information on these supplementary forms was transcribed to the three regular census schedules. Crews of vessels were allocated to the ED's where their ships were berthed at the time of the census. Advance Census Reports The Advance Census Report form, which was used for the first time in 1960, provided for selfenumeration of the 100-percent population and housing items by members of the households. A few days before the census date, copies of the Advance Census Report were mailed to all households on postal delivery routes. The instructions on the form requested the householder to answer all of the questions for each member of the household and to have the form ready when the enumerator called. During his visit to the household, the enumerator reviewed the entries on the Advance Census Report with the respondent and then transcribed the information to the census schedule. If the Advance Census Report was not filled, or if there were omissions or inconsistencies in the data reported, the enumerator asked the necessary questions and recorded the information directly on the schedule. For vacant housing units, the enumerator was instructed to obtain the information from the owner, the landlord, or a neighbor. Field Organization To collect the desired information for the 2,349,544 persons and 520,990 housing units in Puerto Rico required a temporary field staff of approximately 3,300. This included the Area Supervisor, who was detailed from Bureau headquarters in Washington, and his immediate staff consisting of two inspectors, an administrative assistant, a supply clerk and a secretary; six district supervisors; six technical officers; 210 crew leaders; 204 field reviewers; 24 office clerks; and 2,865 enumerators. The Area Supervisor was responsible for recruitment and training of the field staff, obtaining space and equipment for the field offices, distributing supplies, supervising the collection of data, and preparing periodic reports on the progress of the enumeration. The six district supervisors were generally responsible for administering the census program within their assigned areas, while the technical officers were responsible for training crew leaders, handling technical problems, and supervising the field reviewers. The 2,865 enumerators were trained and supervised by the crew leaders, who reported directly to the district supervisors. Under the terms of the special agreement between the U.S. Bureau of the Census and the Puerto Rico Planning Board, arrangements were made with the Department of Education to use school teachers and administrative staff of the Department as enumerators, crew leaders, and district supervisors. All candidates were given one of two selection aid tests prepared in the Bureau to determine eligibility for appointment. The use of school teachers and administrative staff of the Department of Education had the special advantage of providing the census with a well-qualified and well-organized field staff, including supervisory personnel, for the enumeration. Teachers often worked as enumerators under supervisors and crew leaders who were their own superintendents or principals. This fact greatly facilitated the training and increased the effectiveness of the field supervision. The arrangement with the Department of Education also included space in the school buildings for training and general administrative headquarters. Field Review One new feature of the 1960 censuses was a systematic field review conducted by a specially trained staff member at regular intervals during the course of the enumeration. The purpose of the field review was to verify the completeness of the enumeration, to discover any errors which had been made, and to see that errors were corrected. In order to verify the completeness of the enumeration, the crew leader, before the enumeration, visited each ED under his supervision and made a partial prelisting of addresses. This list was given to the field reviewer, who compared it with the addresses entered in the enumerator's listing book. In this way, it was possible to determine on a sample basis whether the enumerator was covering all housing units. The field reviewer also checked the enumerator's work for accuracy, completeness, and consistency, and to determine whether the enumerator was following his instructions properly for the selection of sample units and for handling callbacks. Callbacks When the enumerator found no one at home at the time of his first visit, he was to enter the following information in his listing book: address, name of head of household (obtained from neighbors or from letter boxes), sample key letter, housing unit number, and best time to call back. He was also instructed to reserve a blank schedule in his enumeration book for each unit at which he found no one at home, and to leave under the door a "Notice of Census Taker's Call" specifying the date and time of his next call. To avoid accumulating callbacks, the enumerator was advised to make his callback as soon as possible after the original visit and to have not more than 15 callbacks outstanding at the end of any workday. Timing of the Enumeration The enumeration began as scheduled on April 1, 1960. Previous to that date, the enumerators had completed an intensive training program, including practice exercises and household interviews. By the end of April the enumeration was 97 percent complete, and by the end of May only a few special cases remained to be completed. The taking of the census coincided generally with the Easter vacation schedule. Normally, the schools in Puerto Rico are closed during the week preceding Easter. In 1960, Easter came on April 17 and, therefore, the regular vacation extended from April 9 to April 17. In order to meet the census schedule, teachers who served as census enumerators were released from classroom duties on April 1, eight days in advance of the regular vacation date. This enabled most of the enumerators to complete their assignments before the end of the vacation period and to return to their classes on April 18. Special Problem with Block Numbers One problem which required special attention was the confusion in the assignment of block numbers in the three largest cities--San Juan, Ponce, and Mayaguez. When the enumerators' map copies were reproduced from the master copy, some of the block numbers failed to print, and many of the enumerators had to improvise their own block numbering systems. As a result, there were duplicate block numbers in several areas, and the numbers on the schedules had to be corrected before the identification codes could be punched in the cards. The correction was made by comparing the enumerators' map copies with the master copies and substituting the numbers shown on the latter for the numbers entered on the schedules by the enumerators. This correction was handled by several staff members from the Bureau of the Census and a group of specially trained clerks. PROCESSING THE DATA General After the entries on the schedules were inspected for completeness in the field office, they were transferred to the Decennial Operations Office in San Juan for processing. The processing of the 1960 population and housing censuses for Puerto Rico included the following principal steps: 1. Receipt and check-in of enumeration schedules 2. "General" coding and verification 3. Occupation and industry coding and verification 4. Card punching and verification 5. Mechanical editing of punched cards 6. Ratio estimation 7. Tabulation 8. Result work: balancing of machine runs, posting, and computation of data The Decennial Operations Office in San Juan was opened in April 1960 and closed in September 1961. Employment varied from 30 to 87 persons during the period of data-processing activity. Recruiting and appointment of personnel to positions in the Puerto Rico Office were carried out in accordance with regular Civil Service Commission competitive standards. All persons hired were given temporary appointments for a period not to exceed 1 year. The average length of employment was about 9 months. An experienced statistician and a machine tabulation supervisor were detailed from the Bureau of the Census in Washington to serve as Chief of the Operations Office and Chief of the Machine Tabulation Section, respectively. With the exception of these two persons, all other positions in the Operations Office were filled by locally recruited personnel. At various times, as special needs developed, specialists from Washington were sent to Puerto Rico to handle specific problem areas. The Bureau rented the following equipment from International Business Machines Corporation for the Puerto Rico Operations Office: Receipt and Check-In of Enumeration Books Prior to the enumeration, Minor Civil Division (MCD) Sheets were prepared which listed the ED's within individual areas (city, town, village, or barrio) in Puerto Rico. The MCD Sheets were used by the field offices to insure that each of the ED's was enumerated and that the basic enumeration book and sample enumeration schedules for each ED were submitted to the Operations Office. The receipt and check-in operation involved comparing and reconciling any differences between the field office copy of the MCD Sheets and the original MCD Sheets, and checking in each enumeration book by verifying municipio and ED numbers on the enumeration book labels against those on the MCD Sheets. Coding After the enumeration books had been checked in, the data were coded. Coding involved preparation of entries on a schedule for transfer to a punchcard. Many codes could be punched directly from the schedule entries. Other codes were entered on the schedules by coders prior to punching. General coding.--The general coders entered all required codes except those for occupation and industry (which were coded separately as described below) and also examined for completeness certain items which had been precoded by the enumerator. The following items were coded or examined by the general coders: Sample enumeration schedules In addition, group quarters were coded to differentiate between institutions, i.e., jails, mental hospitals, etc., and noninstitutions such as boarding houses, military installations, etc. The coders used a manual, General Coding Instructions, which gave the codes in full detail, and a summary Code Card which listed the most commonly used codes. General coding was checked by a system of quality control. The work of each coder was completely verified at first, and records were kept on his work. Coders who reached a specified level of accuracy were designated "qualified coders," and thereafter only a certain percentage of their work was verified. Verification of the coding was also checked. The method used (following the pattern used in quality control of the 1950 U.S. censuses) was to make a note of coding errors before the portfolio of enumeration books went to the verifier; after the verifier completed his work, the errors he listed were compared with those previously noted. General coding began June 1, 1960, and lasted 18 weeks. The coders processed the schedules at an average rate of 1,900 lines (persons enumerated) per man-day. Industry and occupation coding.--Questions on industry, occupation, and class of worker were asked for persons 14 years old and over in every sample household enumerated in Puerto Rico, as in the United States. The industry and occupation coding clerks coded responses on industry, occupation, and class of worker and also entered codes for persons serving in the Armed Forces. The coders used a manual, Industry and Occupation Coding Instructions, which gave detailed instructions, an Alphabetical Index of Occupations and Industries, and "Company Name Lists" (one for each of the three SMSA's and one covering the remaining territory in Puerto Rico) to determine the codes for the schedule entries. The verification procedure used was basically the same as that for general coding. Industry and occupation coding started in mid-June 1960 and proceeded at an average rate of 1,600 sample persons per man-day. This operation took 14 weeks to complete. Card Punching After the schedules were coded, information was transferred to punchcards. A card was prepared for each person and for each housing unit. Each column on a card represented a characteristic of that person or housing unit. For example, column 23 on the Complete-Count Population Card was for the sex of the person enumerated. On the card for a male, a hole was punched in the space allotted to "1," the code for "male." When such a card was tabulated, one male was counted. Four separate punching operations were involved, consisting of: 1. Approximately 1,735,000 Complete-Count Population Cards, each requiring 39 columns of punching 2. Approximately 620,000 Sample Population Cards, each requiring 79 columns of punching The punchcard operators began work on the Complete-Count Population Cards in mid-June 1960 and completed this operation by October 1, 1960. The Complete-Count Housing Card punching operation started a week later and was completed by October 28, 1960. Punching for the sample data began the second week in July 1960 and was completed January 26, 1961. The entire card-punching operation required a total of 3,288 man-days of labor and cost $32,566. The punching was verified by a system of quality control. Initially, each punchcard operator's work was completely verified. When he met certain minimum standards of efficiency, he became a "qualified puncher," and thereafter, only a small percentage of his work was verified. Mechanical Editing Mechanical editing involved inspection of the punchcards by machine to detect errors in punching, coding, or enumeration. The machines made three types of checks: (1) checks for impossible codes, (2) checks for inconsistent code combinations, and (3) checks of quantitative data for relationships, magnitudes, and arithmetic. For example, only two entries in the column for "electric lighting" had meaning--a "1" if the housing unit had electric lighting, and a "2" if it did not; any other numbers punched in that column were impossible codes, and the cards having them were withdrawn and corrected. Mechanical editing greatly simplified examination for errors. For most of the card types, unusual or inconsistent reports could be detected in one run through the 101 Electronic Statistical Machine, and the cards removed for investigation. The selected cards were machine-listed, and the listed items were examined for possible errors. Other items on the card sometimes explained the entry or indicated the correct one. If they did not, the enumeration book was examined. If changes were necessary, new cards were punched. If no change was necessary, an explanation was entered on the listing sheet. Then, if the figures were questioned in the critical examination of the tabulations, the listing sheets usually provided the explanation. Inflating the Sample: Ratio Estimates The selection during enumeration of every fourth housing unit and its occupants for the sample, after a random start in each ED, resulted in collection of sample data for approximately 26.4 percent of the total population and 25.7 percent of the housing units in Puerto Rico. The procedure established for inflation of the sample required specified counts for population and for housing from both the 100-percent and the sample punchcards. For this purpose, the cards containing only items collected on a 100-percent basis (representing approximately 75 percent of the population and housing units) were grouped into weighting areas, by municipio and city (selected larger cities) and, in the three largest cities (e.g., San Juan, Mayaguez, and Ponce), by groups of barrio. For each weighting area, the cards were sorted and counts were obtained for each of the following classifications: Male: Heads of households Other members, 14 years old and over Other members, under 14 years old Female: Heads of households Other members, 14 years old and over Other members, under 14 years old Housing units: Owner occupied Renter occupied While these operations were being carried out, a similar series of operations was being performed on the punchcards containing both 100-percent and sample items for the 25-percent sample of population and housing units. First, complete clerical editing, coding, and verification operations were performed, then card punching and verification, followed by a mechanical edit covering the same items as in the corresponding edit of the basic cards. The cards for the sample were grouped into weighting areas, and the cards were sorted into, and counts obtained for, the identical classification established for the basic cards. For each weighting area, the counts obtained were posted on a "Ratio Estimate Control Record." For each weighting area--municipio, or group of barrios and subbarrios--the total card count for each of the ratio-estimation groups was required to be at least 50. If it was not, the ratio-estimation groups were compressed in a specified order until each remaining group contained 50 or more cards. The 100-percent data for each weighting area were checked for consistency between the number of heads of households and the number of occupied housing units. The number of occupied housing units was adjusted to agree with the total count of cards for heads of households. Within each ratio-estimation group, the sample cards were adjusted by duplication or elimination so that the sample cards were one-fourth of the complete count. Thus the sample could be tabulated with a constant weight of 4. In this processing, no sample card was duplicated more than 4 times (i.e., the greatest weight for a card originally in the sample could not exceed 16). The file of cards for persons and housing units that were not in the sample was adjusted so that the adjusted numbers were three-fourths of the complete count. The total counts of population and housing units as obtained from the sample cards thus agreed with the total count of population and housing units obtained from all cards for each weighting area. It was necessary to provide the sample with a uniform integral weight--in this case 4--in order to permit the cards to be tabulated on the IBM 101 (unit count) equipment. The machine then would count each card once, but the resultant tabulations could be clerically multiplied by 4 to obtain the inflated sample results. Individual tracts, places of 2,500 to 10,000 inhabitants, and the rural balances of municipios were not weighting areas, therefore population and housing-unit estimates derived from the sample differed from the complete-count data for some tracts, small urban places, and rural balances of municipios. An additional level of ratio estimation was carried out for some statistics published for Puerto Rico, when the class in question formed part of a larger group for which both a sample estimate and a complete count were available. The improved estimate was made by multiplying a proportion based on the sample data by the figure which represented the complete count of the base of the proportion. The improvement was more important when the characteristic being estimated was a substantial part of the larger group; when the proportion was small, the improvement was relatively minor. Each published report describes the particular tables affected. Tabulating The data collected by the enumerators reached the tabulating process in the form of punchcards. The tabulating machine counted the number of entries in each classification and printed the results. The subject specialists put the general requirements into the form of table outlines which were translated by the machine specialists into tabulation specifications. The specifications indicated how the cards should be grouped and how the control panels on the machines should be wired to provide the data for the proposed publications. The more than 3,000,000 cards punched from the enumeration schedules were sorted mechanically into the groups for which totals were needed. The cards for each municipio were tabulated as a work unit, and each tabulation sheet carried separate totals for each group into which the cards had been sorted. Totals for the municipio were obtained by adding the printed totals for each barrio, sub-barrio, city, etc. The figures on the tabulation sheets were critically examined by the subject specialists before they were transferred to final publication tables. |