I listened to a journalist interviewing a member of an association representing francophones outside Quebec, I believe. What was said was very interesting. The member was complaining, of course, about the demise of the long form. The journalist teased her slightly by suggesting that she tell her people to respond in large numbers to the voluntary questionnaire. Overestimating something is as problematic as underestimating it.
The problem is not the 30%. Statisticians know very well how to work with a questionnaire that applies to 20% of the population, which is a significant percentage. It is the representativeness of the sample that is fundamental. Yet, this is only possible through a mandatory questionnaire. Penalties are another issue. I worked on the history of Statistics Canada. I found some letters in the archives.