Advisory Council on Ethics and Modernization of Microdata Access - Meeting minutes for June 15, 2021

Date and location

June 15, 2021, 8:30 a.m. to 12:30 p.m.
Virtual meeting via Zoom

Members

• Anil Arora, Chief Statistician of Canada (Chair)
• Chantal Bernier, National Practice Leader, Privacy and Cybersecurity, Dentons Canada LLP
• David Robichaud, Ethics Professor, University of Ottawa
• Len Garis, Researcher and Adjunct Professor
• Mary Dawson, Lawyer, former Conflict of Interest and Ethics Commissioner
• Robert Gordon, Executive Director, Canadian Cyber Threat Exchange

Regrets

• Cal Marcoux, Chief Information Security Officer, Canadian Institute for Health Information

Statistics Canada guests/support

Jacques Fauteux, André Loranger, Greg Peterson, Lynn Barr-Telford, Martin St-Yves, Leila Boussaïd, Eric Rancourt, Tom Dufour, Janique Godin, Janice Keenan, Jeff Latimer, Nathalie Brault, Li Xue, Sevgui Erman, Susie Fortier, Yanick Beaucage, Keven Bosa, Martin Beaulieu, Geneviève Jourdain, Marc St-Denis, Farnaz Ahanin, Alexa Tupy

Introductions and Opening Remarks

Anil Arora, Chief Statistician of Canada

Mr. Arora opened the meeting by welcoming members to the fourth biannual meeting of the Advisory Council on Ethics and Modernization of Microdata Access (ACEMMA), and thanking them for their continuous advice and guidance as the agency continues to modernize. He noted that their skills and expertise are of utmost importance to Statistics Canada (StatCan), and are essential as the agency continues to determine how best to serve Canadians with the highest ethical standards and in a responsible manner when dealing with sensitive data sources, while delivering de-identified microdata access to researchers and policy makers that benefits all Canadians.

A recap of the previous meeting was provided. Mr. Arora highlighted ongoing efforts to integrate ethical frameworks and effective governance within all the inputs and outputs of the agency in a systematic and focused manner with transparency to Canadians. He also noted the continued advancements of creating new and innovative methods for collaboration, enhancing the agency's statistical programs, producing more disaggregated data, and making data more open to the research community while adhering to confidentiality requirements and governing frameworks. Additionally, various StatCan ongoing key initiatives were outlined, including developments of the Virtual Data Lab, Virtual Research Data Centre, Client Relationship Management System, accreditation frameworks, and the Confidentiality Classification Tool (CCT) scoring of additional datasets.

Updates on the Virtual Data Lab project

Leila Boussaïd, Director General, Data Access and Dissemination Branch, Statistics Canada
Marc St-Denis, Assistant-director, Data Access and Dissemination Branch, Statistics Canada

Ms. Boussaïd shared updates on the Virtual Data Lab (vDL) project. As the vDL will go into production in October 2021, the goal remains to provide increased access to accredited researchers through the use of authorized workspaces and remote capabilities. She noted that the Protected B Cloud environment should be available for all levels of government, non-profit organizations, and academia, and will provide 24/7 remote access up to CCT 7 score data. Pilots are presently being conducted to gather lessons learned and to work towards an overall improved end-state of the environment. Mr. St-Denis addressed the project's plan and timelines regarding the transition of all current existing partners to the vDL Cloud environment.

Council members expressed support and enthusiasm for the project, and agreed that advancements in technology allow to achieve greater access to data in a secure and virtual way. Members inquired about the security approach for this project, wanting to ensure that both external and internal threats were addressed. A discussion took place on options to mitigate risks such as accreditation and real time monitoring.

Updates on ethical considerations in the acquisition of administrative data, Trust Agenda, and social license

Eric Rancourt, Director General, Modern Statistical Methods and Data Science Branch
Greg Peterson, Assistant Chief Statistician, Economics Statistics Field
Janice Keenan, Director General, Communications Branch
Jacques Fauteux, Assistant Chief Statistician, Strategic Engagement Field, Statistics Canada (Moderator)

Mr. Rancourt shared progress with the Office of Privacy Commissioner (OPC) and accomplishments that have been achieved for data ethics and privacy at StatCan, such as the creation of an internal Data Ethics Committee and Secretariat, the increased execution of data ethics reviews, development of a Framework for Responsible Machine Learning, work to build a data sensitivity scale that will better inform managers about their programs, enhancement to the Necessity and Proportionality Framework, and availability of training in Data Ethics. Mr. Rancourt emphasized the importance of identifying potential social biases, stigmatization and data sensitivity, and asked council members the following questions:

• Do we have a sufficiently elaborated ethical framework to support statistical activities?
• Are there issues not covered or not sufficiently covered?
• How can we move forward with statistical programs and maintain society's support?

Mr. Peterson led the discussion on the acquisition of data. He indicated that the pandemic has brought forth new data requirements and he highlighted the following three: (1) housing prices and implications on citizens; (2) equitable recovery and growth, and; (3) inflation and the acquisition of better information on consumer credit. Mr. Peterson highlighted the engagement work that is being done to continue to build relationships with various stakeholders to determine how best to serve their needs and collaborate with them on data acquisition.

Ms. Keenan led the discussion on the Trust Agenda and social license. A number of initiatives are underway, including the creation of a Communications and Engagement Strategy, enhancement of the agency's Trust Centre to make it more interactive, and the development of enhanced social acceptability metrics. Ms. Keenan emphasized the importance of continued engagement with the Canadian public to enable them to better understand the impact and relevance of StatCan data in their daily lives and their communities.

Council members agreed that there is a need to communicate and engage with Canadians to inform them, and asked what the agency can do to promote the work of StatCan and connect with people who do not visit the website. An extensive discussion on social license took place. Members suggested that partnerships could help improve understanding of the work of StatCan and can help mitigate concerns of some stakeholders. Members would like to see StatCan showcase examples of when society has received helpful information as a result of the agency's statistics, and to continue building on the notion of trust. Mr. Arora thanked the members for their input. He requested if the committee would take on the role of advising on the agency's state of readiness, as there is a desire for the committee to provide guidance regarding the agency moving forward and reasonable next steps in its initiatives. Members unanimously agreed to take on the role.

Managing the landscape in providing disaggregated data to researchers: how can StatCan increase granularity

Nathalie Brault, Director, Centre for Special Business Projects, Statistics Canada
Li Xue, Director, Social Analysis and Modelling Division, Statistics Canada
Jeff Latimer, Director General, Health, Justice, Diversity and Populations Branch, Statistics Canada
Lynn Barr-Telford, Assistant Chief Statistician, Social, Health and Labour Statistics, Statistics Canada (Moderator)

Mr. Latimer began the discussion with a presentation on the intended approach for the agency's Disaggregated Data Action Plan that was recently announced in the 2021 Budget. The plan will focus on contributing to a more equitable Canada by collecting, analyzing, and disseminating disaggregated data to improve insights and decision-making, and support more representative data collection and enhance statistics on diverse populations. Mr. Latimer noted that to achieve the plan's objective, StatCan is engaging with government partners and communities as part of its efforts to maintain social licence, working collaboratively with key stakeholders to collect and acquire new data, and improving access to the data and insights.

Ms. Brault continued the presentation in discussing disaggregated data in a business context. COVID-19 was a driver for increased demand and use of disaggregated data. She touched on the Canadian Survey on Business Conditions, launched in summer 2020. The survey collects information on businesses to identify potential emerging issues, to maintain a regular pulse on business intentions and sentiment and, in the short term, assess economic recovery. Ms. Brault noted that the agency is currently working with organizations such as Chamber of Commerce, the Canadian Business Resilience Network, as well as Black North to ensure that our products and services continue to be relevant to as many Canadians as possible.

Ms. Xue concluded the presentation noting that the goal is to enhance disaggregated data and facilitate research on complex issues and intersectionality. This will be achieved through the innovative approaches to disaggregated data and information (i.e. modeling, geomatics), and engaging with partners on collaborative pilot projects.

Council members were invited to provide input on the following questions for consideration:

• Effective engagement with the communities served by the disaggregated data plan is key. Respondents, businesses owners and individuals alike, are reluctant to self-identify. How can we improve StatCan's engagement while respecting privacy and ethical considerations?
• Canada is a mosaic with known regional diversity. What approaches does the committee recommend for reflecting diversity in an inclusive manner?
• Any general advice or specific issues with regards to the planned direction on the Disaggregated Data Action Plan.

Mr. Arora noted the important role that the agency can continue to play as a data steward and the importance to lead by example. Standards will also play a key role in enabling a secure infrastructure with tools that build trust and capacity.

Council members noted the importance to make the distinction between the use of disaggregated data for policy making and for informing Canadians. With disaggregated data, more information will become available regarding which populations are underprivileged and the disproportionate impact on some groups in society. Members cautioned that the information can be accompanied with complicated ethical questions and issues, however, that the benefits are bigger than the drawbacks and that communicating these benefits and intentions with the public and how we engage will be key. There was general support for the strategic direction being pursued on the Disaggregated Data Action Plan (DDAP).

Responsible Machine Learning

Sevgui Erman, Director, Data Science Division, Statistics Canada
Yanick Beaucage, Chief, Data Science Division, Statistics Canada
Keven Bosa, Chief, Data Science Division, Statistics Canada
Susie Fortier, Director, International Cooperation and Methodology Innovation Centre, Statistics Canada
Martin Beaulieu, Chief, International Cooperation and Methodology Innovation Centre (Data Ethics Secretariat), Statistics Canada
André Loranger, Assistant Chief Statistician, Strategic Data Management, Methods and Analysis, Statistics Canada (Moderator)

Mr. Loranger noted that the Data Science Division and the Data Science Network had been created to help conduct ethical reviews and further advance data science techniques at StatCan. Mr. Bosa continued the discussion with a presentation on behalf of the group. He highlighted that as the use of artificial intelligence (AI) and machine learning methods continues to expand, there is a need for a framework and review process of machine learning at StatCan. Having a framework and review process preserves confidentiality, ensures transparency, maintains Canadian trust in StatCan as their data stewards, and helps produce statistics of quality. Mr. Bosa noted that the agency created a Framework for Responsible Machine Learning Processes that provides guidelines for the evaluation of applications for all statistical programs and projects using machine learning algorithms. The framework focuses on the following four themes: respect for people, respect for data, sound application, and sound methods. Mr. Bosa concluded the presentation and invited council members to provide input on the following questions for consideration:

• Are there any missing elements or gaps?
• Should we move the ethical peer review earlier in the project lifecycle?
• Do you recommend other areas of research in view of enhancing the Framework?
• Should we do more socialization of the Framework across Government of Canada? If so, how?

Council members agreed with the importance of all four components of the Framework, and noted that it is a tool that can be useful and informative and help build credibility into the agency's overall process. Members cautioned that the application of new AI technologies have ethical implications, such as potential issues of algorithmic bias. To address potential ethical implications, it is crucial for the agency to carefully review the data elements being used. Regarding social buy-in, members noted that remaining transparent will be key.

Closing Remarks

Anil Arora, Chief Statistician of Canada

Mr. Arora thanked members and invited them to share their final thoughts.

Advisory Council Action Items

• Next meeting: January 2022 (tentatively)
• Provide Council Members with an update on the Virtual Data Lab and modernization initiatives
• Questions/remarks from Council Members for Statistics Canada to consider:
  • In what ways can we provide more access to granular information?
  • Increasing data access to researchers is a central objective to pursue. How do we maintain momentum and stay the course in the agency's modernization efforts?
  • What opportunities exist to continue holding open dialogue with Canadians, and increase communication and engagement to provide a better understanding of how StatCan data shapes policy and contributes to decision making?
  • As new technologies and methods are applied at StatCan, how can we ensure the continued application of ethics and transparency?
• Council Members could be asked to provide some quotes or contributions.