The census controversy: a travesty of Galilean proportion

In 1633, Galileo was found "vehemently suspect" of heresy. His heretical opinion: holding and defending the belief that the Copernican, heliocentric model of the solar system was true in contravention to the teachings of the Roman Catholic Church. Galileo was placed under house arrest for the remainder of his life and forced to recant, verbally and in writing, any belief in the Copernican model. His book, Dialogue Concerning the Two Chief World Systems, was banned. All because the Copernican model contradicts Biblical scripture. Well, mostly that. The conflict between Galileo and the Church was as much political as scientific or religious. Galileo had made some powerful enemies, people who also opposed Pope Urban VIII, accusing him of being too soft on heretics. So Galileo was in the wrong place, at the wrong time. Nearly five centuries later, the entire affair is one of the most stark examples of the conflict between science and religion.

It was an unfortunate conflict, an unnecessary conflict. Whether science and religion are irreconciliable or incompatible is a much larger debate than I can discuss here, but in this case the conflict seems minor. Galileo was not a villain attempting to derail the Church; he was a good Catholic, earnest in his belief that the pursuit of truth through empirical study and mathematics was a form of devotion to God and appreciation for God's handiwork. The Copernican view may have seemed threatening at the time, but only a century elapsed before the Church removed most pro-Copernican books from its list of banned texts. What happened to Galileo did not have to happen, but it did, a conjunction of personalities and politics clashing to form the zeitgeist of 17th-century Italian science and culture. It is a sad episode in history, for both science and religion, and now we try to move forward.

Or not. Watching today's session of the Standing Committee on Industry, Science, and Technology reminded me of Galileo's trial. There was something eerily familiar about watching the committee grill Munir Sheikh, the former head of Statistics Canada who resigned over how the media represented StatsCan's role in this matter; Ivan Fellegi, former Chief Statistican for Statistics Canada; Don McLeish, president of the Statistical Society of Canada; and many other experts. In both situations, you have an institution in power ignoring facts in the name of political and ideological expediency. While Sheikh et al. aren't quite being forced to recant, there is a comparable intensity and pressure coming from some of the committee members (mostly from the Conservative MPs, but not completely).

It was disturbing to watch David Anderson try to force Sheikh and Fellegi to comment on the content of long-form census questions. As these two tried to make clear, Statistics Canada is not responsible for choosing the questions, only for administering the census. If the government is concerned about the privacy implications for the long-form census questions, maybe it should do something about the questions.

Sheikh, Fellegi, and the other statisticans all agreed that a voluntary census would be biased. The notable exception seemed to be David Tanny, a mathematics professor from York University, who mentioned that the mandatory census has its problems (negative bias associated with the "mandatory" part). That may be true--but it does not make the voluntary census any more accurate a priori. So that's why the compromise favoured by many of the committee's witnesses, including Don McLeish, makes sense: do both. Run the mandatory long-form census and a voluntary replacement, then compare the data sets to see if this alternative is viable. The statisticians are doubtful, but that's what statistics is all about: gather and then report.

Somewhat topically, statistics are what the Conservatives need but do not have. Where are the studies that show this is the best alternative, if an alternative is truly what they desire? Where are the statistics that show Canadians are outraged over the invasion of privacy signified by the mandatory long-form census? The Privacy Commissioner doesn't have them. Where are the statistics showing that Canadians live in constant fear of incarceration should they fail to complete the long-form census? It is one thing for the Conservatives to propose changing the census--which could be improved, most likely--but it is another thing entirely to make unilateral changes to the census, at practically the last minute prior to the 2011 census, without any research to support the changes.

The government's arguments for these changes are ultimately inconsistent. After Sheikh's resignation belied any implication that Statistics Canada recommended the change, the government has fallen back to its original argument: the mandatory long-form census is invasive to Canadians' privacy. Additionally, the threat of incarceration for failing to fill out the long-form census is unreasonable. Since there are viable alternatives to a mandatory census, we should make the long-form census voluntary to address these concerns.

As mentioned above, the claim that the census invades Canadians' private lives is dubious. Even if it does, Canadians do not seem to be very vocal about it. Furthermore, the census provides valuable data well worth the inconvenience of telling the government how many bedrooms one has in one's house--this is the same government, remember, that asks how much money you make and then takes a cut. But I don't see the Conservatives abolishing income tax. Some of the alternatives to a long-form census, such as the citizen registries used in Scandinavian countries, are far more intrusive. Fellegi stressed to the committee that the census data is collected anonymously, that the danger of it somehow being identified with the individual who submitted it is insignificantly low. Hence, there is a difference between being intrusive and being dangerous when it comes to privacy concerns.

So what about the assertion that incarceration is an overreaction to non-compliance? I actually agree. I think most people would agree. As an aside, it's worth mentioning that no one has been jailed for refusing to fill out a long-form census. Fellegi mentioned this during his testimony. Yet the Conservatives insist on creating a false dichotomy around the issue of enforcement: either, they insist, we keep the mandatory long-form census and throw people in jail, or we make the long-form census voluntary. There is, apparently, no middle ground, no alternative.

Since when have jail time or a fine (the other option) been the only punishments for breaking the law? Surely there are methods of enforcing a mandatory long-form census that are more reasonable. It's not my job to suggest enforcement methods--and it is certainly not the job of Statistics Canada or statisticians, despite attempts by David Anderson to get Fellegi and Sheikh to do just that. No, once again, the onus is on the Conservative government to develop a solution. Too bad they neither want to admit it or address the problem. Rather than a good round of antibiotic, which a more robust form of enforcement would be, the government has just decided to amputate the infected limb.

Privacy and enforcement aside, there are good alternatives to a long-form census, right? Yes and no. Yes, alternatives exist. Other countries use some of them. Some of them have more privacy concerns, but ultimately the major issue comes to one of statistical accuracy and integrity. What works for one country may not work for another, and without proper studies on implementing a change in Canada, we cannot just appropriate a foreign model. That would be irresponsible.

Irresponsible is exactly what the Conservative government is being, and it makes me angry. As a mathematician and as a Canadian citizen, I am saddened and disturbed by the government's disregard for the opinions of its civil servants and its experts. This is a rejection of science in favour of a political agenda--and agenda that is far from consistent, as I have demonstrated above. The census provides essential data on life in Canada and its population. It is not only a source of that data, but it is the benchmark against which other surveys--including Statistics Canada's voluntary surveys--are measured. So many groups rely on census data: the government, notably, as well as business groups, religious groups, special interest groups . . . the list goes on almost ad infinitum. And here we have the Conservative government blatantly ignoring opposition, forging ahead with an ill-conceived notion that they are crusading against violations of privacy.

Yet it moves.