Home » Celebrating Black History Month

Tommy Wright

1 February 2020 2,611 views No Comment

Tommy Wright

Affiliation: US Census Bureau
Educational Background: PhD, Statistics, The Ohio State University; MS, Mathematics, University of Tennessee; BS, Mathematics, Knoxville College

Since joining the US Census Bureau in January 1996 as a research mathematical statistician with an interest in probability sampling, Tommy Wright has led the Center for Statistical Research and Methodology (CSRM) (formerly Statistical Research Division). CSRM’s collaborative research focuses on challenges in providing official statistics on the behavior and condition of the nation’s people and businesses. Some of Wright’s current personal application interests include expressing uncertainty in overall rankings based on sample surveys, assessing the variability in census counts treated by a disclosure avoidance algorithm, and thinking about a role for big data with official government statistics.

Between 1979 and 1996, Wright was a research staff member at the Oak Ridge National Laboratory Mathematical Sciences Section, where his collaborative research focused on probability sampling and the design of sample surveys for large national energy-related studies.

Wright has more than 35 years of experience teaching statistics and mathematics at Knoxville College; University of Tennessee-Oak Ridge Graduate Program; University of Tennessee, Knoxville; and—most recently—Georgetown University as adjunct faculty since 2009. He was an ASA/NSF/Census Research Fellow (1993–1996) pursuing research in using probability sampling methods to improve the constitutionally required decennial census count.

An early interest in mathematics shifted slightly when, as a junior mathematics instructor with an MS degree in mathematics (no background in statistics) at a small college, Wright was assigned to teach the course in probability and statistics when all other faculty members declined. But as he learned statistics with the students, he quickly discovered many options regarding what one might do with statistical knowledge, ultimately leading him back to graduate study, but this time in statistics.

Throughout his career, Wright has sought to minimize the complexity and mystery sometimes seen in mathematical and statistical results, which often hide underlying simplicity and beauty.

As an example, recent results bring together his interests in optimal sample allocation, apportionment of the US House of Representatives, and Lagrange’s Identity. During the summer of 2011, Wright realized the problem of how to allocate a fixed sample size among strata seemed similar to the problem of apportioning the fixed number of seats in the US House of Representatives among the 50 states following the census, yet the two problems have different solutions. Only one solution yields integers directly (apportionment method of equal proportions), while the other solution (Neyman allocation) needs to be rounded to integers and, hence, optimality could be lost or at least put in doubt.

Wright showed in a 2012 The American Statistician article that both problems are special cases of a general problem with a general solution that yields the exact method of equal proportions and an improvement on Neyman allocation without the need to round to integers. One can derive Neyman allocation using Lagrange’s identity, and Wright used Lagrange’s identity to give an insightful derivation of the method of equal proportions in a 2014 The American Mathematical Monthly article. In the framework of Lagrange’s identity, Wright showed how to easily obtain several exact integer solutions to extensions of the sample allocation problem in a 2017 Statistics and Probability Letters article. Others have obtained similar extensions, but none makes an insightful connection with Lagrange’s identity.

Wright was born and grew up in Birmingham, Alabama. His professional recognition includes being and elected member of the International Statistical Institute (1989) and fellow of the American Statistical Association (1995).

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)

Comments are closed.