228
B. Efficiency
Once decisions have been made concerning the data to be collected, the technical staff (data
managers, statisticians, data collectors, computer technicians, and the like) must collect and publish
the data in the most efficient manner possible. Selection of the appropriate survey and sample designs
should be based on the type of data needed. Probability designs in which a known portion of the
population is sampled should be used when inferential statements about the population are required.
Probability surveys allow the calculation of sampling errors and allow survey statisticians to make
scientifically valid statements of data precision. If general information is sufficient, a non-probability
design may be acceptable. To achieve efficiency in surveys, the technical staff must be properly
trained and correctly located. Adequate logistical support must be provided at the appropriate time and
place. These requirements are difficult to achieve and often test the administrative and personnel
capacities of the survey organization.
C. Timeliness
Much of the value of agricultural data depends on timing. Policymakers need access to current
information to enable them to make meaningful decisions. By their nature, information on crop yields,
livestock numbers, and market prices must be obtained during the season. It is impossible to determine
these numbers once the crop has been harvested or the livestock sold or moved. Country leaders must
be sensitive to the needs of producing timely data and support the logistical needs of the data collection
units. Unless the amount of data collected is restricted to only the essential numbers, the entire process
is slowed—causing losses to both efficiency and timeliness. To provide accurate data, it is essential
that data collection, data summarization, and data analysis teams work in concert. The important step
of analyzing the data with respect to current economic, climatic, and political conditions must also be
done quickly and accurately. This is achieved by making researchers and policy analysts a part of the
survey-planning team. It is only through the work of the data analysts that the compiled data become
usable information.
D. Accuracy
To state that data must be accurate to be usable in decision-making seems elementary. While
no policymaker would willingly accept inaccurate data as a basis for decisions, it is necessary to do
just that in too many developing countries. Leaders are forced to use available data, the accuracy of
which is seldom known and even less often committed to paper by data managers. Hence the data user
is left with the problem of determining whether the data are accurate. This can only be done by
understanding the source of the data series' sampling and non-sampling errors. Unfortunately, the data
manager's judgement of accuracy is usually clouded by a lack of information.
Before commenting on a data set's accuracy, a manager must have full knowledge of how the
data were collected. What were the practices employed by the field enumerators? Were mistakes or
oversights made? Were the mistakes unintentional or intentional? Inaccuracies caused by these kinds
of mistakes are called non-sampling errors. Non-sampling errors are understood and controlled by
thorough survey planning followed by intensive training and supervision of staff. While it is not
possible to completely eliminate non-sampling errors, a good data manager constantly works to
minimize them. Minimizing non-sampling errors is possible only if data managers are able to insist
upon high-quality performance from their employees. Survey leaders should strive to create an
atmosphere of "expected excellence." However, to do this successfully, a manager must have the