15 Jul Harper’s Reformers vandalize the census

In an interview with CBC Radio’s Jim Brown, Ivan Fellegi, who served as Canada’s Chief Statistician from 1985 to 2008, set forth five ill-effects of the Harper Government’s surprise decision to make a crucial part of the 2011 census voluntary.

• The results will be biased because aboriginals, new immigrants, the poor, those with low educational attainment, and the very well off are less likely to respond. This will deprive Canada of important information about social trends such as income polarization. It will eliminate our best source of information about aboriginal Canadians, immigrants, and minority language groups.
• Municipalities and provinces will lose their main source of planning information for transportation systems, housing, and other local issues.
• The introduction of such a major change will disrupt any comparison from previous censuses, since it will be impossible to know whether a trend is real or simply an artifact of the switch to voluntary reporting.
• The change will deprive private polsters and researchers of the baseline data they use to check the validity of their samples, and adjust them to ensure they accurately reflect the Canadian population.
• The change means we will spend upwards of $100 million to gather unusable data.

Derek Cook, research and social planner for the city of Calgary, adds a sixth: In previous census years, the long form was sent to carefully selected households in each community, producing invaluable planning information at the neighbourhood level. Making the long form voluntary will render this neighborhood data unusable. Says Cook:

If we don’t have that data at the neighbourhood level, we’re crippled…. How do we assess the need for affordable housing if we have no reliable low-income data to base our affordable housing plans on?

The Harper government responds to these problems with specious arguments. Industry Minister Tony Clement says some people found questions about the number of bedrooms in a household or the length or travel to work intrusive, but provides no data. Canada’s Privacy Commissioner says the 2006 census produced a total of two complaints to that office, and an investigation determined that the long form did not offend the country’s stringent privacy laws.

The government says an increase in the number of households receiving the long forms will compensate for its voluntary nature, thus betraying its ignorance of the science underlying statistical sampling. Increasing the size of an unrepresentative sample does not correct its failure to reflect the population being sampled. Randomness is paramount. Voluntary samples, no matter how big, are not random.

The surprise change is a move right out of the US Republican Party playbook: If you don’t count poor people, poor people won’t count. The same holds for minorities.

Mark this as one of those revealing episodes in Stephen Harper’s minority tenure when the carefully tended curtain slips and we catch a fleeting glimpse of the Oz-dwelling figure at the controls. Like the mean-spirited mini-budget tabled and then abandoned in the fall of 2008 under threat of an opposition coalition, it’s a hint of how the Reformers will change Canada if we’re ever incautious enough to give them a majority.