see the new book from Stanford University Press

In carrying out the census, three main problems emerged: 1) a lack of adequately trained personnel plagued the census project in my area, and many others, from the beginning, and was exacerbated towards the end by the mass hiring and training in 2-3 hour sessions of hundreds of enumerators who frequently couldn't tell you what census day meant, or the difference between occupied, vacant and nonexistent housing. 2) In attempts to speed up and control the census work, the district office implemented several changes that administratively speaking may have been brilliant, but had the effect of inadvertently creating a huge amount of duplication, as well as making key elements of the census task virtually impossible to achieve. Thirdly, a general lack of authority about the meaning of key terms and regulations, (that is, a lack of authority to define them to people working in the field) frequently left ennumerators in the dark about what key terms on the census questionnaires meant, and their own rights and responsibilities as census workers.

First, while the census bureau has complained several times about the difficulty of finding good recruits, reporting that the number of applicants was invariably far lower than what they expected, my experience was that it was incredibly difficult to even find out how to be hired -- so much so that I assumed that the census was all full up.2 In mid April, I discovered a number for the census bureau in the white pages (under US government), and after a number of referrals, found a number on which you could leave your name and address on an answering machine. Three weeks later, I got a phone call scheduling me for "testing". I took the testing that same week, but then had to wait another two weeks for somebody to call me to schedule me for "training". Thus, despite the fact that the census bureau was apparently aching for recruits, it took me 5 weeks to get hired. I was lucky, though. Many people didn't get work even after being trained. I met one person from my training class almost 6 weeks later in the last hectic days of the list/ennumerate phase-- It was his first day of work!

Meanwhile, the census was so desperate for people that they even resorted to random telephone calling -- a professor in my department got a call from a census recruiter asking him if he wanted to earn $9 an hour knocking on doors.

Whatever its cause, the lack of personnel had a definite effect on the census operation: in the area I worked in, my supervisor told me that they had started with only 6 people with over 4,000 units to enumerate in one month. The month stretched into 2 1/2, even though dozens of new people were added to the team. One problem was that many people would show up to collect their work, but would never be seen from again. Others would come in occasionally, with only a few cases completed, although they may have reported working full time.

As the project continued, the quality of the training declined dramatically. I was trained in 16 hours in a class of 25. Later, they trained new hires in two 4-hour sessions in lecture halls of 300. At least 2 hours of each session would be taken up with roll-call. Around this time they also engaged in mass promotions of older enumerators to assistant crew leader positions -- I suppose to train the new enumerators in the field since they had no regular training beyond learning how to fill out their time sheets. The problem was that they promoted all the best enumerators, and production plummeted because crew leaders were not supposed to enumerate.3 The training for the crew leaders was also rather rickety by this time--in fact, it was a circus. We had only four hours to learn all the secrets necessary to be crew leaders and answer all the questions we had been asking our crew leaders for weeks (without getting answers most of the time). Our "trainer", a field operations supervisor, started the class by ordering half the class out because there was no overtime (it was a Friday, and most of us had been working all week since Sunday). After further screaming and shouting, they came back, having been informed that an exception would be made in their case. Since we only had four hours, the trainer discarded the training manual, spent an hour having screaming arguments with trainees over timesheets, another hour was taken up worrying because there were only four hours, and most of the rest of the time, to my recollection, was taken up with the trainer's reminiscences about the 1980 census.

In another training class I took, the trainers insisted that double underlined questions on the census form were the last resort questions (that is, the minimal number of questions answered as a "last resort" if respondents are unavailable or completely unhelpful). They are actually questions asked about vacant housing units. However, I was later told by people who took classes from another trainer that he had drowned out protest -- insisting again and again on reasserting incorrect information. It was also during this training that we noticed that the manual was virtually useless in resolving the many different debates about census procedures and definitions that came up, and was even self-contradictory on some points.

The poor quality of training was especially apparent in the last week of the list/ennumerate phase, when, in order to speed things up, the district office centralized the assignment of work to the central office. At this point, crew leaders were no longer responsible for enumerators, and gave work out to any warm body that presented itself, hoping to get something back at the end of the day. Unfortunately, the bulk of enumerators were recent hires who had never been adequately trained, nor had sufficient experience under the guidance of a crew leader. The results were disastrous. One girl wanted to delete a unit because it was vacant -- she wanted to put it in the "delete -- vacant, open to elements" category designed for derelict houses. She didn't know that the census bureau counted vacant housing units! Many felt that it was sufficient to get the names off of mailboxes in order to substantiate a body count! (I later found several crew leader assistants who also thought this was appropriate!) Long forms were all but impossible to complete, it seemed, since almost nobody brought them in filled in completely. Countless errors made the bulk of work turned in useless, although some of it was "saved" by the expedient of asking enumerators if they "remembered" the answer to the missing questions. It is amazing what the human mind can remember under that kind of pressure! (previously, we would have told the same people to go out again, but under the centralized regime we knew that any incomplete work would be given out to anybody.)

This centralization was the climax of a number of changes in the way that census work was carried out. Initially, we were each given a computer printout listing all the houses in an area of 4-5 blocks. Of these, about 35%-50% would be "non-response households that hadn't returned the census questionnaire by the time the printout had been made (about 4/26 in most cases). We would, in our own time, visit each house and attempt to fill out a form for each household. The printout was essential for several reasons -- the most important being that it was the only way to identify housing units that were duplicated in the listing, or that were left out altogether. That is, two households may have lived at the same address during the last census, meaning that two listings existed for the same address. If one form were sent back in, without the printout, we would assume that the other form had to be filled in for that address -- and would effectively duplicate that household. On the other hand, the opposite could also occur: new living units, easily detected with the printout because we could see that they were not listed, are impossible to discover without the printout. Another major problem that could occur for which the printout was necessary to resolve, was what was called "misdeliveries". That is, when several units within a housing complex got their forms mixed up. For example, if apartment 2 in an apartment building inadvertently sent in the form for apt 4, the printout would show that apt 4 had responded and apt 2 hadn't. Since the printout had last names of respondents recorded on them, the "misdelivery" could be caught and corrected by the enumerator, if he was adequately trained, and if he had the printout. Without either, apt 4 would actually be skipped, while apt 2 would be counted twice. The printout also had the benefit of coming with the only manual we had for use in the field.

The reason I'm telling you the importance of the printout is that, in order to speed up the process, the district office required all enumerators to hand their printouts back to their crew leaders, who would then simply hand out addresses to be enumerated 10 or twenty at a time. This change was made because, as I mentioned before, some enumerators had received printout books but had done nothing with them. Thus, these books, as long as they were in the hands of the enumerators, would never get done. However, besides eliminating the safety features just mentioned that the printouts were designed to perform, this administrative move created a whole series of problems. At least in the crew I was working with, with something like 8 crew leaders and assistant crew leaders, the confusion of assigning work out of the same set of books meant that hundreds of addresses were handed out two or three times to different enumerators. This was on top of the duplication of work caused by the fact that many of the printouts repeated whole blocks from other printouts. (We later discovered that there were also massive "holes" of whole neighborhoods that were skipped between printouts.)4

This type of duplication could not possibly affect the body count, however, since they simply represented duplicate counts of the same cases (as opposed to the above mentioned duplication problems, in which the same people could be counted in two different cases), but it certainly did nothing for the public image of the census bureau, to say nothing of the morale of the poor enumerators, to have them bothering people for the second, third, fourth or umpteenth time.

This trend was only amplified when the district office decided to centralize operations in their offices: crew leader distinctions disappeared, and all the crew leaders and assistants, as well as many enumerators and even office staff who didn't know what a questionnaire looked like, were compelled to sit behind desks for 8 hours a day handing out 10-15 forms to each warm body that passed by, and picking up and "correcting" work at the end of the day. Since many of those involved didn't know what they were doing, a huge mess ensued.

Some of these problems were supposed to have been caught in the subsequent "field follow-up" phase in which enumerators checked all the reportedly vacant or deleted listings. For example, in my first day of work on the new phase, out of 10 units listed earlier by enumerators as "vacant", 5 housed people who had lived there well before census day -- one lady had been living there 22 years. I also found "deletes" where houses existed, but this is where problems emerged again: They didn't give us the printout, so we had no way of knowing if the listings were deleted because they duplicated other listings (i.e. the address had been originally listed twice.).5 I was also surprised at the number of apartments that had been previously listed as vacant in the enumerate/list phase that were still vacant when I visited them in the "field follow-up" phase. Few apartments in West Los Angeles go vacant for over two months, and I suspect that they had been listed as vacant simply because they were vacant when visited by untrained enumerators in the previous phase.

The census process, in my experience, was one of the most alienating work situations I have ever had -- and as the district office changed the rules to make it more "efficient", it got progressively more alienating. Initially, enumerators were free to work when they wanted to, reporting in with their work done every few days. In a sense, they themselves controlled the means of their production -- the printout sheets. However, in an attempt to impose greater control over the work process, the district office required crew leaders to retain greater control over the printouts, and eventually usurped absolute control themselves. Not only did this, as I pointed out above, make certain vital features of the printouts unavailable to the people actually doing the work, but it also meant that enumerators had to spend substantial amounts of time every day waiting to be assigned work in the morning, and to have it picked up and checked in the afternoon. Furthermore, the lack of agreement about the definition of key terms and procedures necessary to census personnel created feelings of fatalism and a certain lack of respect for their superiors on the part of those who had to deal with the incertainties of these definitions. Very often, the "correct" way of doing anything would not be known until experimentation had produced a series of forms that had failed edit. Even this was not a sure way of finding out correct procedures, since the editors were just as likely as anyone else to be wrong.

Census procedures and rules also tended to change from moment to moment, heightening the confusion about what is the right thing to do, and creating a certain amount of defensiveness on the part of those responsible for enforcing them. The rules on overtime were never very clear. For a while overtime was allowed for everybody in order to speed up operations. This policy was reversed after a week in which overtime soared, but productivity hardly changed at all. At times, people would work overtime, only to find that it wouldn't be allowed. Exceptions could be made, but very few people were aware of the circumstances in which this was possible. To give a simple but succinct example of changing rules, when I first went into the district office building, I was required to wear a visitor badge despite the fact that I already had an enumerator badge. On a second visit, I asked whether a badge was again necessary. Looking at her watch, the receptionist answered "Well, as of 1:05pm today, you don't need it. Wait five minutes and maybe you will." These are petty examples to merely illustrate the lack of coordination and information that pervaded census operations.

These alienating features created a generally negative attitude towards the census bureau, the federal government and the census process itself. We had a hard time believing that census figures were taken seriously by anybody, but gained acute insight into the source of the federal deficit. Unfortunately, this attitude led to a lacksidaisical attitude on the part of many. Furthermore, the threat of sanctions and the provision of rewards had little effect on quality or productivity. Both sanctions and awards were applied haphazardly with little correlation with efficiency or zeal, and often on a personal basis, since as an organization, the district office had little idea of who was even working for it, let alone the quality of work they did. Supervisors were usually supervisors because they had simply been around longer, or had ben in the right place at the right time. I remember in my final week of work for the census, I was offered the opportunity to again be promoted to assistant crew leader status. When I arrived at the training class, however, it became apparent that most of the students were new hires with no previous training or experience, and thus had to go through enumerator training first. My training was "postponed". The trainer was thoroughly annoyed. Later, he told me that the district office manager had assured him that these were "hand picked candidates" for promotion.

So far, I have focused upon procedural problems, but problems of definition also plagued us lowly workers as we struggled to complete our noble tasks. One of the biggest questions in training, although not the only by far, pertained to the definition of race and hispanicity. No census manual available to me explained this method of measuring race. As a sociologist, I thought that the point of separately asking for race and hispanicity was to separate black hispanics from white hispanics. But in practice, it didn't work. When the questions were asked, the race question was answered first, and some variation of hispanic was answered there by the respondent. The results may be different for write-in questionnaires, where the respondent usually reads all the questions before answering.6 This was not helped by the fact that most interviewers that I know of skipped the list of racial categories altogether.

There were other definitional problems, although they may seem minor -- such as problems over the definition of vacant and non-existant housing. But a number of demographic changes also created serious problems for census workers in their everyday work: for example, security buildings. Security buildings not only provided protection for xenophobic citizens, but also a wealth of questions about the right of census ennumerators (i.e. the census) versus the right of citizens to their paid for privacy.

1. I have already mentioned that the census is carried out by what normally would be considered an oxymoron: a temporary bureaucracy. Whatever they are, bureaucracies usually evolve and grow over long periods of time, allowing for battles over turf and uneasy truce's between branches to emerge that allow at least some work to go on between intra-agency squabbles. Their longevity gives a sense of security to individual members as well as time to at least figure out the ways and byways of the bureaucratic morass. A bureaucracy constructed overnight, on a plan carefully laid out by "experts" is guaranteed to fail to be able to adapt to the personal nuances of the individuals who occupy key offices, or even minor offices. And rules that were written months or years before the people who are to carry them out were even dreaming of doing so, are bound to fail to clearly express themselves in the same manner to every individual who needs to interpret them. Thus, for instance, a definition which may seem clear to one person after reading the manual, may be interpreted differently in training, in another manner by field supervisors, and in yet a fourth fashion by district office quality control staff. The situation was not helped by manuals which systematically defined terms by giving one example of them, and which in some cases even contradicted themselves from one page to another.
2. I wanted to work on "S night", but never found out how to be hired. While I didn't spend 100% of my energy looking for information on census hiring, I was certainly looking harder than your average citizen, and I heard nothing on the radio, TV or press about hiring, except once in early April, when a local TV station reported that the census was looking for people who spoke exotic languages. The implication of the report, indeed, was that all other positions had already been filled, so I stopped looking for a while.
3. One of the best enumerators we ever had fought promotion tooth and nail and, after a week of getting two or three cases completed a day by his entire team of three enumerators, finally went out and enumerated 45 cases in one day. If he had been an enumerator, he would have received a $67.50 bonus.
4. In another area, a bright crew leader had the idea of handing out photocopies of the printout several pages at a time. Unfortunately, they lost track of how many copies they made and had to efectively close operations for a week as they tried to sort out the books.
5. A second form of check was also built into this phase: if people were living now in "vacant" apartments (that is, apartments that were vacant on Census day), we asked them if they had filed forms elsewhere. If they hadn't, we filed them now, for them to be counted at their 4/1 address. This was designed to catch people who had failed to be counted at their old address, but it did nothing for people who moved into units whose previous residents had been counted.
6. I should also point out in passing that while I have heard it said that the census bureau's operating definition of race was whatever the respondent tells them, this should not be mistaken for auto-identification. Many of our respondents were neighbors and landlords.

Visit my home page

See my CV