Abstract:
The paper begins, by discussing the problems that can arise in the design of samples for multi-purpose industrial and consumer surveys. A model of the data collection in such surveys is given in the format of a data hierarchy. This discussion is then applied to a specific panel survey, dealing with fertiliser usage and purchase in G.B. The paper examines the classes of 'minority' variables measured and how the estimation of these was affected by a change in sample design from quota to random. An account is given of how a map-based sample frame for agricultural holdings was created and the development of the stratification, with the aid of prior data. The disproportionate sampling by strata, as well as actual holding selection and projection method is described.