Philip Stark studies inference problems and uncertainty quantification with applications ranging from cosmology, earthquake risk, particle physics, and climate to ecology, food security, election integrity, lottery fraud, and gender bias. He focuses on nonparametric and exact inference tailored for specific scientific goals. He developed methods for auditing elections now in law in California, Colorado, Nevada, Virginia, and Rhode Island. He developed or co-developed methods that are part of the data pipelines of the Øersted geomagnetic satellite and the Global Oscillations Network Group. He has consulted for major corporations and for the U.S. Department of Justice, the Federal Trade Commission, the U.S. Department of Agriculture, the U.S. Census Bureau, the U.S. Department of Housing and Urban Development, the U.S. Department of Veterans Affairs, the California Attorney General, the California Highway Patrol, and the Illinois State Attorney. He has testified to the U.S. House of Representatives Subcommittee on the Census; the State of California Senate Committee on Elections, Reapportionment and Constitutional Amendments; the State of California Assembly Committee on Elections and Redistricting; and the State of California Senate Committee on Natural Resources. See www.stat.berkeley.edu/~stark/bio.pdf
In the News
Any way you count it, the fastest-growing major at UC Berkeley by far is one that long slumbered in obscurity: statistics.
Since 1965, California counties have been required to hand tally one percent of all ballots after an election to validate the machine count, despite the fact that available auditing techniques lack any statistical basis. UC Berkeley's Philip Stark has now provided statistically sound methods for conducting these audits, and a proposed bill, AB 20203, will establish a statewide pilot program to test these methods.