Overview and Brief History - Sampling Spatial Units for Agricultural Surveys - page 38

specific type of crop or crop mixture is cultivated. The holder is a civil or juridical
person who makes the main decisions regarding the use of resources and manages
the agricultural holding. The holder has technical and economic responsibility for
the holding. They may directly undertake all the responsibilities, or delegate
responsibilities to a hired manager .The respondent of an agricultural survey is
the person (i.e., the holder or manager) who has provided data about the holding.
For further details about these definitions, the reader can refer to Chap. 5 .
The aim of an agricultural survey is to estimate several variables of interest for
the total survey area. The estimate of each variable is obtained through an inference
procedure using the values in all, or a sample, of the reporting units (typically the
holdings or land areas). The survey area may be the whole country, a province, or a
region.
The layout of this chapter is as follows. Section 2.2 is devoted to an overview of
using spatial units when sampling, and describes the advantages and disadvantages
of this sampling approach. Section 2.3 contains a review of the main agricultural
surveys based on spatial units. Finally, the last section concludes the chapter.
2.2 The Use of Spatial Units When Sampling Natural
and Environmental Resources
The importance of sampling spatial units has recently been acknowledged for
various practical problems in geographical and environmental studies. In most
cases, spatial units are defined across a geographical domain partitioned into a
number of predetermined regular, or irregular, shaped locations, i.e., pieces of land
territory. Sample information is used when a complete enumeration is too expen-
sive, with the usual operations for estimating the distinctive features of a given
population.
In standard sampling theory, spatial units have traditionally been represented as
a mosaic of areas in which individual primary units are considered to be identical
members of the same population (Haining 2003 ). In the first stage of a multi-stage
sample design, the sampling frame consists of a number of large aggregate units,
each of which contains sub-units. We will define a first-stage unit as a primary
sampling unit (PSU). The investigator selects a probability sample of PSUs, and
then proceeds to the second stage of sampling, in which a probability sample of
sub-units is selected from each PSU. These elements are defined as secondary
sampling units (SSUs, see Sect. 6.6 ).
Data relevant to each PSU (see the next paragraph for some examples) are
frequently assumed to be independent. Therefore, they are selected so that the
second-order inclusion probabilities are as close as possible to the product of the
first-order probabilities.
When population units are geographically distributed, classical random sam-
pling strategies may be inefficient. In fact, nearby locations tend to have more