Considering the difficulties, the Census Bureau does a remarkably good
job at counting people. This article discusses techniques for adjusti
ng the census. If there is a large undercount, these techniques may be
accurate enough for adjustment. With a small undercount, adjustment c
ould easily degrade the accuracy of the data. The Bureau argued that e
rrors in the census were more serious than errors in the proposed adju
stment, using ''loss function analysis'' to balance the risks. This pr
ocedure turns out to depend on quite unreasonable assumptions With oth
er and more realistic assumptions, the balance favors the census. The
story has a broader moral. Statistical models are often defended on gr
ounds of robustness. However, internally generated measures of precisi
on may be critical. If the model is at all complicated. these measures
of precision may turn out to be driven by assumptions not data-the an
tithesis of robustness.