Though Statistics Canada can never be wholly independent, the quality of the data produced through the census depends upon its purposeful detachment from political influence. Even the wording of a census question can dramatically affect how people will respond to it.

For example, before the implementation of the “population group” question in 1996, Statistics Canada used proximate measures—ethnic identity, language, place of birth—to determine the visible minority population of Canada. The category “Black” was added to the list of examples in the 1986 census in order to improve reporting by Canada’s African-origin population and to meet the need for data to implement the Employment Equity Act.

Simply adding this category as an example not only increased the population count from 30,975 in 1981 to 260,335 in 1986, but also demonstrated that previous census counts were inadequate. Using racial proxies, as had been done prior to 1986, only 72 per cent of Black respondents gave “compatible” ethnic origins when compared to their responses to the population group question in 1996. The use of ethnic identity, language, and place of birth was insufficient for enumerating, for example, the Loyalist-descended Black population in Nova Scotia, which would likely have answered “British” for ethnic origins, “English” as language spoken, and “Canada” as place of birth.

The interference of ideologically-saturated decisions about who and what should be counted, and how this counting should take place, is an affront to the goal of acquiring high-quality and accurate data on the Canadian population. As the Canadian Sociological Association argued in its July 21 letter to Minister Clement, “This decision, involving an Order in Council rather than widespread consultation with stakeholders who depend upon quality census data, eliminates our most comprehensive and accurate source of longitudinal data. The social and economic consequences for Canada and our ‘common good’ are profound.”

Yet, this isn’t the first time or place in which political parties have tried to alter the census. While the census and the statistical agency responsible for it are non-partisan, census results and the policy consequences that arise from them often have incredible effects on partisan politics, linked to the two central concerns of government: money and representation.

Recent censuses in the United States are case in point. After the 1994 election, when partisan politics became particularly hostile, census adjustments to reduce a known undercount became the target of partisan animosities in congressional committees. The two parties were equally concerned about counting or not counting the populations that have historically been missed by the census, which are disproportionately racial and ethnic minorities living in urban areas and are far more likely to vote Democrat than Republican.

In Great Britain, where a privacy movement against the intrusive nature of the census actually exists, Thatcher foreshadowed the Canadian Tories’ argument that census questions are an unwarranted incursion of the state into the private lives of individuals. She found many of the questions proposed for the 1981 census “completely unnecessary” and jettisoned them from the final questionnaire. Further, by eliminating these questions she accomplished her additional goal of reducing the overall cost of the census by 17.5 per cent.

Yet the quality of data produced has been more important than ideology in driving the overhaul of Britain’s statistical system over the past 20 years. As public confidence in UK official statistics eroded, in part due to political meddling, the government centralized responsibility for statistics in a new Office for National Statistics (ONS) in 1996. The government’s 1999 White Paper outlined new accountability and governance arrangements for the ONS, including the creation of the independent UK Statistics Authority. These actions signal recognition of the importance of an autonomous statistical agency to government effectiveness.

Canada is not immune to political battles over statistics. Mulroney’s first government originally cancelled the mid-decade census in 1986 but was forced to reverse its decision due to constitutional provisions requiring that a mid-decade census occur. Statistics Canada has also been influenced by tirades against Big Government from Conservatives and Liberals alike. Its budget was cut during both the government-wide program review in the 1990s and more recent incarnations of slashed government spending.

Comparatively speaking, however, our statistical agency has successfully guarded its autonomy from the ideological agendas of the government in power. The system has some built in protections that help work to protect the autonomy of Statistics Canada. For example, the Minister of Industry cannot overrule the Chief Statistician on issues of confidentiality and the Chief Statistician will defend the budget for Statistics Canada before Parliament. In other government departments Deputy Ministers must accompany the responsible minister.

BreakoutIn light of recent developments, this institutional autonomy must be strengthened. According to the UN Statistical Commission, “Compilation and release of data should be free from political interference, so as to ensure impartiality of the national statistical office.”

An eclectic mix of organizations and individuals have expressed grave concern over how the Harper government’s decision will impact the quality of data produced. Equally disturbing, however, is what this means for the future of the integrity of Statistics Canada.

What would it take to strengthen the autonomy of our statistical system? First and foremost, Canadian political elites must recognize, as British officials did in the 1990s, that high-quality information is essential to the health and vibrancy of our democracy. The statistics produced through the census and other surveys must be objective and accurate in order to inform decision-making inside and outside government and to allow the society to question and judge whether or not the government is acting in its best interests.

Once it is recognized that accurate data will increase the government’s accountability to those it exists to serve, an extensive consultation with stakeholders and the general public must occur. The network of stakeholders tasked with advising Statistics Canada is already well-established. It includes the National Statistics Council, a non-partisan body of highly qualified professionals from a cross-section of Canadian society, the Federal-Provincial-Territorial Consultative Council on Statistical Policy, and the various subject-matter advisory committees. These consultations should be augmented by a general public debate on future arrangements for official statistics in Canada. Public confidence and trust in statistics are of the upmost importance. Too many recent and highly controversial decisions have been made behind the closed doors of the Prime Minister’s Office or in Cabinet meetings.

Finally, depending on the outcome of public and stakeholder consultations, it is worthwhile to consider how the autonomy of the statistical system could be strengthened while maintaining the lines of accountability that characterize good government. Two alternatives are obvious.

The first is to make Statistics Canada an arms-length agency similar to the Office of the Auditor General. This would involve severing the agency’s ties with the Minister of Industry, making the Chief Statistician a position appointed by the Governor General in Council for a term with a set limit and removal procedures that require the approval of Parliament. A truly independent Statistics Canada would report and be held accountable to Parliament rather than the government of the day.

The second alternative—one that the UK has already implemented—is to create an independent body that oversees the compilation and dissemination of official statistics. Rather than acting in an advisory function, as the National Statistics Council does, the Canadian Statistical Authority would operate at arm’s-length from the government and would be directly accountable to Parliament. It would function to oversee Statistics Canada and to provide independent scrutiny, monitoring and assessment of all official statistics produced in Canada.

These policy options are both feasible and viable, but ultimately depend on the will of the government to take the necessity of independent, objective, and accurate official statistics seriously. The recent decisions of the Conservative government have redesigned the relationship between autonomy and ideology in our statistical system. In this forced battle between autonomy and ideology, the victor should be clear. Statistics Canada must maintain its autonomy from political influence.