A Report on the General Enumerative Surveys-I

Statisticians are constantly trying to make their methods more precise, not only because of their scientific interest but also to meet the ever-increasing load which modern society throws upon statistical measurement. A notable contribution has been made over the last dozen years by means of the adaptation of probability sampling, along with the interview-type schedule, to agriculture and the development of the so-called master sample materials through the cooperative efforts of the Bureau of Agricultural Economics, the Bureau of the Census, and the Statistical Laboratory of Iowa State College. Ironically enough, the Bureau of Agricultural Economics has not been able to make very wide use of these new methods since the added precision means added cost. We were able, however, to arrange for two general Nation-wide enumerative surveys, both covering items which could not readily be obtained by mailed inquiry. The field work for the first of these was carried forward in January 1947 and for the second in April 1948. The scheduled items were arrived at through Bureau-wide discussion, and the surveys were jointly financed from economic research and agricultural estimates funds, with costs divided between the two on the basis of the number and character of questions asked. In addition, the second survey carried a section on marketing channels and transportation methods financed from an allotment under the Research and Marketing Act. An article based on these data is carried in this issue of the journal. Questions are raised from time to time, and properly so, as to what was done and the results obtained. This is the first of two articles designed to answer such questions. That is, the Division of Special Farm Statistics has been asked for a report written in the form of two related articles, the first describing and to some extent evaluating the mechanics of developing and carrying forward the surveys and the second dealing with the data themselves, how they are being used or where they are published. It should be understood that these articles are not designed either as an outline as to how to carry forward an enumerative survey or as a discussion of the merits or demerits of the enumerative versus other statistical methods. Some light should be thrown on both of these questions, however, by an actual examination of how the surveys were run, of their costs, and the results obtained. In conclusion, attention is called to one final factor: No attempt is made to evaluate the experience gained, especially by those actually in charge of survey operations, in the States and in Washington. But certainly, the Bureau statisticians are now much better qualified to do enumerative survey work and to discuss plans for the future, including the task immediately ahead of advising on the Census of Agriculture for 1950.-0. V. Wells