representatives had pointed to the probability that many multiplant companies would have exceptional difficulty in providing this type of information, particularly on insurance and pension expenditures, for a single plant or an arbitrary combination of plants. Allowing such companies the option of providing information for all plants combined, it was believed, would facilitate response. In the instance of a selection of a single plant, this was accomplished by the following instruction in the box heading of the questionnaire:

In the case of multiplant companies, a report covering the plant identified on the accompanying letter would be preferred. However, multiplant companies may report for all plants if records are maintained on that basis. Whichever alternative is selected, it is essential that all of the information supplied relate to the same unit.

When more than one plant of a company was selected, a flyer was attached to the questionnaire with the following instruction:

A separate report for each plant is preferred.

If your supplementary remuneration programs and practices, as well as the types of records kept, are the same in each plant, and if you feel that a single report for one plant would be representative of your company, then such a report for one plant may be furnished. If expenditure records on supplementary employee remumeration items are kept on a companywide basis, and if you prefer to make one report covering all plants of the company, such a report will be acceptable.

The questionnaires were mailed in April 1954, which allowed ample time for companies to have completed their normal summarization or review of expenditures for the calendar year 1953.

In accordance with usual Bureau practice in mail wage surveys, a selected group of key companies were visited by Bureau representatives for the purpose of putting the questionnaire in the hands of the appropriate official, explaining the background and purpose of the study, and working out arrangements for single-plant or multiplant reporting.

One followup letter was sent to all nonrespondents. Preliminary editing of the returns necessitated a number of requests for clarification of data or for omitted data. Unfortunately, due to time and staff limitations, this could not be done in all instances of omissions or of apparent discrepancies, as the following analysis will show.

The findings of the survey are discussed in this report under three main headings: Characteristics of the Response; Establishment Records and Reporting Practices; and Factors Affecting Expenditure Levels.

Characteristics of the Response

The feasibility of a mail questionnaire survey of this type should be evaluated with respect to: (1) Rate of response, or the proportion of the companies solicited for cooperation which returned usable questionnaires; (2) balance, or the structure of the response as compared with the structure of the original sample, and (3) the reliability of the data as to the manner in which questions were interpreted, the ability of the respondents to provide the type of information requested, and the care exercised in computing or compiling the data. Because no fixed standards exist against which these factors can be measured in a study of this type, evaluations must necessarily be tentative.

Rate of Response

The survey yielded a response rate of 50 percent, or 550 returns out of 1, 105 solicited. The rate of response here is measured by the usable questionnaires received. A usable questionnaire was not necessarily a complete one. The letter accompanying the questionnaire, and the box heading on the questionnaire itself, urged companies to answer as many questions as possible and to indicate the reasons for their inability to answer other questions. Many companies thus returned partially completed questionnaires. These returns were classified as usable if the basic data on payrolls, manhours, and employment were supplied and if Section IV, dealing with records and expenditures for selected items, was answered at least in part. A number of questionnaires were eliminated because of multiple discrepancies or omissions in the data requested, which indicated inability on the part of the respondent to supply reliable information.

The rate of return by industry group (table 1) revealed relatively low response levels for some groups, notably those in which small establishments predominate, e.g., apparel manufacturers (21 percent) and lumber and wood products companies (29 percent). Variations among specific industries within industry groups were of course much wider than these data show. However, the significance of the differences in response rates among industries, insofar as an all-manufacturing survey is concerned, is centered in the totals for

6 To facilitate the analysis of response rates and the substantive data reported, the manufacturing industries covered were divided into three groups, in accordance with industry gross average hourly earnings data compiled by the Bureau for the year 1953. The groupings were established, arbitrarily, as follows: Group I-less than $1.65 an hour; Group II-$1.65 to $1.89; Group III-$1.90 and over. For convenience in discussion, the terms "low wage, dium wage, " and "high wage" groups will be used.

" "me

the three wage groupings into which the industries were classified." The average rate of return for the low-wage group amounted to 42 percent as against 54 and 55 percent, respectively, for the mediumand high-wage groups. The question of whether this differential return has an appreciable effect on average expenditure levels will be examined later.

On a regional basis, returns lower than the average were received from companies in the South, comprising the South Atlantic, East South Central, and West South Central States (table 2). On the whole, variations in response among regions were not large. However, a relatively low rate of response from companies with plants in more than one region was encountered. This appeared to be due to the difficulties of reporting in multiplant situations and (perhaps a related factor) to the unwillingness of several large multiplant companies to participate in the survey.

Although the low response rate (25 percent) for establishments employing fewer than 100 production workers, as shown in table 3, may reflect the inability of small companies to report information of the type requested, it conforms to the usual experience of agencies conducting mail questionnaire surveys. Significantly, however, the response from plants employing between 100 and 500 workers (50 percent) was close to the average for the larger establishments.

Multiplant Reporting

Multiplant companies responding to the questionnaire had the option of providing information for a single plant (separate reports if more than one plant was selected in the sample) or of providing data for a combination of plants or for all plants. Although it was not possible to identify all of the establishments in the sample which were part of a multiplant organization, it would appear from the returns that multiplant companies preferred combination reporting to providing data for selected individual plants. Of the 550 usable questionnaires, 96 covered more than 1 plant-37 covered 2 plants, 38 covered from 3 to 10 plants, 9 covered more than 10 plants, and 12 covered an unreported quantity. Only 16 multiunit companies furnished separate plant reports-13 provided 2 reports, and 3 provided 3 reports.

7

7 As a consequence of multiplant or companywide reporting in a substantial number of cases, the coverage of the survey can be described precisely neither in terms of establishments nor of companies. Strictly speaking, the 550 schedules represent 550 reporting units (whether company or establishment). However, since the

bulk of the returns represent establishments, that term is used in this report for sake of convenience.

Multiplant reporting was especially predominant in the chemicals and primary metals industries. In number of workers covered, the multiplant reports accounted for a disproportionate share of the total response. This factor among others precluded the presentation of data in terms of number of workers.

Other Characteristics of the Response

Requesting data for the calendar year appeared to present relatively minor difficulties.8 Only seven reports covered a fiscal period differing from the calendar year.

A few reports covered a group of workers other than production and related workers. Of these, the most common deviation was the coverage of all employees, followed by "hourly employees" only. These questionnaires, however, appeared to be internally consistent with respect to the information reported and were therefore included among the usable questionnaires.

The accuracy and reliability of the information supplied are difficult factors to assess, as will be stressed at various points in this report. This was undoubtedly a complex questionnaire to many companies; probably a large part of the sample had never before responded to such an inquiry concerning expenditures for supplementary remuneration. Many responses appeared to indicate that the cooperating company normally did not compile precise expenditure data for its own purposes. In a large sense, therefore, the survey was an experimental one, both to the Bureau and to the respondents, and must be evaluated in that light.9

The questionnaires were examined for inadvertent omissions and for obvious discrepancies. Omission of expenditure data for legally required payments was, of course, readily detected; inadvertent omission of other expenditure data practice was reported to be in effect. man-hour, and employment data, and in levels, were less apparent and could be error appeared to be large. In general,

became obvious when the Discrepancies in payroll, practices and expenditure recognized only when the overstatement was more

8 The questionnaire requested data for the calendar year 1953. However, if data had to be computed on another basis, respondents were asked to write in the period covered by the report. Reference to certain problems in calendar year reporting will be made later in this report in connection with legally required payments.

to

9. In letters accompanying questionnaires, some companies expressed concern about their own lack of exact data. One large multiplant company requested a number of blank forms in order conduct its own survey among its many plants. Other companies expressed an interest in appropriate accounting procedures.