In considering these options, a number of factors had to be borne in mind. Firstly, the difficulties of traveling to and within the interior of the state are considerable. Roads can be poor, especially in the rainy season, and there is no aviation infrastructure at all. Secondly, there is no telephone or telegraph communication with many government offices. The only consistent method of transmitting or receiving documents is the postal service - which itself relies on a variety of methods such as once-weekly private buses or trucks to carry mail. Thirdly, there is no computerization within most government offices. Therefore, data is not held in a readily-retrievable form.
After some consideration, it was decided to reject the first option, i.e.: the data collected by the Accountant-General. Though computerized, it was done to a very poor standard and would not anyway have provided sufficient post and personal data. The second option, requiring departments themselves to provide the necessary data, was also rejected. Due to the highly decentralized nature of their structures, as described above, they are generally not able to provide accurate and timely returns. (Indeed, the main cause of the delay in producing the Directorate of Statistics' 5-year study has been the reluctance of the Departments to provide the necessary returns).
The option of employing census enumerators initially looked promising. It was envisaged that they would visit every location in which personal files/service books were held and extract the necessary data directly from source. Using the Accountant-General's computerized data as a basic control check, they would then compile returns for each office, district, etc. to a uniform format and standard of accuracy.
However, in calculating the number of enumerators required, it became clear that there would be a need for approximately 200. Recruiting and training such numbers would have become an unavoidably major exercise in itself. Their deployment across the state, virtually simultaneously, would have required a major logistical, administrative and management structure; and the costs of such an exercise would have been considerable - much more than the state government could have afforded. The only alternative to the employment of such enumerators would have been the use of government employees in the role; but the state government made it clear that it was unwilling to sanction the release of 200 of its staff for such purposes.
Therefore, the only realistic option left was to require the DDOs to provide the necessary data. This option also was not without difficulties. Given the complete absence of any centrally-held information on the exact number and location of DDOs, it was obvious from the outset that it would not be possible to compile a mailing list. Therefore, there would be no direct means of communicating with each DDO and, in turn, no means of checking that each DDO had submitted a return. However, this option was by far the cheapest and therefore the state government selected it.
Late in July 2000, the Finance Department issued instructions to each Treasury to, in turn, instruct their DDOs to provide the necessary information. These instructions were contained in a detailed memorandum - one which provided specimen copies of the format to be used for the data returns. The memorandum indicated that failure to provide the staff returns by the end of the following month would result in the paybill for defaulting establishments not being honored. This was considered to be the most effective sanction possible and was intended to ensure maximum compliance.
Within a month, the returns began arriving in Finance Department. They were stored in no particular order and not checked so as to ascertain which DDOs had forwarded returns and which had not. It was acknowledged that some returns might be as late as 6 months - due to the extreme inaccessibility of some areas, and it was presumed that each Treasury would take the responsibility of chasing up any defaulters.
Of much greater concern was the physical state of the returns themselves. DDOs had used whatever paper they possessed or felt was appropriate. Some returns were set out on standard A3-sized paper. Other returns comprised A4-sized paper stapled or taped together to form larger sheets. There was no standard size used. Nor was there any standard method of completion. Some DDOs had typed their returns, others were handwritten. These latter returns were, in some cases, virtually illegible.
There was a further problem. The DDOs had not uniformly complied with their instructions. In many cases, the establishment data had not been provided. In those instances where it had been, there was no obvious correlation between establishment and staff data. So, for example, vacant posts were not indicated and it was not possible to determine whether departments were under- or over-establishment in particular Classes. Also, some DDOs had transposed columns, others had omitted certain ones altogether. Some had been deliberately facetious. (In the column 'Marital Status', one DDO had written 'so-so' against one name, 'getting worse' against another, and so on).
However, as it would have been very difficult to return questionable forms to DDOs, it was decided to proceed with what was available at the time. Accordingly, all the returns were sent to several local data-entry agencies for processing. A specially-compiled data entry program was used to cut down the number of potential errors in the entry process, (e.g. misspelling of department names or inconsistent use of abbreviations). Despite these and other precautions, the agencies themselves made some key errors of judgment. The most notable of these was separating the various sheets compiled by individual DDOs and giving them to different data entry operators. This meant that the operators did not know which department or DDO a particular sheet of staff names referred to.
After approximately four months, a total of 490,000 staff had been entered. This figure represented all the returns so far received. The total fell well short of the estimated 650,000 posts in the establishment and, allowing for the non-recording of approximately 50,000 vacancies, it meant that potentially 100,000 staff had not been accounted for. Moreover, there were large 'gaps' in the data - which meant that the state government was committed to a substantial verification exercise if the data was to be cleaned up to an acceptable standard.
Read the full case study‚Ä¶