Canada’s Counting Problem

Every ten years there’s a debate in the U.S. around the census.  Typically it’s over whether or not to use sampling techniques to develop a more accurate count of the population.  The practice has been to not use sampling, resulting in what most consider to be an undercounting of the population.  The undercount is considered to benefit Republicans as the people that get missed tend to vote Democratic and the people that get double-counted tend to vote Republican.  Since the census determines levels of funding for services as well as a state’s numbers in the House of Representatives, the debate matters, and will likely persist.

To the north, Canada is preparing for a census in 2011 (it does so every five years compared to ten for the U.S.), but is dealing with a very different methodological debate.  The Conservative government, expressing a concern over the privacy of its citizens, has opted to make the census long form (distributed to twenty percent of the population, the remainder receive a short form asking for very basic information) voluntary.  (Much like in the United States, you can be fined for not answering census questions.)  However, the national privacy commissioner has received all of three complaints about the census since 2001, and Statistics Canada takes stringent measures to maintain confidentiality of census data (much like the U.S. Census Bureau), not releasing detailed census information for decades, if ever.

So the publicized rationale for the change does not reflect a statistically significant analysis of public opinion, but the anecdotes of a few people who caught the attention of Members of Parliament.  Please don’t choke on the irony.

Statisticians are furious (yes, it’s possible).  The government has said it will increase the number of long forms sent out, but since it’s now voluntary, the response rate will decline and the reliability of the resulting data will not mesh well with what has come before.  Census data are critical to many civic and private planning efforts, and this move will make it much harder to prepare effective predictions of the future.  Census data is also critical to those conducting research on demography or the provision of social services.  Given the strong rules in place on disclosure of individual information, the Harper government made the wrong call, if they are really concerned about privacy.