My current research in causal inference is about causation in legal contexts. I have focused on studying the probability of causation (e.g., What is the probability that an individual’s cancer was caused by his exposure to a chemical?), and I have applied this framework to evaluate the use of scientific evidence by expert witnesses in cases of Shaken Baby Syndrome/Abusive Head Trauma. Under the guidance of Edward H. Kennedy, I recently developed a novel, influence-function-based estimator for the probability of causation, which allows for nonparametric estimation of the nuisance functions and still provides confidence intervals under weak structural assumptions. I intend to expand the definition of my causal parameter so it can be properly applied to more realistic settings and at some point be used in court. I am part of the Penn Center for Causal Inference, a group out of statistics and biostatistics, and one of the most stimulating reading groups I have participated in.
Click here to see an app I made to visualize the estimated probability of causation for a randomized trial from Kenya.
My current research in forensic science is about the validity of statistical statements in forensic analysis, about which I have testified as an expert witness in court. I am working on several projects including studying:
- Whether blind proficiency testing yields different results than open (non-blind) testing in the Pittsburgh forensic laboratory, the Allegheny County Medical Examiner’s Office,
- How contextual information about a case influences a forensic examiner’s decision (as well as an algorithm’s decision) about whether two bullet cartridges were fired by the same firearm, and
- What are the error rates of tool mark analysis using a black box study.
I intend to continue doing research in this field in order to reduce the rate of errors in the criminal justice system. I am part of CSAFE, an innovative and impactful group of statisticians and other researchers working to improve the foundations of forensic science. I am also part of the Quattrone Center, an interdisciplinary group housed in the Penn law school that promotes data-based research for the fair administration of justice.
Here is a talk I gave at the Newton Institute for Mathematical Sciences at the University of Cambridge, during the workshop on Bayesian Networks and Argumentation in Evidence, entitled Shaken Baby Syndrome on Trial. (Here’s the link in case you can’t see the video.)
- Cuellar M, Short fall arguments in court: A Probabilistic Analysis, 50 U. Mich. J. L. Reform 763 (2017). Link, video.
- Cuellar M; Causal reasoning and data analysis: Problems with the abusive head trauma diagnosis, Law, Probability and Risk, 2017; 16(4): 223–239. Link.
- Cuellar M; Trends in Underreporting of Marijuana Consumption in the United States, Statistics and Public Policy, 2018,2018; 5(1): 1-10. Link.
- Moral J, Hundl C, Lee D, Neuman M, Grimaldi A, Cuellar M, Stout P; Implementation of a blind quality control program in blood alcohol analysis. Journal of Analytical Toxicology.
- Cuellar M, Gonzalez C, Dror I; Human and algorithmic similarity judgments in forensic firearm comparisons (2018).
- Cuellar M, Kennedy E; A nonparametric estimator for the probability of causation. (2018) arxiv:1607.02566 Link, video.
Invited book chapters
- Cuellar M; “Causes of Effects and Effects of Causes,” Chapter in Statistics in the Public Interest: In Memory of Steve Fienberg, Springer (2019).
- Cuellar M; “The Flawed Statistics of Shaken Baby Syndrome/Abusive Head Trauma,” Chapter in The Shaken Baby Syndrome Controversy (2019) (to submit in May 2019).
- Cuellar M; Mentch L, Spiegelman C (2018), “Association does not imply discrimination: Flawed analyses that lead to wrongful convictions,” in Handbook of Forensic Statistics, Chapman & Hall/CRC Handbooks of Modern Statistical Methods, S 1 Chp 4, forthcoming.