Tuesday, August 3, 2010

Long-form census II

In a strange twist the Statistical Society of Canada has launched a petition to bring back the mandatory long-form census. Strange, because the long-form is not really a census, which by definition involves universal enumeration within a geographic area. The long-form is a survey, which, regardless of care, will introduce some bias into the results.

Surveys are much more than a statistical exercise. And, an important starting point is to note that all responses are voluntary regardless of whether the completion of the questionnaire is mandatory or not - if the survey is not user friendly it will, by its nature, introduce biases.

A fundamental question is whether the data collected is actually useful relative to an information need. Census information should focus solely on information to enable efficient and effective government. More than this is an abuse of the public. Just because a religious group would like certain information, to be able to better target its missions, does not create a need. The group can find other ways to get the information. It should have no say in the matter.

Canada's long-form has been around for some time. Some of the questions now sound dated and are open to interpretation. The survey perpetuates ethnic and social profiling and connects it to housing and incomes. Does this really measure social progress? The survey fails to deal with many important emerging issues including the financial, social and activity issues of pensioners, volunteer activities, early childhood development, adult educational, and so on. These issues could give a better indication of life in Canada than questions on commuting times or number of rooms in a house.

Is a mandatory survey the best way to get information or would special studies provide more? Many commentators support a mandatory survey because they feel we will lose coverage of low income earners, new immigrants, aboriginal peoples, people not fluent in English or French, and the elderly. However, one has to question coverage quality when all of these groups are faced with completing a very technical 40 page survey. I suspect that a large proportion of the surveys returned on behalf of these people were not completed by those targeted.

The survey is simply too complex. It assumes that all members of a household share freely information. It assumes that people keep detailed records of expenditures. It requires too much effort to complete. Unlike members of the Statistical Society, we perhaps don't get quite the enjoyment out of all this. I think that it is fair to ask why the information is so important.

Other countries have given up on long-form surveys. They are collecting data through special studies that focus on specific issues. Canada should do the same.

No comments:

Post a Comment