birmingham and attended the University of Alabama at Birmingham, where I earned a PhD in biostatistics.

Research Areas

Model misspecification in the frequentist framework.
  • How are our analysis results and overall message affected when we misspecify the design matrix, underlying distribution for the outcome, or the working covariance structure?
  • How are analysis results and overall message affected when we break an assumption (or multiple assumptions) of the parametric model?
  • This area generally involves performing simulation studies. We do this so that we know the true value of θ, the parameter of interest and can then quantify the effects of model misspecification.
    • The bias and mean square error quantify the error in parameter estimation.
    • Type I error quantifies the error in statistical inference.
    • Model fit indices quantify the error in model selection.
Computational environmental science.
  • Please see the Computational Geomorphology & Modeling Lab.
  • How can we use mathematics and statistics to answer questions about the natural environment around us?
    • We are interested in modeling and predicting natural phenomena and changes within our local environment.
  • Because the University maintains an active research site on Pensacola Beach, we have natural opportunities for collaboration with the Department of Earth and Environmental Science.
Education: both general STEM and discipline-specific
  • What are the longitudinal effects of student-led interventions in gateway STEM courses?
  • What instructional tools improve student outcomes in the statistics classroom?
  • How are student learning outcomes impacted when STEM courses naturally infuse statistics into the curriculum?
  • When are significant learning gains made during statistics and data science graduate programs?
    • Is there a difference in when learning gains are made between those in statistics, biostatistics, and data science programs?


Want to Work Together?

Are you are a researcher interested in collaborating with us?
  • As a result of my extensive biostatistics training and experience, I truly love collaborating with other researchers, especially students. Please note that our skills are not only applicable to "science" topics -- our skills are applicable to any field where data exists!
  • Please reach out if you think this may be a good fit! It is a wonderful experience to see how students begin connecting the dots, understanding their role as partners in scientific research.
  • The main expectations within collaborations include a reasonable timeline for project deliverables and formal co-authorship for both the lab director and collaborating student on any resulting presentation or manuscript.
Are you a student interested in collaborative statistics?
  • As a collaborative researcher and educator, I especially enjoy including students in collaborative projects. My goal is to model how to form and maintain respectful and productive collaborative relationships.
  • This side of the lab gives students a formal collaborative experience resulting in a non-statistics or data science-focused research product to showcase to potential employers.
  • The main expectations of collaborative students are a willingness to collaborate outside of your field, learn necessary statistical and/or science concepts, and a willingness to learn and improve your R programming skills.
Are you a student interested breaking the rules of statistics?
  • As a curious mathematician and statistician, I have a lot of "what happens to analysis results when this assumption is broken?" questions ready for students to answer via Proseminar or Capstone.
    • If you think about it - you probably do too! Let's consider OLS regression. We know that OLS assumes that the residuals are normally distributed with mean 0 and some constant standard deviation.
      • What happens if the residuals have a Poisson distribution? A uniform distribution?
      • What happens if the standard deviation is not constant? How does this change depending on the level of heterogeneity?
      • How do the observed relationships above change as our sample size increases? (i.e., what are the asymptotic properties?)
  • The main expectations of simulation-based students are a willingness to learn necessary mathematical and/or statistical concepts, a willingness to learn and improve their R programming skills, and to have an inherent understanding that research-related things will go very wrong more than once.