much editing should be done manually and how much by computer, the matter of whether to divert resources to produce advance tabulations, and the value of documentation. 8.1 Manual versus machine editing

Prior to the introduction of data processing equipment, it was necessary to perform the editing functions entirely by hand. This was a very slow and complicated procedure. With the introduction of punch card equipment some editing could be performed by the machines; but here again because of the volume of cards it was a time-consuming process.

compiling the calendar, it is necessary to take account of the availability of qualified personnel, the length of the workday, occurrence of holidays, and similar considerations. Allowance must be made for absenteeism, delays in receiving materials from the field, and unexpected problems that may arise.

Developing a calendar is a two-way operation. An optimum schedule can be forecast to provide the most desirable calendar; then the schedule must be modified by restrictions that adversely affect it. For example, the number of available qualified applicants for key punch operators may be considerably smaller than was anticipated. Therefore, the operation will require a longer period of time than was originally estimated.

The actual operations to be considered as separate functions are:

(1) Systems design and programming

(2) Manual editing-coding and verification
(3) Key punching and verification
(4) Machine editing-coding
(5) Tabulation

These five steps are separate and distinct operations even though the flow of work from one to another will overlap. Throughout all of these steps will be a management review which can be called methods review. This is a continuing review of all processing steps from the beginning to the final day of processing. Another review which is a subject matter review rather than a procedural one also takes place beginning with the first computer run and continuing throughout the computer operation.

9. VOLUME OF WORK BY FUNCTION

In order to adequately estimate the personnel requirements, it will be necessary

to determine the workload by function and to calculate approximate production to be expected for each step of the census processing. For most operations, there is no standard rate of production which can serve as a guide. Persons responsible for estimating workload and personnel requirements can use the results of the testing procedures (discussed in section 5 above) to make time estimates, making allowances for unforeseen difficulties that arise.

9.1 Systems design and programming

The additional workload created by the census activity in the area of systems design and programming cannot be handled by the regular full-time staff of the NSO. The Chief of the Data Processing Division must decide at least a year in advance how many of his current staff can be retrained and promoted to positions of systems analysts, programming supervisor, programmers, computer processing specialists, computer operators, and control and tape storage specialists. Based on that decision, he can determine how many additional programmers and analysts will be needed. In some countries, the supply of qualified dataprocessing specialists will be critical or even non-existent. In Providencia, the NSO has been able to promote some of the existing staff, hire several beginning programmers, and hire several trainees who will be given on-the-job training at the NSO. The NSO planned to start training the programmers before the census is taken and to have programs prepared and tested before the census results are received in the NSO.

9.2 Manual editing and coding

In number of man-hours, the editing-coding process requires more time than any other portion of the census except the enumeration. The actual time schedule, therefore, will depend almost entirely on the number of clerks that are assigned to this operation. Ideally,

it would be best to use a minimum number of highly qualified clerks, train them well, and establish a routine system of quality control. This may not satisfy the time requirements, however.

The number of editor-coders that are needed can be estimated from an examination of the production records that are kept during pre-tests and other testing activities. From these records, the staff can compute the average number of questionnaires edited per day and the number of man-days it would take to complete the operation.

It is important that a realistic average be used. If the estimate is too high, too few persons will have been estimated for the job; on the other hand, if the estimate was too low, too many persons will have been employed and it may not be possible to keep them fully employed.

9.3 Key punching

The estimate of the volume of work for key punching and key verification is dependent largely on the number of questionnaires, number of items on each questionnaire, and the number of cards (records per questionnaire). An additional consideration is the number of columns to be punched on each card.

9.4 Computer processing

The census operations to be carried out on the computer, in Providencia, will consist primarily of (a) the card-to-tape process, (b) the machine edits, and (c) the tabulations. Estimates of the volume of work take into account such factors as the length and number of records, number of reels of tape, number of tables to be produced and printed, number of geographic areas for which each table is to be compiled, and number of sorts and merges. The time required to process the data depends on the quality of the input media

In Providencia's industrial census, it is evident that many pieces of paper are transmitted to and from various parts of the NSO and the field and from one division to another, within the NSO. The prompt transmittal of the materials and the keeping of complete and accurate records are necessary for successful and efficient implementation of the census program. Delay in assembling and sending materials to the field makes it impossible for the field staff to meet their time schedules; similarly, delay in routing materials through the various operations in the NSO may make it impossible to complete the work or to take remedial action based on review of the materials.

10.1 Importance of accurate control

Control is particularly important with respect to data processing. The DPD Control Clerk must verify that every document is received and finally tabulated.

Control records and procedures must be accurate. An error in transcribing the ED No. to the census form, for example, may result in including the wrong form in the data for a county or province. Similarly, illegible handwriting or misfiling may even make it necessary to repeat previous steps.

Control operations should be kept as simple as possible; the essentials must be included but the operation should not be cumbersome. For a one-time operation, like the industrial census, it is usually not necessary or practical to establish detailed procedures and numerous control forms.

10.2 Responsibility for control

The Control Clerk in the Control Operations Branch of DPD is responsible for the movement of materials to and from DPD and within DPD. In order to carry out his functions properly, the Control Clerk should be familiar with the overall plan for the census program and have specific knowledge of all the steps in the data processing operations. He is to verify that all the documents are received in DPD, sort the questionnaires, transcribe certain identifying information to each questionnaire, count the number of forms, assemble them in work units, and prepare progress reports. He also is responsible for reporting to the proper official any delays in sending or in receiving materials and for following through as directed.

10.3 Flow of work

Below is a simplified chart to show the flow of work through the data processing operations. It is assumed that verification of an operation would be performed in the same organizational unit but by different persons.

As the chart indicates, the Control Clerk has an important role in the flow of work. He is responsible for verifying that every document is received in DPD, assembling them in work units, recording the progress of each work unit, and maintaining records of errors made by the editor-coders and key punch operators. He is responsible also for noting where delays occur and reporting them to the proper authority.

(3)

unit and is used as a reference list to locate the ED for follow-up when problems occur during editing-coding.

Master Control List for Skip List Establishments, Form 502 (exhibit V-1-3)

Form 502 has the same functions as Form 501 except that it is used by the Control Operations Branch of MMD. Another difference is that it provides for the control of individual questionnaires for the skip-list establishments instead of all the questionnaires for an ED. Since there are so few skip-list establishments in relationship to their importance, it is felt that it is essential and practical to control by the individual form.

resulting key punch cards as they are processed. It provides the means to compute error rates for editing-coding and key-punching. In addition, it provides a means to insure that all the cards (data) for all the questionnaires are punched and entered into the computer.

(6) Card-to-Tape and Computer Edit Control, Form 513 (exhibit V-6-3)

Form 513 is used by the computer staff
to insure that all data are put on the
computer and are machine edited. By
means of this form, the staff can tell
when all the data for a county has been
accepted and merged with the rest of
the data for a province. It also tells
when a province is completed and ready
for tabulation.

Other processing controls are provided by the Master ED File and Listing of Tabulated Counts (exhibit V-6-1) and the Card-to-Tape Diary (exhibit V-6-2). These provide counts which are compared with the totals on Forms 501, 502, and 503 to insure all the data are entered into the computer. In addition, diaries of machine processing give information on either control totals or the quality of the data; for example, a diary might indicate the number of punched cards that were entered on a tape or the number of editing changes that were made.