Okay, but just to clarify—because this is my technical question—that example that I used would involve bias, would it not? That's not a sampling difference; it's bias. It's bias that would only occur among those people who are forced to fill it out, that marginal group over and above the people who fill it out automatically. It would be a bias that would only occur in a mandatory long-form census and wouldn't necessarily occur in a voluntary national household survey.
On March 8th, 2011. See this statement in context.