Past JSM Sessions
Title: Communicating Statistical Disclosure Avoidance Measures to Data Users
Date: Tuesday, August 9th, 2022
Conference: JSM 2022
Sponsor: Privacy and Confidentiality Committee Sponsored Session
Abstract: This panel session will discuss best practices for communicating information about disclosure avoidance methodologies and associated impacts on analyses of data. The importance and timeliness of this issue has become evident by public response to privacy modernization at the Census Bureau. While the formal privacy methods employed for the 2020 Decennial Census are relatively new, synthetic data methods, proposed for future ACS public-use microdata, have become increasingly popular and viable over the last decade, yet there continues to be suspicion that such data are 'fake' and not useful. We can do better to educate users on how the data are changing, what this means for their work, and to better understand the value and advantages of privacy modernization.
Panelists:
Amy O'Hara, Georgetown University
Danah Boyd, Microsoft Research
Michael B Hawes, US Census Bureau
Erica Groshen, Cornell University--ILR School
-----
Title: Collecting, Protecting and Sharing COVID-19 Data and the Stories They Tell
Date: Thursday, August 12th, 2021
Conference: JSM 2021
Sponsor: Privacy and Confidentiality Committee Sponsored Session
Abstract: As a third of the world population has been in lockdown, nearly half a million people have died of COVID-19, and the world's economies have nosedived, policy makers and the public are looking for answers to questions such as ``Might I be infected if I go to work?'' or ``Does wearing a mask help prevent the spread of the disease?'' Answering these questions requires data, on the infectiousness of the disease, on the efficacy of interventions such as lockdowns, distancing, contact tracing, and protective measures. Both repurposing relevant existing data collections, and the quick and effective design of new data collections are top priorities for informed, high quality decision-making. However, the data are sensitive, and so even in these urgent times of a pandemic, the individual identities need to be protected. Access to data by parties other than those originally collecting them has become important, even critical. Balancing privacy (what to ask), confidentiality (what to keep and how to use it), and the societal value given the issues (lives and livelihoods) have become particularly salient. If and how should private companies publish COVID-19-related datasets? Should national state or local governments publish much or little demographic detail on those infected? What is to be done with tracing apps now appearing as society starts to exit the lockdown? Will governments use the data ``for good'' or for possibly nefarious purposes? How will attitudes of citizens change with respect to their willingness to provide data, their knowledge of how data can or should be protected? This session will investigate the current challenges to privacy protection in the times of the COVID-19 pandemic.
Speaker 1: Damien Desfontaines, and co-authors, Google. Privacy in Google Community Reports
Speaker 2: Amaç Herdağdelen, Alex Dow, Bogdan State, Payman Mohassel, Alex Pompe, Facebook. Protecting privacy in Facebook mobility data during the COVID-19 response
Speaker 3: Sarah LaRocca, Facebook. Privacy by design when launching a global COVID symptom survey
Speaker 4: Jennifer Childs, Aleia Clark Fobia, Casey Eggleston and Yazmin Garcia-Trejo, U.S. Census Bureau. Privacy and Confidentiality Views towards Surveys Collected in the Era of COVID-19 Slides
Speaker 5: Frederic Gerdorn, University of Mannheim and Frauke Kreuter, University of Maryland, University of Mannheim & IAB. Privacy attitudes, attitudes to contact tracing Slides
Discussants: Alan Karr, AFK Analytics and Aleksandra Slavković, Penn State University.
-----
Title: Private Data for the Public Good: Formal Privacy in Survey Organizations
Date: Tuesday, August 4th, 2020
Conference: JSM 2020
Sponsor: Privacy and Confidentiality Committee Sponsored Session
Abstract: Differential Privacy will be applied to various 2020 Census Data Products to help protect the privacy of individuals. Where does formal privacy go from there? Survey organizations provide many types of data products, including survey sample data, administrative data, census data, business data, data from partial frames. Research continues, for example, as to applying formal privacy methods to survey data from complex samples, such as handling survey weights, accounting for the additional noise in variance estimation. The impact of the formal privacy movement on various data products such as both large-scale surveys and the numerous smaller-scale surveys remains unclear. Applying formal privacy methods to the survey organization environment brings forth challenges under tight budgets and limiting its impact on the utility of existing data products. This session will begin with discussion on Census Bureau’s formal privacy research agenda for complex survey statistics and then discuss the current state of formal privacy methods in relation to survey organizations.
Speaker 1: Frauke Kreuter, University of Maryland, University of Mannheim & IAB. Will differential privacy affect social science research workflow? Slides
Speaker 2: Quentin Brummet and Brandon Sepulvado, NORC at the University of Chicago. The effect on data utility of using differential privacy techniques to produce estimates of the cost of child care Slides
Speaker 3: Aleksandra Slavkovic, Penn State University. How to achieve optimal statistical inference under formal privacy — a general framework and specific examples Slides
Speaker 4: John Abowd, Census Bureau. The formal privacy research agenda for complex survey statistics Slides
Discussant: Jerry Reiter, Duke University