Using Government Statistical Systems to Target Businesses and Vulnerable Population Groups: Examples and Issues William Seltzer Fordham University Presentation at session on Data Confidentiality and Privacy Issues GAO brown bag seminar, Tuesday, June 11, 2013
I. Overview My topic today is an aspect of data confidentiality not usually addressed: the deliberate diversion of data and data systems, routinely used for statistical purposes, to target businesses and vulnerable population groups as part of government investigatory, intelligence, regulatory or prosecutorial activities.
Overview - 2 • I shall introduce my topic by briefly identifying five examples such diversions or targeting drawn from US experience and three from such experiences in other countries. • I will then review some concepts that are useful in looking at these examples and in thinking about the problems they raise. • With these concepts in mind I return to these eight examples first to better understand them and then to say something about their impact on future government statistical work. • Finally, I will say something about prevention and mitigation.
Overview - 3 • I could not attempt to cover so much ground in the brief time we have today, but for the extensive body of research that my colleague, Professor Margo Anderson of the University of Wisconsin-Milwaukee and I have carried out on the topic over past decade and a half. • Copies of papers presenting the results of this research, or links to them, may be found at https://pantherfile.uwm.edu/margo/www/govstat/integrity.htm References to a number of them were also cited in the announcement for this session and are reproduced at the end of this PowerPoint.
II. Eight Examples • A. From US Experience: Locating Japanese Americans in World War II Oil pricing and anti-trust investigations after the 1956 Suez Crisis and resulting oil embargo St Regis Pulp and Paper anti-trust investigation National Center for Education Statistics after 9-11 Arab Americans, the 2000 Census, and the War on Terror
II. Eight Examples - 2 • B. Experience from Elsewhere: Jewish Holocaust in World War II (two examples: Germany and the Netherlands) South Africa and the beginning of Apartheid in 1950 China and the Cultural Revolution, 1966-1967 These eight cases are a selection from a much larger set of examples where a system used to compile statistics was transformed into a tool of investigation or repression.
II. Eight Examples - 3 The full set of instances so far identified includes a range of both democratic and totalitarian countries, although the scale and extent of destructiveness tends to far greater among the totalitarian countries. Furthermore, in the case of democratic countries such diversions have usually arisen in times of special national stress.
II. Some Sets of Concepts to Better Understand These Examples A. Purpose of data collection • Statistical (e.g., population census, labor force survey) • Administrative (e.g., data on building permits filed) • Mixed (e.g., birth and death registration) B. Data Sources • Censuses • Sample surveys • Registration systems and other administrative reporting systems
II. Some Sets of Concepts - 2 C. Data Types • Macro data (data aggregated to large areas) • Meso data (data aggregated to small operationally useful geographic areas) • Micro data (data for individual population unit or firm) D. Type of Statistical Confidentiality Law • No law • Weak law (only protects against “unauthorized” disclosures) • Strong law (individual protection only) • Strong law (both individual and group protection)
III. Methods of Targeting and Disclosure • A. US Examples Japanese Americans in World War II (1940 Population Census carried out initially under the strong confidentiality provisions of the Title 13 of census act; meso and micro data) Energy Statistics pricing and anti-trust investigations after the 1956 Suez crisis (routine administrative data on oil production and prices provided on a voluntary basis to the US Bureau of Mines; micro data) St Regis Pulp and Paper anti-trust investigation (data provided in the economic census to the U.S. Census Bureau under strong confidentiality provisions of Title 13 of the census act; micro data)
III. Methods of Targeting and Disclosure - 2 • US examples (continued) National Center for Education Statistics after 9-11 (sample survey and administrative data gathered under the protection of the NCES strong confidentiality act; micro data) Arab Americans, the 2000 Census, and the War on Terror (2000 Population Census; apparently meso data) • Examples from other countries Jewish Holocaust in World War II: Germany (1939 Population Census; micro data) and the Netherlands (population registration system; micro data) South Africa and Apartheid (1950 Population Census and population registration system; micro data) China and the Cultural Revolution (population registration system; micro data)
IV. Known Consequences for the Agency Involved or the Statistical System • A. US Examples Japanese Americans in World War II (First, patriotic pride, then nothing, then denial, then drip by drip disclosure of what happened, apparently some residual level of public mistrust; improved efforts at data stewardship by the Census Bureau) Energy Statistics pricing and anti-trust investigations (Federal statistical system unable to prevent disclosure encouraging Federal Trade Commission and the Justice Department to move more aggressively to seek disclosure of business data)
IV. Known Consequences for Agency or Statistical System - 2 • US examples (continued) St Regis Pulp and Paper anti-trust investigation (US Supreme Court decision setting aside the statistical confidentiality protections of Title 13 in the circumstances of the case, then a data provider revolt that lead to a strengthening of Title 13 that effectively counteracted the Supreme court decision)
IV. Known Consequences for Agency or Statistical System - 3 • US examples (continued) National Center for Education Statistics after 9-11 (Patriot Act and successor legislation turned the NCES statistical confidentiality law into a weak one; initial and ongoing impacts on response rates or data quality unclear; eventually unknown degree of legislative amelioration from the 2002 Confidential Information Protection and Statistical Efficiency Act [CIPSEA]) Arab Americans, the 2000 Census, and the War on Terror (Public embarrassment other impacts not clear)
IV. Known Consequences for Agency or Statistical System - 4 B. Examples from other countries Jewish Holocaust in World War II: Germany and the Netherlands (After World War II, very strong statistical confidentiality laws, in many European countries, including extending confidentiality protections from individuals to groups and limitation of questions about vulnerable population groups; in Germany growing backlash lead to cancellation of population censuses from the 1980s to 2011)
IV. Known Consequences for Agency or Statistical System - 5 • Examples from other countries (continued) South Africa and Apartheid (First, Census and Statistics agency proud and pleased with added resources it received to establish population register based on the micro data on race in the 1950 Census so as to enable the new government to institute apartheid; eventually the agency overwhelmed by the task and fell into several decades of neglect, only restored in the post apartheid era) China and the Cultural Revolution (Impact on statistical system not known beyond the fact that a few years after the cultural revolution ended the variable “class status”, which had been used to target many victims in the cultural revolution, was removed from the population registration system)
V. Some Methods of Prevention and Amelioration • A. Care about gathering 100 per cent data on “sensitive” topics and related classification categories • B. In general, sampling safer than complete count information • C. Decentralization of data storage • D. Frequent exchange of information and discussion of confidentiality and disclosure issues among statistical agencies, with data providers, and with supporters in Congress, professional societies, and the press
V. Some Methods of Prevention and Amelioration - 2 • E. Orientation of intelligence and investigatory agencies on quality issues in large-scale statistical systems • F. Training in statistical ethics and methods of disclosure control • G. Careful review of tabulation, analysis, and data dissemination plans from legal, ethical, statistical policy perspectives, including the reputation of the agency as a responsible data steward • H. Research into what happened and its impact
Thank you William Seltzer email@example.com firstname.lastname@example.org
Some Relevant Writings and Presentations • William Seltzer, The Role of Ethics in a Federal Statistical Agency, A presentation at the US Census Bureau, June 9, 2010 • Margo Anderson and William Seltzer, "Federal Statistical Confidentiality and Business Data: Twentieth Century Challenges and Continuing Issues," Journal of Privacy and Confidentiality 1 (Spring 2009), pp. 7-52; Comment on Article by Anderson and Seltzer, by C. L. Kincannon, pp. 53-54; Rejoinder, by M. Anderson and W. Seltzer, pp. 55-58. • William Seltzer, “Statistical Confidentiality and UK Population Censuses: Recent History and Prospects.” Paper prepared for presentation at the International Statistical Institute, 57th Session, 2009.
Some Relevant Writings and Presentations - 2 • William Seltzer and Margo Anderson, "Using Population Data Systems to Target Vulnerable Population Subgroups and Individuals: Issues and Incidents," in Statistical Methods for Human Rights, Jana Asher, David Banks, and Fritz S. Scheuren, eds. (New York: Springer, 2008), pp. 273-328. • William Seltzer and Margo Anderson, "Census Confidentiality under the Second War Powers Act (1942-1947)," Paper prepared for the Annual Meeting of the Population Association of America, March 30, 2007, New York, New York.
Some Relevant Writings and Presentations - 3 • Margo Anderson and William Seltzer (2007). "Challenges to the Confidentiality of U.S. Federal Statistics, 1910-1965."Journal of Official Statistics. 23(1): 1-34. Incident Table. • Margo Anderson and William Seltzer (2006). “Discussion of Habermann, ‘Ethics, Confidentiality and Data Dissemination.’”Journal of Official Statistics. 22(4): 641-49. • Margo Anderson and William Seltzer. "Federal Statistical Confidentiality and Business Data:&n! bsp; Twentieth Century Challenges and Continuing Issues," Paper presented at the meeting of the Federal Committee on Statistical Methodology, Arlington, VA, November 2005.
Some Relevant Writings and Presentations - 4 • William Seltzer, "Official Statistics and Statistical Ethics: Selected Issues," paper prepared for the International Statistical Institute, 55th Session, 2005. • William Seltzer, "On the Use of Population Data Systems to Target Vulnerable Population Subgroups for Human Rights Abuses," Coyuntura Social, No. 32, June 2005, pp. 31-44. • William Seltzer and Margo Anderson, "NCES and the Patriot Act," Paper presented at the Joint Statistical Meetings, New York, August 2002. Abridged version available in 2002 ASA Proceedings, pp. 3153-56.
Some Relevant Writings and Presentations - 5 • William Seltzer and Margo Anderson, "The Dark Side of Numbers: The Role of Population Data Systems in Human Rights Abuses."Social Research, Summer 2001, Vol. 68 Issue 2, pp. 481-513. • William Seltzer, "U.S. Federal Statistics and Statistical Ethics: The Role of the American Statistical Association’s Ethical Guidelines for Statistical Practice," Paper based on a presentation at the Washington Statistical Society, February 2001.
Some Relevant Writings and Presentations - 6 • William Seltzer and Margo Anderson, "After Pearl Harbor: The Proper Use of Population Data in Time of War," Paper presented at the Annual Meeting of the Population Association of America, March 30, 2000, New York, New York. • William Seltzer, "Population Statistics, the Holocaust, and the Nuremberg Trials,"Population and Development Review, Vol. 24, No. 3. (Sep., 1998), pp. 511-552. • SOURCE for these and other relevant materials, see https://pantherfile.uwm.edu/margo/www/govstat/integrity.htm