Warning! Two-topic post!
There's a discussion on CNN right now, where Rick Sanchez is talking to a guy from the Census Bureau about why we have to count everyone instead of using statistical methods to take a sample, and extrapolate the population numbers from there. Evidently, Rick's List is an "audience-driven" show, where Sanchez presents stories based on viewer questions; this explains some of his eye-rolling as he tries to hold up "his" side of the interview ("TV ratings extrapolate the opinons of a thousand viewers from a poll of a hundred, and we know how well that works.").
To me -- and, I suspect, anyone who's really studied and used statistical methods -- the answer is obvious
The U.S. Census is one of the rare opportunities to get the baseline data upon which we can base our statistical analyses.
In the Geospatial Analysis/Remote Sensing field, we call this "groundtruthing". It doesn't matter how good you think your digital data is -- at some point, you have to get down on the ground, take a look at the place you're mapping, and make sure the Map Resembles The Territory.
It's funny -- I'll lay odds that the guy who posed the question on Rick's site is also one of those people who bitches that "statistics don't mean a damned thing -- they can make'em say anything they want." Too many people will lambaste statistics as a lazy shortcut that fabricates meaningless data -- until they find themselves in a situation where rigorous, complete data collection inconveniences them.
And, yes, statistics can be misused, massaged, and abused. More often than not, it's because the people reading them aren't doing so fairly or rigorously, and the people viewing them don't really know how to read them.
This segues into a subject that was running through my head earlier this morning:
The people who are most resistant to accepting the principals of Evolution by Natural Selection in a biological context are those who most eagerly accept the same principle in an economic context. They call it "Capitalism". Darwin cheerfully admitted that he got a lot of ideas from Adam Smith's Wealth of Nations.
There are those who will argue that the biological and economic systems are very different, and you can't assume that a model that works for one will hold true in another.
True enough, but from a century or two of observation, the model holds more true in the biological context.*
Even more, they tend to embrace it in a social context, as well, condemning programs that "coddle" the poor. If the poor were worthwhile -- in other words, if they were fit -- they wouldn't need support. If they were worthy, they wouldn't be poor, now, would they? So it's Right and Natural to leave them to their own devices.
Creationists tend to be Social Darwinists.
*Actually, the model works just fine in either context -- the mix of stable periods, instabilities, conditional oscillations, and mass extinctions look very similar whether you're looking at graphs of the fossil record or of economic trends. When you're on the ground in the middle of it all, however, the Panglossian hypothesis that the Invisible Hand of the market will produce the most desirable results depends heavily on how "desirable" you consider a regular pattern of decimation.