How would they do that? They would count the number of people in every house in your neighborhood but if they couldn't get a hold of you, they would impute 4 people to your household because the average number of people per house in your neighborhood is 4. They will assume you breed about as often as your neighbors. Will this work? It will, because while you might have only one child, in the next neighborhood they do this, the person they miss may have one more child than their neighbors. I'm grossly simplifying, of course, but this is the basic idea. Things even out and they haven't thrown you off from their census.
This kind of averaging is done everywhere. It was used by telephone companies in designing switching systems -- the average phone call, they believed, would be three minutes. Before the Internet and staying connected because it is so tough to get connected to the local ISP, it used to be three minutes. The telephone network worked and prospered and gave users dialtones promptly. Such averaging is done by insurance companies to know just how low a quote they can give a person of your age, gender and location and still make a profit.
Politicians use it to conduct polls based on which they decide whether or not to lie about their sex life. So, what is the problem if the Census Bureau uses it?
You know the answer to that one, but the question still bears asking. Why shouldn't the Census Bureau use the most accurate system available?
