It is the basis for calibrating almost all research conducted in a vast number of social sciences, especially economic analysis, sociology and political science, three areas with which I am particularly familiar.
Every time we conduct a survey, we must assess its value by comparing it to census data. For example, it is possible for housewives to be overrepresented in a survey. In fact, with a telephone survey, the people most likely to respond are those who are at home. They are often women. Consequently, in most of our surveys, there is overrepresentation of housewives and underrepresentation of youth, who cannot be reached at home during the day, in the evening, or even late at night.
The only way to take into account the fact that our survey has flaws is to recalibrate the survey based on census data. We use the latter to correct our sample. That is why the census data must be the most reliable. We use the data to correct all our other surveys. In most areas, particularly the social sciences and humanities, it is vital to have a completely reliable survey. In fact, all the others have flaws and we truly need the census data to correct them. That is the case for research.
If census data were to become less reliable one day, we would have serious problems with the accuracy of our research in the six-year period after the census. It would be a very big problem for the social sciences and humanities.