A selection of my academic publications

Investigating Cancer Inequalities in Urbanizing Texas with Plausible Reasoning
The Annals of the American Association of Geographers (2024) DOI:10.1080/24694452.2024.2425807

This article examines changes in the social geography of colorectal cancer (CRC) incidence following the spread of improved preventive technology circa 2003, namely screening colonoscopy. It adopts a realist approach to social science methodology, and a political-economy of health perspective on disease prevention. PDF (open access)

Spatial Uncertainty and Probability
with co-author Yongwan Chun. Oxford Handbook for the Spatial Humanities (forthcoming)

This chapter introduces basic concepts from the frequency-based theory and from the rival, epistemological tradition of probability theory. We sketch out some connections between probability theory and ongoing discussions of methodology in the humanities, singling out Franco Moretti’s ‘operationalist’ method of literary analysis for constructive critique. The chapter then provides a theoretically-grounded orientation to spatial data analysis, including exploratory analysis, spatial regression, and geographical theories of spatial structuring or relational space. To illustrate, we re-analyze state prison sentencing data from a previously published study of Florida’s convict leasing program.

Plausible Reasoning and Spatial-Statistical Theory: A Critique of Recent Writings on ‘Spatial Confounding’.
Geographical Analysis (2024) https://doi.org/10.1111/gean.12408 (open access)

Statistical research on correlation with spatial data dates at least to Student’s (W. S. Gosset’s) 1914 paper on “the elimination of spurious correlation due to position in time and space.” Since 1968, much of this work has been organized around the concept of spatial autocorrelation (SA). A growing statistical literature is now organized around the concept of “spatial confounding” (SC) but is estranged from, and often at odds with, the SA literature and its history. The SC literature is producing new, sometimes flawed, statistical techniques such as Restricted Spatial Regression (RSR). This article brings the SC literature into conversation with the SA literature and provides a theoretically grounded review of the history of research on correlation with spatial data, explaining some of its implications for the the SC literature. The article builds upon principles of plausible inference to synthesize a guiding theoretical thread that runs throughout the SA literature. This leads to a concise theoretical critique of RSR and a clarification of the logic behind standard spatial-statistical models.

Building spatial conditional autoregressive models in the Stan modeling language
OSF Preprints (2021) https://osf.io/preprints/3ey65 (open access)

This paper details some of the computational methods used to implement (proper) spatial CAR models in the geostan R package. A comparison of Markov chain Monte Carlo (MCMC) samplers shows that these custom Stan models can sample about 10 times faster than CAR models in the popular platform Nimble (measured as effective sample size per minute of sampling). The paper models censored mid-life U.S. county mortality rates to demonstrate how geostan can support Bayesian spatial modeling with Stan.

Modeling community health with areal data: Bayesian inference with survey standard errors and spatial structure with co-authors Yongwan Chun and Daniel A. Griffith.
Internt’l J. of Environ. Research and Public Health (2021) https://doi.org/10.3390/ijerph18136856 (open access)

This study examines the potential for sampling error in the U.S. Census Bureau’s American Community Survey (ACS) estimates to influence research findings when ACS estimates are used as covariates. Of particular concern are predictive models that may be used to assess needs for service provision. Based on the reported standard errors, the quality of ACS estimates differ considerably across survey questions, place, scale, and demographics. We show how survey standard errors can be incorporated into a Bayesian hierarchical model for spatial regression and/or disease mapping, and we show why standard (non-spatial) measurement error models are inadequate for small-area ACS estimates. The method is now available part of the geostan R package.

The Making of Florida’s ‘Criminal Class’: Race, Modernity, and the Convict Leasing Program, 1877-1919
Florida Historical Quarterly (2019)

Under the State of Florida’s convict leasing program (1877-1919) approximately 14,000 Floridians and visitors served sentences of hard labor, working for private interests like phosphate mines and lumber companies. This article draws on over four decades of reports on the prison system by its administrators in the Florida Department of Agriculture, geographic sentencing data, data on prisoner characteristics, minutes from the Board of Pardons, and additional materials held in the Convict Lease Subject Files in the Florida State Archives. The study engages with a number of questions revolving around the inter-connected themes of forced labor, industrial interests, violence, disability, and racial ideology. PDF (pre-print with images); PDF (the FHQ article)

Florida Historical Quarterly cover