By Malcolm Cunningham, Institute of Cognitive Science, Carleton University
Summary of the Issue
Following the recent Canadian federal election, there has been an upsurge of public discussion about reinstating the mandatory long-form census. The previous government canceled it in 2010 citing privacy concerns and coercive data collection methods (people compelled under threat of prosecution to complete survey questions) as main reasons for its demise. The debate, it seems, is quite simple; either you support reinstatement or you do not. A debate cast this way, however, reflects only particular interests—those of organizations that consume census data and the census creator itself–at the expense of people who provide data. I argue that the data provider perspective is largely missing from the national discussion and needs to be included. Moreover, now is a fortuitous time to consider a new viewpoint as the newly elected federal government brings promise of change. We have an opportunity not only to reinstate the long-form census but also to improve it by adding this new perspective and redressing rather dated legislation.
The 1971 Statistics Canada Act remains the basis of the mandatory long-form census. Yet since 2000, Canadian public institutions (e.g., hospitals, schools, universities, colleges) have recognized and acted on the need for rigorous research protocols to guide ethical collection of human information. Thus while the Statistics Canada Act languishes in an old interpretation with dated protections for people who provide census information, publically funded research has moved on. Simply reinstating the mandatory long-form census, therefore, is to re-admit the status quo without question. So while this blog acknowledges a need for large public data it does so by challenging readers to consider how we might improve protections for data providers in both Act and census. The goal is to provoke reasoned discussion about ethical implications of census taking in the 21st century that lead to constructive solutions.
Who Am I?
I am an adjunct-researcher in the Institute of Cognitive Science at Carleton University. I received a PhD in measurement and evaluation from the University of Toronto in 2012. I use large-scale public data for research purposes and have experience with developing research proposals requiring protections for people supplying information.
What Perspective Does This Blog Represent?
There are three recognizable stakeholders in the mandatory long-form census debate–the survey creator, data consumers, and data producers. Perspectives of two stakeholder groups completely dominate national discussion. There are well-positioned arguments from the census creator, Statistics Canada, and various groups whose daily business involves consuming large amounts of census data (e.g., media organizations, corporations, researchers). These arguments often speak to the need for mandatory census data collection as a vital tool in our understanding of a changing Canada (although, notably, Statistics Canada has recently softened this position). As these groups represent important vested interests in maintaining and even expanding census creation and use, it should come as no surprise they strongly support the status quo, reinstatement. But theirs is already a well known position and this blog would add little by reflecting their views. Representing perspectives of the largely silent group that produces this data, however, is another matter. You and I are the ones asked to fill out the forms. We are more vulnerable to data breaches and clearly have the most to gain or lose when decisions about the mandatory long-form census are finalized. By reflecting a hitherto unreported perspective, therefore, this blog represents these largely silent stakeholders in an effort to introduce a new dimension to the national discussion.
This is not a political blog. Although the long form census was terminated under the previous federal government I am in no way associated with any government players or the party involved with its demise. Nor am I associated with any of the newly elected federal government players or their party. The truth is, I am not associated with any party. My interest is purely with the substantive issues and crafting a better and more durable solution.
The Issue in Brief – Research Ethics and the Long-Form Census
Statistics Canada does not conform to a modern understanding of research ethics. A search of the Statistics Canada website reveals an appeal to potential participants about the importance of complete information and reassurance that data is collected only in the strictest of confidence. Participation in two census forms is currently mandatory (Population and Agriculture) where one’s failure to comply, neglect to answer, answer falsely or knowingly deceive may result in fines or even imprisonment. There is far more attention paid to data–the products of various surveys—on the Statistics Canada website than to the participants who supplied this data. An ombudsman exists specifically for businesses who complete surveys but not for individual Canadians. Indeed, beyond statements about the importance of complete data, Statistics Canada says little about purposes of surveys, what they entail, risks and benefits, or potential uses of collected data.
At present, whether or not Statistics Canada should conform to principles consistent with the Tri-Council policy on research ethics is an open question. Some may say that Statistics Canada data is strictly collected for census purposes and not research. It follows that such data should be exempt from protocols otherwise imposed on public research. But this is really a disingenuous argument. Although there is no question Statistics Canada data is used for census purposes, researchers from coast to coast use the same data to conduct research and have done so for decades. To say that any datum will not be used for research purposes, therefore, is to wilfully ignore the integral use of all Statistics Canada data. Even the corporation’s own assurances about confidentiality grows increasingly tenuous as our ability to link new databases together affords a level of analytical granularity unimaginable in 1971. So the question of Statistics Canada’s adoption of a modern interpretation of research ethics is arguably far more about a growing urgency to protect those who participate in supplying survey information in today’s world and less about maintaining definitional boundaries that were once more meaningful.
Adoption of a modern research ethics policy consistent with the Tri-Council document would signal important changes;
- Survey design, data collection, analysis, and data availability would be subject to approval by an independent ethics board (every university in Canada currently has such a board). This board would serve as the final arbiter in situations were there are questions about ethics.
- Statistics Canada would redesign the website to more directly appeal to survey participants. It would offer more information about purposes of surveys (including potential research and commercial uses), what they entail, and what risks or benefits might accrue
- Direct contacts would be available at Statistics Canada in the event participants have questions or concerns
- Unless successfully argued in front of an independent ethics board, mandatory data collection would a thing of the past. Only in exceptional circumstances should the corporation be able to successfully argue for coercive data collection methods.
- Voluntary and informed census participation would likely signal a change in the corporation’s approach to survey design, administration and analysis. Attention to design details (e.g., shorter surveys, more attractive formats) would be related to completion rates. It may be necessary to include incentives to entice greater compliance. Large representative sampling designs may have to give way to smaller targeted (perhaps linked) designs.
- Statistics Canada currently generates revenue from organizations and individuals wishing to access to data repositories. A modern ethics protocol would require the corporation to be completely transparent, signalling potential conflicts of interest.