Russell Almond is a senior research scientist in the Automated Scoring
Group at Educational Testing Service in Princeton, NJ. He received
his undergraduate degree from the California Institute of Technology
(Caltech) in 1983 and received his Ph.D. from Harvard University in
Statistics in 1990.
He is a statistician with extensive training and interest in computer science, particularly artificial intelligence. His research interests include representations of uncertainty in artificial intelligence (in particular graphical models and Bayesian networks), use of artificial intelligence techniques in data analysis, human factor design for data analysis and education, Bayesian computation, and statistical model engineering. He has been a regular contributor to both the Uncertainty in Artificial Intelligence and the Artificial Intelligence and Statistics conferences and has served on the program committee for both. In 2005 he chaired the Bayesian Application Workshop at UAI.
Together with R. Mislevy and L. Steinberg, he was a part of the effort to create Evidence Centered Design (ECD)—a process for designing assessments based on explicating the underly evidentiary arguments that support their claims. The ECD framework is broad enough to support not only familiar Item Response Theory models, but also a new class of diagnostic assessment models capable of supporting complex, constructed response tasks based on Bayesian networks. This work won the National Council on Educational Measurement Award for Outstanding Scientific or Technical Contribution to Educational Measurement in 2000.
He has also been part of several large software development efforts since coming to ETS. These include Portal—a tool for managing the knowledge used in the Evidence Centered Design process,—Biomass—a prototype assessment for measuring ability to perform scientific reasoning in the context of genetics,—and ACED—a prototype diagnostic assessment aimed at 8th grade mathematics standards. He is the architect and lead designer of StatShop—a suite of tools for supporting the use of Bayesian networks and other measurement models with the ECD process. Currently, he is working on smaller project involving integrating evidence over time and analyzing constructed response data.
He also served as ETS's representative on the IMS Global Consortium's working group for Question and Test Interoperability (QTI). There he was a key contributor to what is becoming a widely adopted specifications for sharing assessments and item banks as well as assessment results. His contributions include a four process abstract model of assessments (with L.Steinberg and R. Mislevy) which served as a reference model in the construction of these specifications.
For the past several years he has been a volunteer mentor with Science Mentors 1-to-1 a local non-profit, science based, environmentally focused, one-to-one mentor program that inspires, motivates, and empowers under served high school students to develop academic discipline, critical thinking and personal responsibility, in order to be active stewards of their lives, communities, and the environment.
Last modified: May 29, 2009