Search The Web

Today's Headlines

Saturday, May 1, 2010

There Has Got To Be A Better Way To Do This!

So, I have been immersed for about 8 hours a day for the past few days, learning some of the details of how the census is completed. Obviously, some of this learning has come from actually working on various aspects of this process step by step.

I don't have personal experience with any of the census bureau's activities that come before the individual enumeration. But some of this was explained to us as we worked during our shifts at the local census office. These early steps start with preparing a list of the addresses of every known living quarters in the US. The addresses are then classified as either housing quarters or group quarters.

Group quarters do not get census questionnaires mailed to them. Instead census enumerators visit them personally to account for all persons living in group quarters (like dormitories, monasteries, hostels, long-term stay hotels, old-age homes, etc., etc.). Then comes the enumeration of long-term patients at hospitals, prison inmates, residents of mental institutions, etc. I was also told that enumerators actually went out one night to common hangouts for homeless people and counted them up too. That should have been an exciting field-trip, I am sure!

Then comes the mailing out of the census forms to all the known individual households in the US. The responses that get mailed back are collected, scanned and entered into computers by early April. If everybody in the US to whom a census form was mailed had returned their forms, I would have no employment at the census bureau right now. After the beginning of April is when the Non-Response Follow-Up (NRFU) phase of the enumeration starts. This is the phase of the operation I have the most familiarity with, at least as far as the office procedures that go into are concerned.
US Census
I don't know how old these procedures are. I was joking that some army general in the 1880's probably came up with this procedure, and it has been faithfully adhered to because nobody thought to write up (or even think up) new procedures. Maybe the only changes have been the introduction of a little bit of automation into the procedure.

What I can tell you is that the procedure is atrociously wasteful and inefficient. It is extremely labor-intensive when it does not have to be. I have already written about how pretty much all of my first day at work in the census bureau consisted of stuffing envelopes. The level of intelligence required for the work starts at that point, and gets worse every step of the way! Let me walk you through what we did, and the final product of our labors.

Basically, the NRFU phase of the census divides the entire geography of the US into Districts. These are called Crew Leader Districts (CLD's) because each district has a crew leader to whom a bunch of enumerators report. Within each district is a bunch of blocks. These are census blocks, not city blocks. These census blocks are then divvied up among different Assignment Areas (AA's). An enumerator is in charge of one or more AA's. Census blocks can belong to a single AA or be covered by several AA's. It depends on the number of households and addresses in that census block. Some AA's contain several census blocks. Some census blocks belong to several AA's. It depends on the residential density of different census blocks.

At the beginning of the NRFU operation, the census bureau identifies every census block in which at least one household did not return their census form. This helps them figure out which AA's and CLD's have to be involved in the NRFU operation. The first step in the NRFU operation is then to print out maps that cover each of these census blocks. These maps are then stuffed into map envelopes on which we put the CLD and AA numbers. This is what I did on my first day of work at the local census office.

After that, a list of all addresses in the census blocks corresponding to each AA is printed out. This address list also includes households that have returned their census questionnaires. Those addresses are marked off so that the enumerator knows he/she does not have to visit those addresses. But the addresses are all there so that as the enumerator goes around the neighborhood, they can identify addresses that are not on the address list, and add them to the address list. This is important since new houses are constantly being built in the US (not so much in the past few months it seems like, but that is a different story for a different time!). The enumerator also takes notes about addresses that don't exist anymore (demolished homes, etc.). To enable the enumerator to note down new addresses in their assigned areas, a couple of address add sheets are also printed out for each AA.

Then a list of census blocks included in the AA is printed out. I have no idea how these so-called block lists are used by the enumerator, but they are printed out for them by AA. We then print out (or included pre-printed copies) of a couple of pages of legalese, a confidentiality notice about the contents of the binder, and then a cover page for the binder. As you may have guessed by now, all the stuff that is printed out thus far (including the map envelopes), are all 3-hole-punched, and put into a binder. There is one binder for every AA.
Census Questionnaire
The most important thing that goes into the binder, though, is an envelope with census questionnaires. There is one questionnaire for each household in the AA that did not return their census questionnaire by mail. Each questionnaire is personalized by printing out and sticking a label on it with the household address. The label also includes a bar code that enables the census bureau to keep track of the questionnaire on the computer by scanning it.

I was involved in printing maps, stuffing them in envelopes, printing out address add sheets and 3-hole punching them, printing out address lists and 3-hole-punching them, printing out and 3-hole-punching block lists, copying and/or printing and 3-hole-punching the confidentiality warnings and other legalese, and copying and/or printing and 3-hole-punching the cover sheets for the binders. I was also involved in assembling these binders for various AA's and CLD's. There is a very specific order in which all these materials go into the binders, so that enumerators always know where to find what in their binders.

To give you an idea about the numbers involved, our local census office is in charge of about 50 CLD's, and each CLD contains about 50 AA's. Therefore the total number of binders we put together was about 2,500. I was also involved in printing out questionnaire labels and sticking them on the enumerator questionnaires. Now, each AA contains between 20 and 50 households that have not returned their questionnaires (one AA actually had 220 households that needed follow-up. I have no idea how 220 census questionnaires fit into that AA's binder). I was also involved in printing out address labels and sticking them on all the questionnaires that go into several dozens of these binders.

I have made it sound as if each person is put in charge of a CLD or AA, and asked to prepare all the material for which they are in charge. It did not actually work that way though. Each step in the process is mostly independent of the other all the way until it comes to putting together the binders, so in each shift I worked on different aspects of the operation. One day, all I did was print and stuff maps in envelopes. Another day, all I did was print and stick labels on questionnaires. In another shift, I worked on miscellaneous things like printing out address add sheets, address lists, block lists, etc.

All these preparatory steps then led to big piles of all the material that goes into binders being stored in the office in boxes and cabinets all over the place. Then, everyone in a few shifts, just picked up the piles of material corresponding to different CLD's and put together all the binders corresponding to that CLD. These assembled binders then went into more boxes. Towards the end of the week, there were so many boxes of stuff in the office that it was getting difficult to move around without bumping into some. And boxes were stacked and layered in such a way that some of them were impossible to reach without significant difficulty and/or danger!

Every step of the way also involves quality control. After the maps are stuffed into the map envelopes, a different group of people opens up each map envelope and checks to ensure that the maps stuffed into that envelope belong in that envelope and also that no maps belonging to that envelope are left out. Similarly, after the binders are assembled, a different group of people go through the binders to make sure that all the material belongs to that binder and also that the binder contents are in the correct order.

You would think that once the binders are put together, and assembled into boxes by CLD, we would be done. Not so fast though. The next step in the process is even more mind-numbing than all the stuff that was done so far. You see, people continue to mail in census questionnaires all the time. Any questionnaire that the bureau receives after it has started putting together the enumerator binders is considered a late-mailed return (LMR). Once the binders are put together, a list of LMR addresses is then printed out for each CLD and AA.

The next step in the process is to take these LMR's into account. The way it works is as follows: We open up the binder of each AA in which there is at least one LMR address. We then look up the LMR address(es) in the printed address list (which in part of the binder) for that AA, and strike the address(es) out with a pen. We then go through the enumerator questionnaires for that AA, and take out the questionnaire(s) corresponding to the LMR address(es). As you can imagine, this is extremely tedious and time-consuming work. It is also prone to errors as many addresses can be close to each other, and identifying the correct address to strike out or the right questionnaire to pull out is more a matter of sharp eyesight and mental alertness rather than intelligence.
Stacks of Boxes
At the end of a long shift, when you are physically worn out from lifting and moving heavy boxes of binders, and mentally ready to fall asleep because of the repetitive work you have done so far, it is easy to slip up and make mistakes in the LMR step of binder preparation. The quality control for this step therefore consists of scanning the barcodes of the pulled-out questionnaires into the computer, and letting it check to make sure that those addresses are indeed part of the LMR list.

At this point, the binders are considered fully ready. They are now ready to be "checked out" to the field. This once again involves pulling out each binder from the box, and scanning into the computer the barcode on the label we stuck on the binder's cover sheet. This tells the computer (and therefore, the powers that be at various levels of the organization) that our office prepared so many binders by such and such date. All this information is used to track progress on the census enumeration and to set targets and deadlines for the various local census offices.

The problem is that any operation that involves anything to do with the computers is hit-or-miss as far as the census bureau is concerned. The computers at the local offices are quite new and work flawlessly. But the back-end system is some massive system sitting somewhere (probably in Washington, D.C., but I have no idea and neither did my supervisors) remotely. And this back-end system is one of the most unreliable I have ever seen in my life. I did not work a single full shift in which the back-end system did not crash at least once. More often than not, the back-end system would go down and stay down for hours on end. This system is called the PABOCS, and stands for PAper-Based Operational Control System.

I can write a whole separate post about my trials and tribulations trying to use the computers to do simple things like printing out labels or address lists. Suddenly in the middle of an operation the system would kick me out and force me to log back in. Sometimes, the information you want will fail to be retrieved from the databases it is stored in (cryptic error messages on the computer screen would be followed by R-rated tirades from the mouths of office clerks!). Other times, the system would not let you log in at all.

We learned to get stuff that needed the computers taken care of and done whenever they were in a good mood and not wait (even interrupting breaks), because if we waited 10 minutes, we did not know whether we could get whatever we wanted done or not.

And this computer system is not even taxed by running it 24/7. It is brought down at 11PM every night and does not come up until 8AM the next day. It is down during weekends (from 11PM Friday until 8AM Monday). I am assuming all this downtime is scheduled for maintenance and updates. And in spite of these scheduled downtimes, the system refuses to stay up when it is supposed to. If I wrote a mission-critical software application that crashed on a daily basis or regularly went down for hours on end, I would be standing in line at the local unemployment office pretty quickly. But, I guess about 50% up-time with a few million bugs is good enough for government work! As for the accuracy of the actual census figures stored on these computer systems, it is anybody's guess . . .

Yesterday, we completed the final step in the process I have described so far. We scanned and checked out every last binder in our office (the computer cooperated with us and allowed us to get it all done, probably because we were keeping out fingers so crossed I have to use pliers to uncross them now!). So far, so good. We got our jobs done, and we have some breathing room before the enumerators start collecting the information and bringing it back in. But why do I say there has to be a better way to do this? Well, where exactly do I start?
Forest aerial view
First of all, it is 2010. It is not 1950 or even 1990. What is with all this paper? Our local office alone must have devoured a few small countries' worth of virgin forest to produce all the paper we consumed in producing the binders we were responsible for. I am not even counting the census questionnaires that got mailed out to everybody earlier in the year. The mail-in response rate in the area allotted to the local census office is higher than 75%. Thus the mountains of questionnaires I see now for the NRFU operation are less than a quarter of the questionnaires that were originally mailed out!

And then there is all the other stuff that goes into these binders. The maps, the additional address sheets, the address lists, confidentiality statements, and all kinds of other junk. And each binder contains at least one paper envelope for maps and another for questionnaires. And the binders themselves are not environmentally benign either. Nothing is ever reused, it looks like. The binders are brand new, and I have no idea what happens to them after the enumerators are done using them. We certainly did not see or use any binders from the 2000 census during my work at the local census office.

And don't even get me started on the processes. Even assuming all this had to be paper-based, there is automation that can do all this in a fraction of the time it took an army of us to do it manually. Sure, none of us would have gotten paid then, but I don't think the primary purpose of the census bureau is to be an employer of last resort. I thought it was to enumerate the population of the US efficiently! Can you imagine your credit card company relying on an army of low-paid workers to print out your statements, stuff them in envelopes and mail them to you? You would probably get your neighbor's statement instead of yours more often than not. And it would probably be 20 days late too!

And can't they buy paper that is already 3-hole-punched instead of us having to punch through 20 sheets at a time with hand-held punches? Because of the use of different 3-hole-punchers in the office, the holes on different sheets of paper never line up. Assembling the binders took longer than it should simply because people had to add stuff to the binder sometimes one sheet of paper at a time instead of collecting everything together and slipping them onto the rings in the binders. In some cases, people sloppily punched holes that were not fully inside the sheets of paper, so such sheets could not even be put into the binders without a new copy being printed out and punched correctly again. In one case, a 3-hole punch had been so badly adjusted that the distance between the holes on the paper did not correspond to the distance between the rings at all. I had to punch new holes in these sheets to add them to the binders.

And the whole LMR step is just typical of government. There is just no other way to describe it. Couldn't they just give a list of LMR addresses to the enumerators and ask them to check addresses against that list before visiting them? The theory I heard about this is that we office clerks are cheaper on a per-hour basis than enumerators. So, it is better for us to go through the LMR process than to ask the enumerators to do it on the fly! And what is with having to print labels on sheets of labels, then peel them and stick them on various materials (census questionnaires, binder cover sheets, map envelopes, etc., etc.) instead of printing directly on these materials? I am sure this was how it was done in 1950, and it worked, so heck no, they are not going to change now!

Ideally, the whole process would use computers and the internet as much as possible. First of all, obviously, I would hire competent programmers who actually know a programming language to design back-end systems that can withstand actually being used! Then, I would have a web front-end to all of this. All that households would get in the mail is a card that gives them a web-site URL and a PIN. Households would go to that URL, log in using that PIN and answer the census questionnaire online. No paper questionnaires, no army of data entry clerks interpreting scribbles and squiggles to enter the information in the questionnaires into the computer. The card would also have a phone number that one can call to have a paper questionnaire mailed to the household if that is their preference. Only such forms would need to be scanned and entered into the computer manually. Several forests would breathe a sigh of relief when that is done.
handheld computer
Next comes the NRFU process. Once again, enumerators would use hand-held computers (a low-powered netbook or something similar is all it takes) instead of lugging around tons of paper questionnaires assembled into binders by an army of minimum-wage workers. GPS units being so cheap, I would save paper and the labor associated with the preparing maps for the enumerators by equipping them with customized hand-held GPS units. Instead of boxes of binders filled with heavy paperwork, the enumerator would then travel around with one net-book and one GPS unit. All the information would be entered into the net-book and either uploaded to the census database using WiMax technology (thus the net-books would remain "online" all the time and not store any information locally unless the WiMax network fails, so that the chances of data-theft due to the loss or theft of the net-books would be minimized or eliminated), or stored locally and then uploaded to the database every evening in the office when the net-book is plugged into a LAN.

And believe me, if UPS and FedEx drivers can be taught to use hand-held computers, so can census enumerators. Or change the recruitment tests to emphasize experience with computer use rather than asking inane questions about how to arrange a bunch of letters in alphabetical order or how to arrange dates in chronological order (the fact that such questions have to be asked at all, and that some people don't get them right points to a much bigger problem). If you want to see a sample census test, take a look here! Oh, and of course, the test is paper-based, not taken on computers or anything like that. Even my children take their MAP (Measure of Academic Progress) tests on computers, but the census bureau has no money or expertise to set up kiosks to administer these tests efficiently on computers!
census test
Maybe I will live to see the day when all this is reality. Maybe not. I am not holding out much hope, to be truthful. This is an organization that has 10 years to prepare for every census. And it still can't even get a back-end computer system on which the whole process relies, working correctly more than about half the time. Maybe it will always be a source of temporary, low-paying employment every 10 years. Maybe during tough economic times, the government should do a census every 5 years or 2 years or whatever it takes to get the employment rate up! If the primary purpose of the census bureau is to actually provide a boost to employment rates, perhaps, there is no better way to do this! Well, actually there is a better way in that case: turn off all the computers, and have office clerks write address lists and other materials that go into the binder by hand!!

1 comment:

Term Papers said...

I have been visiting various blogs for my term papers writing research. I have found your blog to be quite useful. Keep updating your blog with valuable information... Regards

Visitors Country Map

Free counters!

Content From

In the News

Article of the Day

This Day in History

Today's Birthday

Quote of the Day

Word of the Day

Match Up
Match each word in the left column with its synonym on the right. When finished, click Answer to see the results. Good luck!



Spelling Bee
difficulty level:
score: -
please wait...
spell the word:

Search The Web