h. Preparing a weight card (steps 9 and 10) i. Calculating the final weights and punching them into the work cards (steps 11-14 ) j. Transferring the data from the punchcards to computer tape (step 15) k. Tabulating the data (step 16) 1. Preparing the computer printouts for typing for publication (steps 17-19 ) Receipt and Check-In As questionnaires were received from the field offices, they were checked against the control records. As part of the check-in, it was necessary to determine if the property was mortgaged and, if it was, that the required Lender Questionnaires were attached. Screening A screening operation was devised to (1) remove all out-of-scope cases, (2) separate largerental-property cases from the others because they required some additional processing, and (3) set aside some cases having serious problems which had to be resolved before they could go through the regular processing. Property questionnaires were reviewed to see that the correct questionnaire had been used (i.e., homeowner properties on H-10 forms, rental and vacant properties on H-11 forms). Properties reported on the wrong forms were transferred to the correct ones. The presence of more than one property control card and questionnaire for a property was evidence that a property was in the sample more than once. (Since the basic sampling unit was the dwelling unit, a property with more than one dwelling unit had more than one chance of being in the sample.) The number of property control cards for each property, i.e., the number of times the property came into the sample, was entered in a code box on the H-10 or H-11 questionnaire(s). This figure was later used as part of the adjustment of the weight. The amounts of the mortgage payment as reported by the owner and by the lender were compared. If they differed, the case was set aside for investigation, since this was an indication that the two respondents might not have reported for the same mortgage. Finally, cases which were thought likely to represent sampling problems were set aside during the screening operation. Clerical Editing and Coding The primary purpose of the clerical edit was to identify problems, which were then resolved by subject-matter specialists on the basis of all the available information on the questionnaires for the property. For example, an owner might have reported that he assumed the mortgage from the former owner, but failed to report either the amount of the loan or the date of assumption. In this case, the date the property was acquired was used as the date of mortgage assumption; the amount of the mortgage at the time of assumption was obtained by computing the unpaid balance as of that date, using information from the Lender Questionnaire relating to term, interest rate, and the original date the mortgage was made. The data on both the properties and mortgages were examined for reasonableness, consistency among the items, and completeness. Coding was done, as necessary, along with the editing. Some items were precoded; others--such as amounts--were punched as entered. The coding and editing were divided into two parts, for (1) the owner questionnaires (H-10 or H-11), for all properties, and (2) the Lender Questionnaires (H-12), for mortgaged properties. The number of pages of the lender questionnaire and the fact that there might be more than one per property made punching directly from the questionnaires impractical. All mortgage data, from both the owner and the lender questionnaires, were transcribed (or coded, for items requiring coding) to a "mortgage transcription sheet." This transcription sheet contained separate columns for a first mortgage, a second mortgage, and a third mortgage. Transferring Data from Components of Change H-8 Characteristics Schedules to Homeowner The decision to use the same basic sample for both the Residential Finance Survey and the Components of Change Survey meant that certain items of information collected for the Components of Change Survey on the H-8 Characteristics Schedule for the owner, his family, and the dwelling unit could be used for the Residential Finance Survey, thus reducing costs and putting less of a burden on the respondent. The necessary data from the Components of Change Survey H-8 Characteristics Schedules were transcribed to the Residential Finance Survey H-10 Homeowner Questionnaires following a matching procedure. Match of Large Rental Properties and Area-Sample Properties The decision to supplement the area sample with a sample of large rental properties made it necessary to check each area-sample rental and vacant property against the universe of large rental properties to eliminate those area-sample properties which had already had the opportunity of being selected in the sample of large rental properties. Every area-sample property with five or more units was matched against the universe of large rental properties, because owners' responses indicated that some properties with less than 50 units were in the list of large rental properties. Card Punching The data on characteristics of each property and its owner were put on one basic data punchcard, and the mortgage data for mortgaged properties were put on another basic data punchcard. If a property had two or more mortgages, a separate basic data punchcard was prepared for each mortgage. Mechanical Editing After the information on the schedules had been edited, coded, and punched to a basic card, each card was edited mechanically for blanks, impossible codes, unreasonable amounts, and inconsistent entries. The mechanical edit of the punchcards repeated many of the clerical checks for consistency among the items and included some additional checks for consistency and reasonableness. Rejected cards were first checked for punching errors by reference to the original schedule or the mortgage transcription sheet. If the rejected card did not have punching errors, the coding on the schedule was checked by reference to the same instructions as used in the clerical edit. Rejects which were not the result of punching or coding errors were resolved by subject-matter specialists. Imputation of Responses for Blanks Nonresponses to items were handled in three ways during the processing operations. First, part of the intensive clerical editing of each questionnaire received in the central processing office was designed to eliminate nonresponses for as many items as possible on the basis of other information reported for the property or mortgage. For example, "year built" was allocated on the basis of the year the property was acquired and whether the property was acquired new or had been previously occupied. Mortgage items were also handled in this manner; for example, payments for principal and interest were derived on the basis of reported interest rate, term, and amount of mortgage. Nonresponses to items for which data were transcribed from the H-8 Characteristics Schedule were handled by allocation partly on the basis of other items on the schedule but primarily on the basis of information reported for the preceding unit in the same land segment. Finally, when related information was not available for imputing the missing item, an amount was arbitrarily assigned for certain property items such as "purchase price." The assigned amounts were then edited for consistency with other amount items. Preparation of Work Card Following the completion of the mechanical edits, the information on the owner card and the mortgage card or cards (if any) for each property was transferred in a highly compressed form to one punchcard (the work card), using punchcard equipment. With a few exceptions, only the items to be included in the final publications were on the work cards. All data relating to characteristics of junior mortgages, therefore, were excluded. In order to reduce the electronic computer programing task, the data on the work cards were in a form ready to tabulate. Wherever necessary, an item was recoded to reduce it to one or two columns, and, to save card space, two basic items were recoded into one column on the work card whenever feasible. Transferring Weights to Work Cards The data-processing steps involved in assigning a weight to each case before the data on the work cards could be transferred to tape for tabulating in the electronic computer were the following: 1. A listing of H-10 homeowner property cards and H-11 rental and vacant property work cards was compared with a listing of H-8 Characteristics Schedules from the Components of Change Survey which contained the weight for each dwelling unit after the first-stage estimation procedure. A new card--a weight card--was punched, containing this weight, certain identification items, and location of the property. 2. This first-stage weight, which was for the dwelling unit, was adjusted if the property contained more than one dwelling unit, to reflect the property's additional chances of being in the sample. 3. The weight was then adjusted by the factor produced by the ratio-estimation process which compared the raw-weighted counts with related figures from the 1960 Census of Housing. The final weight, and also the location of property information which came from the H-8 listing, were transferred to the work card, which was now ready for the computer. Tabulating the Data The tabulation program was divided into two parts--(1) homeowner properties, and (2) rental and vacant properties--and was reduced to five major tables. (See chapter 9, "The Publication Program.") Some additional tabulations which had been scheduled for publication were to be prepared if funds and time permitted. The tabulations for the homeowner properties were prepared for each of the 17 selected metropolitan areas, the four major geographic regions of the United States, inside all SMSA's, and the United States. The tabulations for rental and vacant properties were limited to U.S. totals. The data on the punchcards (i.e., work cards) were transferred to magnetic tapes, tabulated on the electronic computer, and the results printed on the high-speed printer. The printouts were then ready for the final work to prepare the statistical tables. Preparation of Tables To save programing resources (in short supply at the time), the residential-finance, homeowner tables were programed for the electronic computers without identification of the columns or stubs, without subtotal and total columns, and without the medians for these columns. In lieu of posting the data, the printouts were pasted to preprinted forms containing the stubs and column headings. For homeowner property tabulations, clerks then computed the necessary subtotals, line totals, columnar totals, and medians. For rental and vacant property tabulations, the computer program yielded all the required totals. The tables were then reviewed by subject specialists, and any necessary corrections made. The tables were typed, then a machine check of all columns was made to detect any errors. When any errors found had been corrected, the tables were sent to the printer along with previously prepared text. Quality Control To hold clerical, processing, and punching errors to a minimum, 100-percent independent verification was used for all manual operations in the data processing. Quality-control records were maintained and clerks who demonstrated unacceptable quality of performance received additional training or were assigned to other tasks. PUBLICATION PROGRAM The publication program was based on the following decisions made in the light of the priority of needs expressed by the users: 1. The publication was to consist of two parts--one for homeowner properties and the other for rental and vacant properties. The part on homeowner properties was to be published first. 2. Separate sets of tables on homeowner properties were to be presented for each of the 17 selected metropolitan areas, the four major geographic regions, the United States as a whole, and all the metropolitan-area part of the United States as a whole. In addition, two tables for the United States as a whole were to be prepared to present data on homes with nonwhite heads of households. 3. The tables on rental and vacant properties were to present only data for the United States as a whole. 4. The issuance of data on homeowner properties was to be expedited by using press releases and accompanying summary tables to be prepared for each area as soon as the posted work tables became available. No separate formal publications were to be issued for the separate areas. (The use of press releases was adopted to eliminate the work and expense and delay in release of data which would have resulted from the preparation of a separate publication for each area.) The homeowner property part of the residential finance volume, therefore, was to include the tables for all areas. 5. For both parts of the publication--the one on homeowner properties and the one on rental and vacant properties--the tables were to show (a) the mortgage status of residential properties, (b) whether the first mortgages were insured by Federal Housing Administration or guaranteed by Veterans Administration or were conventional mortgages, in terms of number of properties and amounts of first mortgage and total mortgage debt, and (c) the extent of participation of the various types of mortgage investors--i.e., banks, insurance companies, individuals, etc.--in residential financing. The tables were to present data on the characteristics of the mortgages, properties, and owners in terms of these classifications. In the tables for homeowner properties, the detailed characteristics were to be shown only for the one-unit properties since these constituted the overwhelming majority of the homeowner properties. The first press release for homeowner properties was issued in December 1961 for the Washington, D.C.-Md.-Va. SMSA. It contained data on the number of homeowner properties, their mortgage status, the aggregate amounts of outstanding first and total mortgage debt, the number with junior mortgages, and additional information on the characteristics of first mortgages classified by Government insurance status. The release was also used to inform the public about the availability of photocopies of the complete tables from which the press release tables were extracted. Similar releases were prepared for each of the remaining 16 SMSA's and for the United States as a whole, the latter in July 1962. Volume V of the Census of Housing, Residential Finance, part 1, Homeowner Properties, was published in February 1963. The tables presented in this report are the following: Table 1 shows the number of 1- to 4-unit homeowner properties and the number mortgaged and unmortgaged. For 1-unit homeowner properties, data are shown on characteristics of the properties and owners. Table 2 shows the number of mortgaged 1- to 4-unit and 1-unit homeowner properties by insurance status of first mortgage, i.e., whether insured by Federal Housing Administration, guaranteed by Veterans Administration, or conventional. Table 3 shows the total and average amount of first mortgage debt outstanding on mortgaged 1- to 4-unit and l-unit homeowner properties by government insurance status of the first mortgage. For mortgaged l-unit homeowner properties, the amount of first mortgage outstanding debt is distributed by selected first mortgage characteristics. Table 4 shows the total and average amount of total mortgage debt outstanding on all mortgages on 1- to 4-unit homeowner properties by insurance status of the first mortgage. Total outstanding debt is distributed by selected mortgage, owner, and property characteristics for mortgaged l-unit homeowner properties. Table 5 shows the total number of 1- to 4-unit and l-unit mortgaged homeowner properties by type of holder of the first mortgage. The data on mortgaged l-unit homeowner properties are distributed by mortgage, property, and owner characteristics. Tables 6, 7, and 8, presented only for the United States as a whole, repeat table 5 for properties with mortgages insured by Federal Housing Administration, for those with mortgages guaranteed by Veterans Administration, and for those with conventional mortgages, respectively. Part 2, Renter and Vacant Properties, is expected to present data, similar to those in part 1, on the financing and characteristics of rental and vacant properties in the United States. Special runs of the magnetic tape or the punchcards may be made for users on a cost-of-preparation basis by arrangement with the Chief of the Housing Division of the Bureau. |