Survey Data Collection and Processing - Sampling Spatial Units for Agricultural Surveys - page 224

• Definition and exact location of the point provides a standard definition to be
applied for the basic size of the point. It corresponds to a circle with a 1.5 m
radius and specific rules are presented for ambiguous points.
• Land cover and land use provides further detailed instructions and examples of
classifications of land cover and land use.
• Water management information is provided.
• Transect gives information about how a transect should be mapped. It describes
the linear features, walking rules, accessibility, visibility and photo-
interpretation, superposition of linear features, some exceptions, and how to
record the transect information in the survey form.
• Quality of the work reminds the surveyor of the importance of quality in the
survey, and describes the elements of good quality fieldwork.
• A short overview of the valid codes for all entries is given.
• Annexes are provided. They include, for example, the land cover and land use
lists, a list of plant species, and the bio-geographical regions of Europe.
Appropriate test procedures for the registration should be based on repetitive
recordings of a sample of questionnaires, using comparisons between the first and
second recording to identify inconsistencies and errors. Indicators of the quality of
the recording can then be based on the ratio of wrong bytes to bytes checked. More
specific indicators can be calculated by referencing to particular variables or to each
single operator.
9.4 Data Editing
Data editing regards the detection and imputation of missing or inconsistent values
in the variables measured or observed in the survey. It is usually performed using
computerized procedures, making this activity fully automatic or computer-assisted
(Atkinson and House 2010 ). These values, which will be hereafter referred to as
errors, are those that violate a set of formal logic rules. These are called editing rules
and relate to set ranges for individual variables, to the relationships between vari-
ables, and to rules that have been established during the compilation of paper
questionnaires. Data editing procedures can be also interactive, meaning that only
the error detection is automatic. The corrections are manual or assisted, and are
made using a computer. As in manual revisions, the purpose is to make corrections
to the data that minimize the effect of errors detected during the subsequent
processing phases, and thus reduce the errors in the produced estimates.
A systematic description of this topic is very difficult, because it is very
extensive. For further details refer to De Waal et al. ( 2011 ).
In the remainder of this section, the term imputation is considered synonymous
to data correction. In this context, we emphasize the possible ways of performing
automatic data editing and the possible sources of error, which can introduce
additional biases into the data rather than reduce them. Finally, we give some