Thursday, October 08, 2009

Criminals at Your Front Door?

A blog posting by The Hill's Eric Zimmerman seems to confirm our worst fears about the 2010 Census.

Citing a new report by the Government Accountability Office, Mr. Zimmerman reports that the Census Bureau hired as many as 200 employees with criminal records that should have disqualified them for federal employment.

The problem was the result of poor finger-printing techniques by the bureau. According to the GAO, over 35,000 workers were hired for the recently-completed address canvassing process--despite the fact that their fingerprints could not be process.

Without clear, readable prints, the FBI could not complete a full background check on canvassing personnel. But the Census Bureau still hired the workers, and at least 200 criminals slipped through the cracks and participated in the canvassing effort.

However, it is unclear how much contact these individuals had with members of the public; during the canvassing operation, census workers compared actual addresses and residences with those in the existing database and made corrections. Census workers were required to knock on the door of each home as a part of canvassing, and provide information on the process if a resident answered.

Amid continuing concerns about the census--and those hired to conduct it--we forwarded a link to The Hill post (and the GAO report) to a former military colleague. Our friend (we'll call him Bob) worked for the Census Bureau earlier this year, after being laid off by a major corporation. He served as a Field Supervisor for the bureau, managing canvassing operations in nearly a dozen counties in the Mid-Atlantic Region. He has since moved on to an executive position at a non-profit organization. Here is Bob's reply:


I read the blog post (and the GAO summary) with a great deal of interest, given my recent tenure with the Census Bureau. Can't say that I disagree with any of their findings. You'll note that the problems with canvassing went well beyond criminals who couldn't be identified through the background check. The GAO says the bureau has no clue when it comes to cost estimation, one reason that canvassing went over-budget by at least $88 million.

The fingerprint operation, based on my experience, was absolute chaos. We had to complete two fingerprint forms on each employee, literally the day before computer training was scheduled to begin. The prints were necessary not only for background checks, but also to log onto the hand-held computers (HHCs) used by canvassing personnel. Without their prints in the database, individual workers--known as "listers" in the trade--couldn't use their computers, which meant that canvassing would be delayed. We literally had 24-36 hours to get everyone in the system, so the door-to-door effort could begin. Lots of potential for mistakes--including sloppy finger-printing.

Now, consider the size of the effort. I was one of seven supervisors in my region. Each of us had seven or eight teams of listers that we were responsible for--anywhere from 110-135 personnel in all. Everyone was fingerprinted on the same day, and all of those prints had to be rushed to the district office, where other personnel literally worked all night to get them into the database --before they were sent to the FBI. In my experience, the background checks (based on fingerprints) didn't begin until well after the workers were in the field. Why not fingerprint workers at the time they were hired? No one could answer that one.

The problem with the prints can be easily explained: most of the supervisors charged with taking them had no experience in the process. As you know, there is a definite technique involved in recording clear, readable prints. Luckily, I had lots of practice from my days in the military. And, I instructed my crew leaders to find personnel with finger-printing experience in law enforcement or the military. If they didn't have anyone with those skills, I made sure that we had someone qualified to help them out. By Census Bureau standards, our prints were very good (at least, that's what my boss at the district office told me).

Interestingly, the fingerprints did reveal one individual with a questionable past. About two weeks into canvassing, I got a phone call from my supervisor, who told me that one lister's prints had resulted a "match" in the FBI database. Our procedures in that situation were very clear; the worker was immediately removed from the canvassing operation and placed in stand-by (non-pay) status. We collected the individual's equipment, and the lister was instructed to contact the FBI for resolution.

To this day, I have no idea what sort of "flag" was raised by the FBI background check. The individual was not arrested (so it wasn't anything like an outstanding felony warrant), but they never returned to work. So, to some degree, the process worked. But, by the time we identified potential issues with this individual, he/she had been on the job for at least two weeks, and visited scores of homes.

A few final points. The canvassing process was a rush job from start to finish. Originally, the operation was scheduled to last eight weeks, but we completed the task in less than 40 days. There was tremendous pressure to accelerate the job, to save money and make our bosses look good. While field workers (including supervisors like myself) were temporary hires, top managers at the district office were guaranteed 1-2 years of employment, at $20-25 an hour. Not a fortune, but considerably more than the $13.25 listers were earning. The regional office had a sizable, permanent civil service staff.

With the push to "get it done," efforts at quality control were often laughable. While we had a QC team that evaluated our work, many of those personnel were poorly trained. On multiple occasions, my field workers discovered QC checkers were evaluating the wrong area. Most were far less proficient with hand-held computers and maps than our production listers. We also found that some of our work "failed" because the original database was so screwed up that no one could figure it out.

Incidentally (as the GAO observes), this is the first census that used computers for field canvassing. Previously, the census bureau used paper lists and maps. As you might expect, we discovered plenty of "pencil-whipping" from the past operations. One of my crew leaders worked in both the 1990 and 2000 Census, and discovered mistakes in her area dating back at least 20 years. And remember: canvassing is supposed to correct those problems. With the rush to finish the job this time, we questioned whether any of our updates will actually wind up in the revised database. In other words, census teams in 2020 will still be fixing mistakes from 2010--or even earlier.

Lastly, a word about ACORN. There's been a lot of justifiable concern about that organization's relationship with the Census Bureau. But in my particular district, I couldn't find anyone who was associated with ACORN, or had been hired through that its local office. However, my district covered suburban and rural areas, so ACORN probably had little interest in our area. I can't speak for hiring in neighboring districts, which included major urban areas.

I can report that the bureau's hiring practices often bordered on incomprehensible. As we were wrapping up canvassing--and laying off staffers--the local HR office was still interviewing and offering jobs to new personnel. They were supposedly gearing up for the next phase of the operation. Never mind that it was 4-5 months away, and those recently laid-off workers supposedly had first crack at being hired for the second phase of canvassing.

Bottom line: my experience as a census supervisor was unique, to say the least. Lots of good people in the field; my crew included retired teachers, executives and military personnel; single moms, college students and even a retired NSA analyst. As a group, they were hard-working and extremely conscientious. Unfortunately, leadership above our level was often incompetent, populated by a bunch of yes-men (and yes-women) concerned only about protecting their own jobs, and kissing the ass of their boss.

Against that backdrop, it's not surprising that a few criminals wound up going door-to-door. If it's any consolation, listers were not allowed to enter homes during canvassing, even if residents asked them to come inside. However, we had at least one "census imposter" who tried to get into a local home during the operation--we found out only because the home owner knew one of my crew leaders and contacted him.

That's something else the Census Bureau doesn't like to admit; the decennial process also attracts con men and crooks who pose as listers, and trek through neighborhoods in search of potential victims.




tfhr said...

Does total criminal headcount include Charlie Rangel?

Just wondering.

Rich said...

What's next?
Government run healthcare of course.

Storms24 said...

Is door-to-door canvassing or 10-minute written surveys really the most efficient or most accurate means of collecting this data?