Busted Handhelds and the Census

Census Bureau Director Louis I. Kincannon tried to reassure a Senate subcommittee at a hearing yesterday that the agency's plan to switch to handheld devices for the 2010 census will go off without a hitch. In the process, he noted (according to this Washington Post report) that of 1,388 handhelds the agency has tested so far, "only five had problems out of the box."

That may be an impressive number for these types of devices. But it still gives me pause. With the Census planning to use more than 500,000 handhelds, that would mean more than 1,800 wouldn't function -- and that's assuming the failure rate of the test project holds as the effort scales up. And remember, these numbers refer to devices that fail "out of the box" -- in other words, don't work at all right from the start. The figures don't include devices develop problems later on, or are dropped or otherwise damaged in the data-gathering process.

By the way, since he's too modest to mention it on his own blog, I'll note that Allan Holmes' story on the handhelds from the July 15 issue of Government Executive played a key role in the hearings, serving as a focus of questioning of Kincannon by Sen. Tom Coburn, R-Okla.