it would not be possible to obtain accurate information on this subject in the Components of Change Survey. For single demolitions in residential sections that remained for the most part intact, it was generally possible to determine from respondents living nearby whether the authorization was private or government. For large-scale demolitions where whole residential sections had been demolished, it was difficult and time consuming to obtain a reliable respondent in the vicinity. This item was retained on the basic 1956-59 enumeration schedule at the request of some of the users of housing data but with the understanding that the data would represent, at best, only rough estimates and were not to be published.

Some new procedures designed to raise the quality of the enumeration were instituted in this pretest. To improve coverage, the supervisor or crew leader prelisted all dwelling units in the first two or three area segments before assigning them to the enumerator. During the field review of the enumerator's work, this prelisting was compared with the dwelling units enumerated in the segments by the enumerator, and all discrepancies were immediately reconciled. Another method to improve the quality of the enumeration was to assign each enumerator a small number of area segments at the beginning of the enumeration. When enumeration of these segments was completed, the enumerator reported to the field office and his work was immediately given a complete review, and all discrepancies and errors and the necessary corrective actions were discussed with the enumerator.

Extensive observation by the Washington staff and comments provided by enumerators, crew leaders, and reviewers served as a basis for determining methods and ways to improve the techniques of the survey. An analysis of the enumeration documents was also made.

PREPARATORY WORK

Screening National Housing Inventory Materials

One of the first major tasks in the preparatory work for the enumeration of the 1959 Components of Change Survey was to ascertain the extent to which National Housing Inventory schedules had been lost or mislaid in the interval between the two surveys, and to evaluate the impact of the incomplete or missing 1956 materials on the 1959 survey. Another important preparatory task was the detailed examination of the National Housing Inventory area-segment maps. Screening for missing materials and examining the maps were combined into one clerical operation. From the middle of February until the second week of August 1959, the clerical staff examined the contents of approximately 22,000 National Housing Inventory segment folders.

There were approximately 100 area segments for which there were adequate maps but no basic National Housing Inventory schedules. For these segments, a special enumeration and editing procedure was established. There were approximately 2,500 area segments for which maps were entirely missing. New maps were prepared for 2,000 of these and the remaining 500 segments were deleted from the program. For another 800 subsegments, the boundaries as delineated on the maps did not meet the specifications for the 1959 survey. Of these, 150 required special delineation in the field, and for the remaining 650, boundaries were redefined and several special enumeration and processing procedures were established.

About 700 segments were found to be too large. To reduce their size, Current Population Survey enumerators subdivided these segments into subsegments and returned the maps to Washington. In Washington, the maps were reviewed and edited to assure definable subsegment boundaries; sample subsegments were selected to represent the segments, and the information necessary to adjust the weights for these segments was obtained.

New-Construction Universe

In order to improve the estimates of new construction, a separate universe was established. In the areas covered in the 1956 National Housing Inventory program, the new-construction universe was established by updating the 1950-56 new-construction universe used in the 1956 survey. Large clusters of 25 or more new units built during the period 1957-59 were identified by enumerators who made inquiries of informed local persons, such as building permit officials, tax assessors, and builders. Whenever possible, the enumerator obtained subdivision maps, but if none was available he prepared a map for each cluster according to specific instructions in his manual. In the areas that were not covered in the 1956 National Housing Inventory survey, the same procedure was used for establishing the new-construction universe, except that large clusters of new units built during the 10-year period 1950-59, instead of 1957-59, were identified and delineated.

The maps of new-construction clusters were transmitted to Bureau headquarters in Washington where the regular area-segment maps were screened for possible overlap. If the regular area segment was partially or completely overlapped by a new-construction cluster, the basic enumeration procedure for the overlapped segment or part of the segment was changed. A sample of the new-construction clusters was selected and then returned to the field for segmentation into smaller land areas. The subsegmented maps were then sent to Washington where a sample of about 8,500 segments was selected. Intensive editing of these operations was conducted in the field offices and in Washington.

The delineation of large clusters of new contruction and the subsequent segmentation of selected clusters was done by Current Population Survey enumerators. The training consisted of approximately 2 hours of reading the specifications for delineation and segmentation of the clusters, and completing several practice exercises.

The identification of large clusters of new construction began in early April 1959, and the bulk of the work was completed by the middle of August 1959. Work on subsegmentation and selection of the new-construction sample could not begin until mid-August and was not finished in some areas until December 1959, several weeks after enumeration had begun.

Mailing Materials to the Field

All materials for the Components of Change Survey were shipped first to Jeffersonville and from there to the various field offices. The flow of materials to Jeffersonville started when records for 5,000 1950 ED's were shipped from Washington in April 1959. The rest of the 1950 materials were sent periodically throughout the next few months, the last shipment on September 1, 1959. The initial shipment of 1956 National Housing Inventory materials began immediately after the screening operation was completed in August. The mailing of the balance of 1956 materials continued periodically until October 1, 1959, when the last shipment was received in Jeffersonville.

All new materials such as the components-of-change schedules and segment folders were mailed to Jeffersonville from the second week of August through September 15, 1959.

Transcription of 1956 Data to Questionnaires, and Other Preparatory Work

Pretranscription of selected data from the 1956 National Housing Inventory to the basic enumeration schedules was required for all segments which had been in the 1956 National Housing Inventory program. The pretranscribed data included items utilized in the 1956-59 and 1950-59 comparisons, such as complete address of the 1956 dwelling unit, number of rooms, and number of dwelling units in structure. Items that were to be used in the later clerical processing were also transcribed--e.g., the 1956 dwellingunit number and the 1950-56 comparison. Subsample units for which data on characteristics were to be collected were designated on the basic enumeration schedules for all segments.

This preparatory work, which included also the preparation of segment folders and related materials for all the segments, including those that were not enumerated in the 1956 National Housing Inventory, was carried out in Jeffersonville. The operation began during the middle of August 1959 and was finished by the middle of October.

Field Organization

In order to facilitate the enumeration of approximately 26,000 segments located in 650 counties and cities in the United States, 22 SCARF offices were established. These offices were classified into three groups for administrative purposes:

Group A consisted of 12 SMSA offices located in cities with a permanent Census Regional Office and administratively responsible to the Regional Office. They were responsible for the enumeration of their respective SMSA's and also for the enumeration of all the national sample except that done by group C offices. These offices were located in Atlanta, Boston, Chicago, Dallas, Detroit, Los Angeles, New York, Philadelphia, Pittsburgh, St. Louis, St. Paul, and Seattle.

Group B consisted of the five remaining SMSA offices and were located in cities that did not have a Regional Office. They were located in Baltimore, Buffalo, Clevelɛ nd, San Francisco, and Washington. They were responsible only for the enumeration of their respective SMSA's.

Group C consisted of five offices responsible only for the enumeration of the remainder of the national sample segments outside the regions covered by the 17 SMSA offices, and was located in cities having Census Regional Offices. These SCARF offices were under the administrative direction of the Regional Offices, like the offices in group A. The offices were in Charlotte, Cincinnati, Denver, Kansas City, and New Orleans.

The three groups of offices were staffed in general with the following personnel:

In all offices in groups A and B, a number of regular Regional Office enumerators were assigned to the SCARF enumeration. They were supplemented by temporary enumerators who had passed a qualifying test. All personnel in the group C offices were regular Regional Office enumerators.

The regular Regional Office enumerators received their usual hourly rates which averaged $1.82 an hour. All temporary enumerators and almost all the temporary employees serving as clerks received $1.69 per hour. A few clerks received the same remuneration as the crew leaders which was approximately $1.95 per hour.

Training Program

Training materials.--Work on revising the enumerators' manual began immediately after the Philadelphia pretest. For the final program, two manuals were prepared, one for each basic enumeration procedure. The Enumerators' Manual numbered Form 60-08-101 contained detailed instructions for enumerating the area segments which had been in the 1956 National Housing Inventory program. The Enumerators' Manual numbered Form 60-08-101a contained the instructions for the area segments which had not been in the 1956 National Housing Inventory.

Two training guides were prepared also, but because both basic enumeration procedures were used in all areas, it was necessary to include in each training guide not only the instructions for the basic procedure for the area but also supplemental material for the alternate procedure. The training guides included the lectures to be given by the trainer and also exercises, examples to illustrate the difficult points, mock interviews, and written tests. Two film strips which had been used in training for the 1950 census, "Dwelling Unit and Type of Structure" and "Condition of Unit," were used again for the 1959 training program.

A training innovation for the Components of Change Survey was the use of a "VU-graph" projector, to project slides on a screen or wall, which permitted modifications of the slides to be made with a special pencil by the instructor. For example, the instructor could record the correct answers on the schedule projected on the screen during a mock interview.

Supervisors and crew leaders.--The training of SCARF supervisors and crew leaders was done in five locations by teams from Washington headquarters. It began September 21, 1959, in Boston, Los Angeles, and Pittsburgh, and a week later, September 28, in New Orleans and St. Louis. Training for the supervisors lasted 5 days. The crew leaders attended 32 days of the sessions. In addition, at least one member of the regular staff of each of the Regional Offices attended the first 2 days of the training.

During the training period, the supervisors and crew leaders actually enumerated some of the segments while being observed by members of the training team, to obtain practical experience with survey concepts and procedures and to become acquainted with some of the difficulties encountered in enumeration. Also, a half day was devoted to practice training by the supervisors and crew leaders. This gave them the opportunity not only to familiarize themselves with the training guides and visual aids but also to receive suggestions for improving their training techniques. In addition, it gave the supervisors the opportunity to evaluate the training ability of their crew leaders; in some cases, the supervisor then elected to train the enumerators himself or to combine the two training sessions for the enumerators into one and use the better crew leader as the principal trainer.

Enumerators.--In most of the 17 SMSA's, enumerators were trained in two groups, each group conducted by one crew leader. In a few SMSA's, the supervisor elected to combine the sessions. The sessions for the new enumerators began October 19, 1959, and continued 2 days. The regular Regional Office enumerators were given only 1 day of training because they were already familiar with many of the concepts and definitions.

Offices in group C did not begin training until the following week. Since the personnel were all regular Regional Office enumerators, they required only 1 day of training.

In all areas, each enumerator was observed by the supervisor, crew leader, or technician from Washington as he conducted the enumeration in the first few segments. Enumerators who had appeared weak in understanding of the instructions during training were the first to be scheduled for observation.

THE ENUMERATION

Enumeration Forms

The basic enumeration documents and control forms are described briefly below. Facsimiles of the schedules are reproduced in a separate publication.1

H-7, Inventory Changes Schedule.--This was the basic enumeration document for determining the components of change in the area segments that had been included in the 1956 National Housing Inventory. It contained pretranscribed addresses and selected data for the dwelling units reported in the segment in the 1956 National Housing Inventory. It was used by the enumerators to list and compare all 1959 dwelling units in the segments with the data for the 1956 National Housing Inventory units. The subsample of 1959 units for which detailed data on characteristics were to be collected was predesignated on the schedule.

H-7a, Inventory Changes Schedule. --This was the basic enumeration document for establishing the components of change in the area segments that had not been included in the 1956 National Housing Inventory. It was used by the enumerators to list and compare all 1959 dwelling units in the segments with the 1950 census records. It was similar to the H-7 schedule in design except that the enumerator was to record the addresses and selected data for the 1950 dwelling units for purposes of the 1950-59 compari son and the determination of when to fill the H-8 Characteristics Schedule in these segments was established by another form (see H-113, below).

H-14, Address Sample Schedule.--The Address Sample Schedule was generally used in areas where the H-7a was used. The H-14 was used to record the 1950 dwelling units in whole structures that no longer were in existence or no longer contained dwelling units.

H-8 Characteristics Schedule. --This was a FOSDIC schedule which was used by the enumerator to obtain detailed data on characteristics of a subsample of the 1959 dwelling units in each area segment. Space was provided for the clerical transcription of selected 1950 and/or 1956 data for dwelling units classified as "same."

H-113, Enumerator's Control Record.--The Enumerator's Control Record was stapled on the back of the segment folder, or enumerator's kit, and was used by every enumerator. Section I contained control items essential for determining that all schedules were properly filled for each unit in the segment. In addition, in areas where the H-7a schedule was used, selection of the sample of 1959 dwelling units requiring the H-8 Characteristics Schedule was predesignated in section I. Section II was used only in areas where the H-7a schedule was used; in the lower-left corner, serial numbers of 1950 dwelling units for the H-14 Address Sample Schedule were pretranscribed. Section III was used to record special enumeration instructions for a particular segment.

H-117, segment folder. - -The front of the segment folder, or enumerator's kit, contained all the identification items, such as the area-segment number and the primary sampling-unit number, the name of the city and the county, and the type of enumeration. The front of the segment folder also contained code boxes to be used in the office processing and space for entry of certain control data by the field offices.

Enumeration Procedures

1950-59 procedure. --In segments which were not in the 1956 National Housing Inventory program, primarily in the eight new SMSA's, the 1950-59 procedure was used. The enumerator located a specific area segment by using the grid coordinates on the segment map. He then precanvassed the segment by walking around the entire segment. During the precanvassing, he familiarized himself with the exact boundaries of the segment and at the same time listed on the back of the segment folder the addresses of all structures existing in the segment.

The enumerator then searched the 1950 census P-1 schedules to find all 1950 units with the same addresses as those listed in the precanvass. He transcribed certain data for these units, such as address, number of rooms, and number of dwelling units in the structure, from the 1950 schedules to the H-7a Inventory Changes Schedule.

1

U.S. Bureau of the Census, Survey of Components of Change and Residential Finance of the United States Census of Housing, 1960: Principal Data-Collection Forms and Procedures, Washington, D.C., U.S. Govt.Print.Off., 1962, 39 pp.

He then proceeded to list on the H-7a schedule all living quarters in the segment and to determine the 1950-59 components of change. He obtained the answers for the 1959 items, such as address, number of rooms, number of dwelling units in the structure, and year built. By comparing the 1959 entries with the 1950 entries, he established and marked on the schedule the appropriate 1959 component, i.e., same, conversion, merger, from quasi, from nonresidential use, moved to site, or new construction. For conversions, mergers, and other additions except new construction, he also obtained information on "year of change." At the same time, he entered the 1950 components, e.g., same, conversion, merger, changed to residential use, etc.

For a 1959 unit reported by the respondent as new construction or added from other sources, the enumerator again searched the 1950 census records to verify that the unit did not exist in 1950. If a 1950 unit was found for the same address, the enumerator asked enough probing questions to establish definitely either that the unit had been built since 1950 or that it had existed in 1950.

If the 1950 census records were incomplete or missing, the enumerator determined the correct classification by asking the occupants or informed neighbors a series of questions relating to the past and present status of the unit.

After all necessary entries for a dwelling unit had been recorded on the H-7a schedule, the enumerator checked the box opposite the same 1959 unit number on form H-113, Enumerator's Control Record, to indicate that the unit had been enumerated. If the 1959 unit number was circled on the Enumerator's Control Record, he also completed an H-8 Characteristics Schedule for the unit.

For all segments which were not in the 1956 survey except the new-construction segments, H-14 Address Sample Schedules were used to enumerate losses of 1950 whole structures. Because it was not feasible for the enumerator to identify and list on the H-7a schedule the 1950 addresses for structures that were no longer standing in the segment, a sample of six possible 1950 serial numbers had been selected and transcribed to the Enumerator's Control Record. The enumerator entered these serial numbers on the H-14 schedule and searched the 1950 census schedules for these same serial numbers. Usually about three 1950 serial numbers were found. He then copied the addresses of the 1950 structures containing these dwelling units onto the H-14 schedule and determined whether or not the structure was still standing and still contained dwelling units. If the structure was no longer standing or no longer contained dwelling units, he entered the type and year of change--entirely quasi, entirely for nonresidential use, unfit, demolished, moved from site, or other loss--and recorded on the H-14 schedule the 1950 serial numbers of all units reported in the structure in 1950.

1956-59 procedure.--In segments in which National Housing Inventory data had been collected in 1956, i.e., in the nine SMSA's and the remainder of the national sample, the enumeration procedure was similar to that in the eight new SMSA's except that the 1959 dwelling units were compared with the 1956 instead of the 1950 dwelling units.

The enumerator located and precanvassed the segment. He then entered on the H-7 Inventory Change Schedule the data for every 1959 unit existing in the segment and determined the 1959 and the 1956 components by a unit-by-unit comparison with 1956 data which had been transcribed to the schedule.

The design of the survey required an H-8 Characteristics Schedule for essentially the same subsample that was used in the 1956 National Housing Inventory. The units that had been in the 1956 subsample were predesignated on the H-7 schedule by a circle in which the corresponding 1959 dwelling-unit number was entered and the enumerator filled an H-8 Characteristics Schedule for each 1959 dwelling unit entered within the circle.

For units that had not been in the 1956 subsample--new construction, other additions, and those dwelling units missed entirely by the 1956 enumerator--a special section was provided on the H-7 schedule where the H-8 characteristics subsample for such units had been predesignated.

Special Situations

For the National Housing Inventory segments which were partially overlapped by the 1956-59 newconstruction universe, two different enumeration procedures were required and two separate H-7 schedules were used. Poorly delineated 1956 National Housing Inventory segments also required special enumeration procedures. In addition, the updating of the information on 1950 dwelling units which were lost to the inventory by 1956 but which might have changed their status again by 1959 required still other procedures. Detailed instructions for enumerating these segments were placed inside the segment folder, and a brief resume of the special enumeration instructions was given on the H-113 Enumerator's Control Record.

Callbacks

Except in unusual cases, the enumerator was instructed that callbacks were not generally necessary for the H-7 and H-7a Inventory Changes Schedules because the required data usually could be obtained