from a landlord, superintendent, or neighbor. Because of the nature of the items on the H-8 Characteristics Schedule, the enumerator was required to make three visits to the dwelling unit in an attempt to complete a Characteristics Schedule before returning the segment folder to the field office. In midNovember 1959, the offices were instructed to limit the enumerators to only two visits per segment, to expedite completion of the program and to reduce costs. Timing of the Enumeration The enumeration began on schedule during the week of October 19, 1959, in the 17 metropolitan areas and the following week in the remaining areas, and was expected to be completed in all areas on December 19, 1959, two months later. In late November, it became apparent that the cost of enumeration would exceed the estimated budget. The number of segments in each of the eight new SMSA's was reduced from 1,400 to 750, and the staff in each of these offices was reduced to one crew leader, one reviewer, and 15 to 20 enumerators; at the end of December, the remaining work in the eight new SMSA's was transferred to the Regional Offices and these SCARF offices were closed. By the scheduled completion date, December 19, about 90 percent of the enumeration was completed; one month later, 95 percent of the segments had been enumerated. Enumeration of the remaining segments was continued by a greatly reduced staff. On March 25, 1960, all field operations were stopped to prevent conflict with the 1960 population and housing census operations. Seventy-eight segment folders were returned to Jeffersonville unfinished. Adjustments were later made for these segments by adjusting the weights of completed segments of similar type. The average time spent to enumerate an area segment was 4.9 hours in areas in which the basic 1956-59 procedure was used and 5.7 hours in areas where the 1950-59 procedure was used. Progress Reports The enumerators maintained cumulative progress reports on a special form and once a week reported by telephone to their crew leaders, who posted the information on another form. The crew leaders summarized the information and reported the results to the field office, which combined the reports from the crew leaders and submitted teletype reports to Washington on the cumulative progress of the enumer ation. Items reported included the number of segments assigned, the number completed, the number of hours worked, and the number of hours spent in travel. Quality Control of the Enumeration Procedures used to secure and maintain the quality of enumeration included the following: (1) direct observation of all enumerators during the initial assignments, (2) comparison of a prelisting with enumeration results in certain segments, (3) field office review of the enumerator's work, (4) additional observation by the supervisor or crew leader when required, and (5) verification of enumeration by reinterview. Instructions on the quality control procedures were included in the Field Operational Manual. Prelisting segments.--Two of the first five segments to be assigned to each enumerator were prelisted by the crew leaders prior to the training period. The crew leaders visited the segments and by direct interview obtained the name of the household head and the address of every dwelling unit in the segment and entered the information on a prelisting sheet. During the office review of the first group of segments, the units on the prelisting sheet were compared with the 1959 units entered by the enumerator on the H-7 or H-7a Inventory Changes Schedule. If there were discrepancies, the segment was assigned to the crew leader for followup, and if the enumerator was in error, the discrepancies were immediately called to his attention and corrected. Observation of enumerators.--The assignments for enumeration were scheduled so that each enumerator was observed by the supervisors, crew leaders, or other trained personnel during the enumeration of initial segments. Field office review.--The office review procedure was divided into two phases: the full review and the partial review. The first five segments completed by each enumerator were given a full review in the presence of the enumerator, to give the reviewer the opportunity to point out mistakes and the necessary corrective steps to be taken. After the review of these first segments was completed, only every seventh segment was given a full review, and a partial review was performed for the remaining segments. Both the full and partial reviews included checks to determine if all of the required schedules had been filled and if the identification items on each schedule had been entered properly. In addition, certain critical items on each schedule were examined for incomplete or inconsistent entries: The 195059 or 1956-59 comparisons were examined in detail, as well as the selection of the sample for the H-8 Characteristics Schedule. For segments in which the 1950-59 comparison was made, the 1950 records were searched and reviewed to verify the entries on the H-7a and H-14 schedules. In the full review, the assignment of the 1959 dwelling-unit numbers, addresses, and the adequacy of the FOSDIC marking on the H-8 Characteristics Schedules were also checked. In addition, a tally of all missing or incomplete items was made. Some discrepancies were corrected by the reviewers on the basis of the information provided by the enumerators. In the case of segments for which errors could not be corrected with reasonable certainty, or for which there were several missing entries on a schedule, the material was returned for followup enumeration. When the segment was returned from the followup enumeration, only the items which had been questioned were reedited by the reviewers. Reobservation of enumerators.--If an enumerator appeared particularly inaccurate during the initial observation, or if a consistent pattern of errors was found in his work, the enumerator was scheduled for reobservation by his crew leader. During the reobservation, the crew leader pointed out mistakes and referred the enumerator to the manual when necessary. If the quality of his work continued to be below acceptable standards, the enumerator was dismissed. Reinterview by crew leaders.--A sample of households selected from the followup segments was reinterviewed by the crew leaders. If an enumerator completed his assignment without a followup, the sample of dwelling units for reinterview was selected at random from segments which had passed review. In segments which required followup visits, only the units regarding which errors had been made were scheduled for reinterview. The crew leader filled a "Reinterview Report" form for each reinterview and called the enumerator's attention to any errors or discrepancies found. Phase II Enumeration The ratio-estimation procedure for estimating the number of dwelling units added to the housing inventory by new construction in the 1950-59 period required a second visit to the area segments to determine how many of the housing units enumerated by the 1960 census enumerators fell into the Components of Change Survey segments. This was referred to as phase II enumeration. Training and enumeration for phase II were conducted in June and July 1960 under the supervision of the Census Regional Offices. The enumerator was supplied with the following materials: 1. The segment map 2. The H-141 phase II schedule (on which the dwelling-unit data from the Components of Change Survey basic enumeration were pretranscribed) 3. The appropriate 1960 census listing book The enumerator visited the segment and listed all structures in it, residential and nonresidential, on the H-141 phase II schedule. He searched the 1960 census listing book for all housing units enumerated in those structures and transcribed data for those units to the H-141 schedule. Each dwelling unit enumerated in the Components of Change Survey was pretranscribed on the H-141 schedule, and the enumerator's last step was to indicate on the H-141 schedule whether each of the Components of Change Survey units was in the list of units he had transcribed from the 1960 listing book. The sample counts of 1950 and 1960 units from the regular enumeration and from the second visit were used with the 1950 and 1960 census totals (which were based on 100-percent enumeration) to calculate a growth factor for the decade which was applied to new-construction units for the 1950-59 period. In effect, phase II made it possible to provide an improved estimate by taking advantage of the correlation between counts of new construction and the growth between the 1950 and 1960 housing censuses. Phase II counts were also used to adjust the weights of segments for which there was some indication that the enumerators' canvass might have been inadequate. Processing was divided into two major phases: (1) the clerical processing in Jeffersonville, Ind., and (2) the mechanical processing on conventional and electronic equipment in Washington, D.C. The general clerical operations consisted of the following: 1. Receipt and control 2. Clerical editing and coding of the comparison items on the H-7, H-7a, and H-14 schedules 3. Transcribing 1950 and 1956 characteristics data to H-116 and H-116a transcription forms and H-8 Characteristics Schedules 4. Clerical editing and coding of selected items on the H-8 Characteristics Schedules With the exception of some preliminary editing on conventional punchcard equipment, the mechanical processing of the data was done on electronic computers (Univac I and Univac 1105). Approximately 160 programs of varying magnitude and complexity were required to process the counts and characteristics for the Components of Change Survey. The large number of programs was necessary because (1) the tabulations of the 1950-59 comparisons were different from those for the 1956-59 comparisons; (2) different weights and geographic boundaries were used for the two comparisons and for the national versus the SMSA tabulations; and (3) the design of the different schedules differed so much that separate programs for each schedule were required for many of the processing steps. Listed below are the principal steps for the mechanical processing of the data; all except the first were done on the computers. 1. Edit of the punchcards for inconsistent 1950-59 and 1956-59 comparison entries 2. Preparation of a Master Segment Record which contained geographic identifications, various segment weights, etc., for each segment in the survey 3. Collation of all schedules with the Master Segment Record on a segment-by-segment basis 4. Computer edit for inconsistent comparison entries 5. Ratio estimation of the counts of components 6. Computer edit of the Characteristics Schedules for inconsistent entries 7. Ratio estimation of the characteristics subsample 8. Recode of the characteristics in preparation for the tabulation 9. Tabulation of the characteristics After step 4 above, the counts and characteristics for the 1950-59 comparison, which was to be published in part 1 of volume IV of the 1960 Census of Housing, were processed in the computer separately from the 1956-59 comparison which was to be published in part 2 of the same volume. In order to insure early publication of the counts of the components and some of the basic characteristics for the 1950-59 comparison, part 1 was divided into part 1A and part 1B, and the data for part 1A were processed first. Thus, the components-of-change data were processed in the following order: part 1A, part 1B, and part 2. Receipt and Control Operation A receipt and control operation was set up in Jeffersonville to insure that all necessary materials were received from the field offices, and to file and maintain records so that an orderly flow of the materials to each of the processing steps could be maintained. A master file card was prepared for each area segment, and a listing was prepared from the file cards to record the location of each area segment during the many steps in the clerical operation. Clerical Editing and Coding of Comparison Items Since the items used for the 1950-59 and 1956-59 comparisons were the most critical ones in the survey, these items on all of the basic enumeration schedules were given a complete and thorough edit. Editing and coding the H-7 Inventory Changes Schedules.--For the segments which had been in the 1956 National Housing Inventory, the 1956-59 comparison items were edited first, to verify that the enumerator had accounted for all 1959 dwelling units in the segment as well as all 1956 units and had made the proper entries for the comparison. Related items were used in editing the comparison items-year built, year of change, and, when necessary, additional items such as the addresses, number of dwelling units in the structure, and number of rooms for both the 1956 and 1959 units. The enumerator's notes in the "comments" section of the schedule were also used. The 1950-59 comparison was established for units in these segments by a coding operation which utilized the 1956-59 comparison items and the 1950-56 comparison that had been transcribed from the National Housing Inventory records. Chart 1 illustrates how the 1950-59 comparison was coded for some of the units. Editing the H-7a Inventory Changes and H-14 Address Sample Schedules.--The editing of the H-7a Inventory Changes Schedule on which the 1950-59 comparison was obtained by direct enumeration was similar to the editing of the H-7 schedule. However, use was made of the 1950 census records instead of the 1956 National Housing Inventory records in the editing. An H-14 Address Sample Schedule was required for all segments which had not been in the 1956 National Housing Inventory except new-construction segments. Entries indicating that the 1950 dwelling units corresponding to the P-1 1950 census schedule serial numbers had not been found by the 1959 enumerator were verified by searching through the 1950 census records. If the 1950 unit was found by the editor, the H-14 schedule was set aside for return to the field. If the 1950 address on the H-14 schedule represented a structure which was still residential, no further processing was necessary because this structure was not lost from the inventory. If the address represented a structure that was classified as entirely for nonresidential use, unfit for human habitation, demolished, etc., the editor searched through the 1950 census records to verify that the enumerator had entered on the H-14 schedule the 1950 P-1 serial numbers for every 1950 unit in that structure. The enumerator's notes in the section on the schedule for comments were also used in the editing. Chart 1.--ILLUSTRATION OF OFFICE DETERMINATION OF 1950-59 COMPARISONS FOR AREA SEGMENT WHICH HAD BEEN IN 1956 NATIONAL HOUSING INVENTORY Editing the special and combination segments.--Except for minor modifications, editing for combination segments where both H-7 and H-7a Inventory Changes Schedules were used and editing for segments which required special enumeration procedures was done according to the regular specifications for editing of the H-7 and H-7a schedules. In combination segments, the units in each set of schedules were separately identified and each set of schedules was processed separately. Verifying the characteristics subsample.--The H-8 Characteristics Schedules were verified to determine that they had been filled for the proper 1959 units in the characteristics subsample. Where no bias was apparent, H-8 schedules filled for units not in the characteristics subsample were accepted. Where H-8 schedules were missing, the basic identification and components-of-change data were transcribed from the H-7 or H-7a schedules to blank H-8 schedules, for use in subsequent collation operations and in the Residential Finance Survey. Problem referrals.--Most of the problems in the editing and coding which were not covered by the clerical specifications were resolved by the clerical supervisors or by housing-subject specialists. Some of these problems required detailed examination of the 1950 census records. Missing or incomplete entries on the basic enumeration schedules (H-7 and H-7a Inventory Changes and H-14 Address Sample Schedules) were supplied where possible by the technicians. A number of segments were returned to the field for reenumeration. Transcription Operation To prepare the data on counts and characteristics of the 1950 and 1956 components for tabulating, a transcription operation was performed. For the 1950-59 comparison, data for all 1950 units which were changed by conversion or merger or which were lost to the inventory were transcribed to an H-116a Transcription Form for the 1950 data. Similarly, for the 1956-59 comparison, data for all 1956 units that were involved in a conversion or merger, or that were lost to the inventory after 1956, were transcribed to the H-116 Transcription Form for the 1956 data. The 1950 and/or 1956 data for units that remained the same between 1950 and 1959 and/or between 1956 and 1959 were transcribed to the H-8 Characteristics Schedules. In addition, 1959 items enumerated on the H-7 or H-7a Inventory Changes Schedules were transcribed to the corresponding H-8 Characteristics Schedules in FOSDIC markings for all units in the subsample. Clerical Editing and Coding of Selected Items on Characteristics Although the H-8 Characteristics Schedule was basically a FOSDIC schedule, entries for the household occupying the unit and their previous residence were recorded by the enumerator in a conventional manner. These entries were coded in special FOSDIC markings in the Jeffersonville office. During the coding operation, inconsistent or missing entries in the related items were edited. However, the major edit of the items on the H-8 schedule was done in the computer. Coding and editing household items.--Data for the household characteristics were coded based on entries in the following items for each member of the household: relation to the head of the household, sex, and age. Missing entries in these items were edited by using data in related items and special allocation sheets which contained predetermined categories for relationship and age. Coding previous residence for recent movers.--For all households that had moved during 1958 or 1959, the complete address of the previous dwelling unit (street number, city or town, county, and State) was recorded. The location of the previous unit in relation to the location of the present unit was coded (e.g., "outside SMSA, different State"). Verification Procedures A verification procedure for each of the major clerical steps was instituted to improve and control the quality of the work. The results were also used to establish a criterion of acceptable performance for the editors, coders, and transcribers. Verification of clerical editing and coding of H-7, H-7a, and H-14 schedules.--Verification of the editing of the basic schedules for regular segments was done on a line-by-line (dwelling unit-bydwelling unit) basis. All segnents completed during the first 5 days of editing were verified on a 100percent basis. Errors were recorded and during the qualifying period the errors were pointed out to the editor-trainees individually. If the error rate was more than 3 percent during this period, the trainee failed to qualify as editor and was assigned to other duties. After the qualification period, verification of the editing was done on a sample basis, one segment from each work unit of 10 segments. Because of the varied editing procedures required for special and combination segments, the editing for these segments was verified on a 100-percent basis. Other verification procedures.--The transcription of 1950 and 1956 data to the H-116a and H-116 transcription forms and to the H-8 Characteristics Schedules was also verified. To qualify as a transcriber, an error rate of less than 4 percent had to be maintained for the first four work units (40 segments). Once the transcriber had qualified, one out of every 20 segments was verified. During the sample verification, if the error rate rose over the established maximum, two work units were verified until the transcriber either performed acceptable work or was removed from the operation. The coding of the selected 1959 items was verified on a 100-percent basis. Preparation of Schedules for Mechanical Processing Card layouts and punching specifications for each type of schedule were prepared, and a card was punched for each dwelling unit. Each H-116a and H-116 transcription form represented one 1950 or 1956 dwelling unit, and each line on an H-7 and H-7a Inventory Changes Schedule represented one 1959 dwelling unit. The punching was verified on a 100-percent basis. The H-8 Characteristics Schedules, which were FOSDIC schedules, were microfilmed by a procedure similar to the one for microfilming the 1960 census schedules, described in part I, chapter 8. A breaker sheet was prepared for each tabulation unit. It contained such identifying information as the tabulation-unit number, the geographic area code, and the number of H-8 schedules in the tabulation unit. Clerical Editing of Phase II H-141 Schedules After the phase II enumeration had been completed, all related materials were returned to Jeffersonville where the H-141 phase II schedules were edited and the basic 1960 housing unit counts for each segment were prepared for tabulation. The clerical editing was carried out by a small group of clerks in Jeffersonville during the period July-November 1960. The editing procedure consisted mainly of searching the 1960 census listing book and verifying that all housing units in the segment had been transcribed to the phase II schedule. If some 1960 housing units had been missed, they were added; conversely, if some housing units had been transcribed erroneously, they were deleted. The total number of |