STUDIES OF RECORDKEEPING PRACTICES It has long been recognized that the collection of statistical information is limited by the content of the records kept by the reporting concern, some of which may be maintained solely to enable the concern to comply with a law or regulation, such as those relating to income taxes. Of greater interest, however, are the records which concerns maintain for their own information, since it is these records from which most of the statistical information about their activities must come. The Bureau's studies of the recordkeeping practices of manufacturing concerns have taken many forms. Many of its discussions about present or prospective statistical reporting problems are concerned directly with the records regularly maintained by the reporting companies. Questionnaires sometimes include questions about the type of records used as a source of the reported information. For example, the 1954 Census Form MC-D9a, "Water Use," which obtained supplemental information from plants using 20 million gallons or more of water, asked the respondent whether the answers to four sets of questions (on water intake, water recirculation, water treated, and "other inquiries") were based primarily on meter records, on engineering estimates, or on other sources. In addition, more systematic studies of recordkeeping practices are undertaken from time to time. Two such systematic studies are described later in this section. Recordkeeping practices vary not only from company to company and from time to time, but also among departments, divisions, or other administrative and operating sections within a company. The number of sets of records and the possbility of variation among them, of course, increase with company size, although even in a small plant an engineer, a superintendent, or some other supervisory worker may maintain informal records. Often only a small fraction of the detail systematically recorded at the many operating levels ever reaches the "front office," even though the main outlines of these records have been prescribed by "top management." Since in all but the smallest concerns records are the only source of information about what is going on, they are an important managerial tool. Recordkeeping is intimately bound up with lines of authority, delegations of responsibility, and other attributes of a concern's internal organization. Consequently, a concern can have internal reporting problems. Just what a company can report depends, in part at least, upon the level of its records that it is willing and able to tap. A large concern that is unwilling to go beyond the weekly or monthly summaries circulated among a small number of executives usually can report only limited detail about the operations of its various plants. The concern's willingness to dip deeply into more basic records is affected by its evaluation of the worthwhileness of the survey and its estimate of the costs of compiling answers to survey questions. The cost estimates, in turn, are affected by the interest of the operating departments in the survey results and by other factors affecting their willingness to supply "front office" personnel with information for this purpose. Two efforts of the Bureau to find out more about the recordkeeping practices of manufacturing concerns and how these practices affect the quantity and quality of available statistical information deserve further attention. Survey of components consumption A survey conducted for the Munitions Board on the consumption of a detailed list of components also obtained valuable data on manufacturers' records and on their methods of estimating certain figures reported in the census. This survey was confined to the metalworking industries, and the amount of detail required was much greater than is usual in a census. The subsample of plants to which this questionnaire was sent was drawn from the annual-survey sample for 1949, so that some check against the 1949 Annual Survey returns was possible. Important among the findings of this survey was that there had been some misunderstanding and consequently erroneous reporting of cost of materials, supplies, containers, etc. In some cases capital account items had been included in this figure; in some other cases, taxes, labor, depreciation, and other "burden" items were included. Overestimation of the cost of materials was easier to detect than was underestimation. Nevertheless, it was found that purchases for resale and the cost of contract work were sometimes omitted. The effect on the aggregates was negligible. Field interviews have been used in a similar vein to test the content of data on the value of shipments. Because of the small number of firms interviewed the results are not significant statistically, yet they uncovered some tendency for value of shipments to include the contribution of the sales offices or sales branches of multiunit concerns and, in a few instances, delivery costs. Both items were explicitly mentioned in the instructions for the 1954 Census of Manufactures. 1958 Census preparatory survey As a part of the planning and preparation of the 1958 Census a series of interviews concerning recordkeeping practices was arranged with executives of about 60 of the larger manufacturing companies the summer and fall of 1957. In these interviews the Bureau attempted to obtain more precise information on the usual recordkeeping practices of manufacturing companies, the composition of the figures that had been reported on census questionnaires, and the general availability of specific figures in the records of manufacturing companies. Although information of this type is essential to the long-range objective of the Bureau, the more immediate concern was with the content of the 1958 Census questionnaires and reporting instructions. The interviews covered the recordkeeping and reporting practices for inventories, materials consumption, expenditures for selected items, the valuation of capital assets and of shipments, distribution of shipments by class of customer, and company organization of recordkeeping activities. As a guide to the interviews a booklet was prepared containing many of the questions on which information was desired and providing space for recording the results of the interview." The questions asked in these interviews often were quite detailed. Regarding inventory records, for example, an effort was made to find out whether the company being interviewed maintained separate records in terms of quantity and/or value for the inventories of production materials, fuels, containers, other supplies, work in process, and finished goods. For each of these categories, questions also were asked about the frequency with which the company records were summarized; whether the summaries were made for departments, plants, or other subdivisions of the company; how often physical inventories were taken; how the inventories were valued; whether the methods used for valuing inventories were the same throughout the company or whether they differed by category or by operating division; the "definitions" of the various inventory categories used within the company; how products shipped to sales branches or to other plants for further processing are handled in the company's records; and how these shipments have been reported on censuses and annual surveys. For a few selected materials, parts, and components a detailed record of how they appeared in company records often was obtained, and the problem of obtaining material inputs by product class, rather than by plant, was a subject of discussion. 8 The information obtained in these interviews is reflected in the wording of the questions and the instructions for the 1958 Census of Manufactures. The "Supplemental Inquiries for 1957" questionnaire also shows some of the results of these interviews. Through this survey the study of recordkeeping and reporting practices is extended for maintenance and repair expenditures to all of the concerns reporting on the Annual Survey. In addition to requesting the dollar amounts of total expenditures for maintenance and repair for 1957 and separately for "(1) salaries and wages paid to own employees for maintenance and repair activities," and "(2) other maintenance and repair costs (include purchased materials and supplies used in maintenance and repairs and maintenance and repair services purchased from other business establishments)," this questionnaire asked for "yes" or "no" answers to the following questions: Since approximately the same questions were to be asked of several companies, this booklet was considered a report form within the meaning of the Federal Reports Act of 1942. Consequently, it was assigned a "form number" and was approved by the Bureau of the Budget. Needless to say, the results of these interviews are accorded the same confidential treatment as is other information obtained for statistical purposes by the Bureau. 8 This survey was conducted as part of the 1958 Census of Manufactures, under the terms of Public Law 85-207 (August 1957), which provides the Census Bureau with clear and unambiguous authority to make preliminary and supplementary surveys in connection with each census and authorizes the use of sampling except for the determination of population for apportionment purposes. The sample for MC-D11, Supplementary Inquiries for 1957, is the sample selected for the Annual Surveys for 1955-58. A. In your report for 1957 on Form MA-100 (Annual Survey of Manufactures) did you include in Item 6A (Cost of materials, parts, supplies, etc.) the materials and supplies purchased for use in maintenance and repairs? B. In your report for 1957 in Item 6 of Form MA-100: (1) Did you include the cost of maintenance and repair services purchased from other business establishments in Item 6A (Cost of materials, parts, etc.)? (2) Did you include the cost of maintenance and repair services purchased from other business establishments in Item 6E (Cost of contract work done for you by others on your materials) ? The results of these interviews also are evident in the approach to the distribution of manufacturers' sales by class of customer adopted for the 1958 Census. For 1947 and 1954 no effort was made to collect data on this subject, partly because of the deficiencies in the 1939 data, which had been collected for establishments as a whole, and partly because a 1953-54 series of interviews with trade associations and individual companies in selected industries appeared to indicate that product-class reporting by plant, with few exceptions, was not feasible. The approach adopted for 1958 calls for a company report and requests multiunit concerns to provide separate estimates for broad product groups. The use of estimates has been stressed, and the instructions emphasize that accounting-type accuracy is not required. INSTRUCTIONS AND QUESTIONNAIRE DESIGN There is ample evidence that the design on the questionnaire affects the kind of responses obtained. The 1947 Census question 12a on the cost of “materials, parts, containers, and supplies (except fuels)," was so crowded that the cost figure was not reported on about 40 percent of the returns (fig. 2). Apparently the question was simply overlooked by many of those who were preparing the return. The failure to obtain separate figures for clerical and routine officeworkers has already been mentioned. The 1947 return placed the "definition" of the various employee categories in a box over the entire inquiry (fig. 1, above). A pretest had shown that these figures could be reported, but in the pretest the definitions were given on the line on which the figures were to be reported. Basically, many elements of questionnaire design belong more in the realm of art rather than science. It is obvious that good design requires ample space for the entry of the reported figures, that the form should be suitable for typewriter use, that the questions should form a logical sequence, and that all of the entries should be clearly and unambiguously labeled. Yet, all such statements are obvious only if each of us supplies our own definition of what is ample, of what is a logical sequence, etc. The design of a reporting questionnaire requires minute consideration of |