I concede that with any count or survey of a large population, you're never going to count everyone you're after. However, the only logical conclusion is that the undercount is fairly consistent over time, barring any other big changes, which also means that comparing today's TB rate to any earlier year's rate will be comparable in a meaningful way. So today's rate is still lower than 10, 15, 20 years ago.

Otherwise, if you really think that such a large discrepancy is possible last year, then you're left with the alternative conclusion that there could be a hugely variable miscount every year, which makes the numbers completely useless. Is the TB rate now really the lowest since the gov't started keeping count? I don't know, could be a miscount. Was there really an epidemic from 1986-1992? I don't know, could've been a miscount. Is there less TB now than 50 years ago? I think so, but I can't be sure--could've been a miscount.

By the way, the guy hanging out under the plywood in the desert would not be considered "unemployed" if he wasn't actively looking for work. Once you quit looking, you're magically dropped from the ranks of the unemployed. Sounds odd, but it does make perfect sense when you're talking about retirees or people who quit work to take care of their children. And the gov't doesn't rely soley on the unemployment benefit rolls to count the unemployed. They also call tens of thousands of households every single month, so if they call your house and your unemployed friend is crashing on your sofa, that's how the gov't would find out. Then they use that data to estimate what the unemployment rate for the entire country is.