I think that, essentially, we have to distinguish between an opinion poll and a survey with factual questions. The census questions are not opinion questions; they are really very factual questions. With very factual questions, the problem you raise is much less of an issue. Forcing people to express an opinion and take a stand can be problematic. When people are asked very concrete, factual questions, the risks you mention are much lower. The problem does not really come up in practice. I think we have to draw a clear distinction between these two types of surveys.
The census is not an opinion poll. It is made up of factual questions that are easy to answer. The information people have to provide is not a barrier to answering, and, in terms of what the census requires from them, the price they have to pay is very low. I think we need to count on that. We need to talk about the reliability of the data, even though the process used to gather those data is mandatory for people.