Apparently I'm not the only one who thinks that the idea of sending out a voluntary long form in next year's census is a bad idea. Since this seems to be a file to which I will be returning from time to time, here are a couple of points to consider:
1) Yes, it is a big deal. If response rates vary with the income and education levels, then you won't have a random sample of income and education levels. There is a rather large amount of evidence in the sampling design literature documenting the fact that people with lower levels of education and income have lower response rates, and so these groups will be systematically under-sampled.
Sample selection bias is a recurring problem for virtually all surveys in which participation is not mandatory, and it is possible to correct for it - but only if you know the true distribution of the underlying distribution. For example, if you know that those in poverty are (say) 10% of the population and only 5% of your sample, you can re-weight your observations to make the sample representative. But in order to do this, you need to know that the 'true' proportion is in fact 10%. Usually, the source for this sort of information is the census. But if the census itself suffers from selection bias, there's no easy way to fix it.
2) Concerns about privacy are overstated. Anyone who has had dealings with Statistics Canada will tell you that they are ferociously - and at times irritatingly - determined to protect the privacy of those whose information is stored in their data bases. Researchers never see the data. They are obliged to send their estimation codes to StatsCan professionals, who run the programs and return the output to the researcher. That goes for all other non-StatsCan government employees as well.
And if the question requires looking at a subsample that is so small that there's a chance that individual respondents could be identified, then the request is refused.
3) The implications are more wide-ranging than you might think. According to popular cliché, we are evolving towards a 'knowledge economy' - and knowledge requires data. And the usefulness of much of these data depends crucially on the anchor of a reliable census. Here is an incomplete list of people who need this information:
- Investors. Suppose you're planning to start a business somewhere, and a crucial factor is a supply of well-educated workers.
- Marketers. Yes, it's all good fun to mock marketers, but if they can't identify what will sell, people lose their jobs.
- Policy-makers at all levels of government. How do we know if a school's poor performance can be explained by socio-economic factors if we don't have reliable data for income and education levels?
- Academic researchers. 'Nuff said.
The federal government made this move without consulting the people who use census data. It should reconsider and reverse its decision.