I have extensive experience across a wide range of data and stastical models, including experimental data for causal inference (e.g., t-tests, ANOVA, ANCOVA), observational data for prediction and inference, including survey data, routinely-collected and administrative data (linear regression and its generalized forms, structural equation models, Bayesian networks), and also longitudinal data, such as panel designs and intensive longitudinal designs (multilevel models, additive models, vectorized autoregressive models).
My background is psychology so I tend to work with person-centered data, to understand how people change over time, and what is driving those changes - including exogenous forces and within-person causes.
But I’m also generally interested in how to model change over time, especially models which allow causal inference.
Communication is a key part of statistics so data visualization (dataviz) is a necessary and critical tool. Joe Berkson is credited with coining the IOT (Intraocular Trauma Test (Edwards et al 1963), a plot so clear that it hits you between the eyes. My PhD supervisor always said “a good graph beats statistics”, and I’ve always aimed to design interesting and good-looking plots with a clear message. I’m also interested in the cognitive biases and processes we bring to data vizualization and how people understand graphs and plots. Di Cook’s Nullabor package is a different tack on how to use perceptual psychology to understand data. Over time I’ve learned that modelling and plotting are often two sides of the same coin. Every plot implies a model of the data, and I love working with data, using models and plots to understand it.