US census takers fight angry Americans for their data
They barely even vote. Why would they fill in a form?
The results from the US 2010 Decennial Census are guaranteed by the end of the year. However, those who collected it - called enumerators - already know quite a bit about the state of the nation. There's no good news.
The census, conducted once every 10 years, hired a few hundred thousand Americans to collect names, ages and race data, using pencils and forms. It was an ad hoc operation, hiring people swiftly off the street, usually after a 30-minute test of skills and a telephone interview, not at all like the processes now instituted by human resources throughout corporate America. If you were vigorously proactive, had good reading comprehension, could read a map, sort and alphabetise, interpret regulations and schedule yourself independently with very little supervision, you were pretty much in. I know. I was one of them.
In the cratered US economy there was no shortage of motivated and qualified people for the jobs. The economic distress from the Great Recession hit every age and qualification demographic in the US, excluding only the very wealthy and the national security complex. So the crews in Pasadena had the old and young, from the high school educated to those with advanced degrees - everybody from the ranks of the unemployed, underemployed or those trying to squeeze out extra dollars for overstretched budgets.
While the census overlords always talked about data capture, conjuring up the image of a vast network of machines scattered across the US that were humming away, hoovering up all the data - the reality at street level was analog. Pencils, sharpeners, lots of erasers. Optical character recognition limits required everyone be regularly admonished over penmanship. Questionnaires were regularly sent back to be reworked. "No more water stains!" was the command to a downtown Pasadena crew one week.
Actually, although everyone from regional leaders to enumerators had their drink glasses and water bottles close, sweat stains were probably the culprit. At the beginning of the southern California summer, that was what dripped on the papers as you stood at the door, taking down the answers of the American people. Crews were dispatched to cover blocks of addresses printed in master binders; in my case, one book per block due to population density.
The book listed those who had filled in their census forms by the deadline along with those who hadn't, the latter of which were called NRFU's (pronounced nar-foos) for non-responders. We were hard after them.
Early on, it became clear that faults in data harmonisation and insufficiency in speed of processing guaranteed that some people who had sent in their forms properly weren't in address books with which we were serviced. When you showed up at their door, they'd invariably say they'd returned their form. Since this was one of the standard responses of census dodgers, one had to shrug shoulders and patiently ask them to redo it. Most people were good with that. Some weren't. They'd get cranky, rightly so. When their initial returns would eventually be registered, the books in the field would be rechecked for these addresses, and if they hadn't yet been hit, crossed out. It was usually too late to be effective.
The process was also subject to computer-generated error. Theoretically, the enumerators had the wherewithal to catch glitches and inform their directors. But in my experience, no one was interested. The pace of data collection was so accelerated that the manager with whom I had contact simply didn't want to hear about snafus of any kind.
A case in point, which was not rare, was the dispatching of enumerators from two separate crews to canvas the same address. Normally, the two enumerators would not cross paths on such a computer-generated duplicate. They'd simply hear at the door that someone had already been there, a message delivered with the usual exasperation. Since there was very little effective intercommunication between crews in the city, there was often no way to correct the problem, which was to list one data set as a duplicate for deletion. Very occasionally, you would bump into another enumerator working the same neighborhood on one of these error-caused runs, which would be cause for comparing address books. At which point it always became clear the dispatching had published the same non-responding address in two different volumes.
Maybe the people running the census from the top were interested in such errors being winkled out. At the bottom, no. The best you could hope for was to have your crew leader, who was a step lower than the supervisor, hold a questionnaire return as a potential duplicate so that it could be reconciled later. And they would do so. However, whether this resolved the issue was impossible to determine. From experience, I'd speculate it didn't. In the end, they were probably thrown into the giant data suction.
Sponsored: Hyper-scale data management