See my CV for a complete list of papers and talks.
Publications
Submitted
Josephs, N. and Upton, E. (2024). Hypergraph adjusted plus-minus. arXiv preprint arXiv:2403.20214.
Jain, D., Kher, S., Liang, L., Wu, Y., Zheng, A., Cia, X., Plantinga, A., Upton, E. (2023). Improving and Evaluating Machine Learning Methods for Forensic Shoeprint Matching. arXiv preprint arXiv:2405.14878
De Veaux, R., Plantinga, A., Upton, E. (2023). Age and Performance in Sports.
Published
Congdon, E., Liu, S., Upton, E. (2024+). Exploring the Trajectories of Age-Related Cognitive Decline via Stacked Ensembles. To appear in Psychology and Aging.
Upton, E. and Carvalho, L. (2024+). Modeling Urban Crime Occurrences Via Network Regularized Regression. To appear in The Annals of Applied Statistics.
Upton, E. M., LaRocca, J., Rudolph, A. E., Young, A. M., & Havens, J. R. (2024). Examining the social and behavioral dynamics of substance use in a longitudinal network study in rural Appalachia. Addictive Behaviors, 108060.
DiGiorgio, A. L., Ma, Y., Upton, E. M., Gopal, S., Robinson, N. J., Susanto, T., & Knott, C. D. (2022). Famished Frugivores or Choosy Consumers: A Generalist Frugivore (Wild Bornean Orangutans, Pongo pygmaeus wurmbii) Leaves Available Fruit for Nonfruit Foods. International Journal of Primatology, 1-22.
De Veaux, R., Plantinga, A., Upton, E. (2022). Are the Handicaps Fair? Age and Participation Effects in the Dipsea Race. CHANCE, 35:4, 40-49.
Upton, E.M., Rudolph, A.E., Ward, P.J., Havens, J.R., and Young, A.M. (2022). Extent and Implications of Omitted Ties on Network Measures in a Longitudinal Social Network Survey of People who Use Drugs. Drug and Alcohol Dependence, 109554.
Rudolph, A.E., Upton, E., Young, A.M. and Havens, J.R. (2021). Social network predictors of recent and sustained injection drug use cessation: findings from a longitudinal cohort study. Addiction, 116(4), 856-864.
DiGiorgio, A.L., Upton, E.M., Susanto, T.W. and Knott, C.D. (2020). Wild Bornean orangutan (Pongo pygmaeus wurmbii) feeding rates and the Marginal Value Theorem. American Journal of Primatology, 82(10), e23183
Rudolph, A., Upton, E., McDonald, M., Young, A. and Havens, J., (2020). Peer Influence of Injection Drug Use Cessation Among Dyads in Rural Eastern Kentucky. International Journal of Drug Policy, 85, 102604.
Network Data
Advances in data collection have led to the increased presence and availability of network-indexed data, that is, data that has an inherent structure. Networks are composed of nodes (actors, vertices, egos, units) and edges (relationships, ties, links) and are found in a number of diverse fields including the social sciences, technology, and biology. My research centers on developing novel modeling methods and computational advancements that improve our understanding of this progressively more prevalent data structure.
My methodological research has been motivated by the following two goals: (1) to understand or explain a vertex attribute of interest in a manner that utilizes information from both the network’s structure and pertinent covariates and (2) to offer a logical and flexible framework to model graph edges or edge responses that can capture basic network properties while retaining computational feasibility. My research in these areas extends to the following topics: Bayesian modeling, hierarchical modeling, Bayesian variable selection techniques, exponential random graph models, and latent space network models.
Figure 1: Liz’s PhD Support Network, 2019
Social Network Analysis
Have you ever heard the sayings “Bad association spoils useful habits” or “Birds of a feather flock together”? Are these true? Can we prove or disprove the idea that the people we hang out with, our social network, has an impact on the decisions we make in life? Do we change our own behavior to match that of our friends, or do we self-select friends based on similiarities in behavior? Questions like these can be analyzed with social network analysis. We can quantitatively assess the social structures in a network and the corresponding behaviors of individuals that compose said network. We can also draw inferences on the relationships between the social structures and behaviors.
My recent research in this area attempts to understand the independent and combined influence of individual and network (egocentric and sociometric) factors on risk behaviors among marginalized populations. I am also analyzing the data collection methods used to gather social network data on these populations in order to improve the accuracy of corresponding analyses.
Education and Pedagogy
I truly enjoy sharing my passion for statistics with others. Thus, my future research plans also include the development of pedagogical tools to better serve the needs of our students studying statistics and data science in the undergraduate classroom. I would love to tackle this endeavor through a social network analysis lense. Project idea anyone?
For Students
Interested in a summer research position or an honors thesis? I have included some possible topics below but there are endless possibilities. Feel free to set up a meeting if you are interested in discussing.
Possible topics:
-Regression models for network data: how can we incorporate network structure (and dependence) in our regression framework when modeling a vertex-indexed response?
-Identify effects shaping network structure. For example, in social networks, we often see transitivity (we have a mutual friend, thus, we are more likely to be friends). How can we capture or test this effect, and others, in a regression framework when modeling edge-indexed responses?
-Extending models for multilayer networks. Current methodologies combine edges from multiple networks in some sort of weighted averaging scheme. Could a penalized multivariate approach yield a more informative model?
-Developing algorithms to make inference on large networks more efficient.
-Any topic in linear or generalized linear modeling (including mixed-effects regression models, zero-inflated regressions, etc.).
-Applied statistics research. In collaboration with a scientist or social scientist, use appropriate statistical methodology to answer an interesting scientific question.
Possible colloquium topics:
-Any applied statistics research project/paper
-Topics in linear or generalized linear modeling
-Network visualizations and statistics
-Network sampling methods
-Networks as applied to…anything!