247
A sampling frame should be developed which allows for the collection of data at various
administrative, ecological, and geographical levels. This would facilitate the initiation of smaller, more
cost-effective survey designs to collect detailed data for specific needs.
With a accurate sampling frame in place, most future surveys should be based on probability
designs. Probability surveys allow the statistician to make statements about the estimates' precision.
These statements in turn allow the data users to calculate a band of confidence in using the data.
Decision-makers would gain confidence in both the MAFF and the CSO if the organizations moved
to more probability surveys and concentrated on collecting and summarizing fewer, higher-quality data
sets.
Scott suggests that some movement has begun within the CSO to combine the noncommercial
and commercial surveys. This is an excellent idea. When combined, the resulting survey should be
much more cost-effective and efficient than the existing separate surveys. However, as of April 1994,
separate questionnaires were still being administered to the two categories of farms. One small
improvement has been made. Instead of sending and receiving commercial questionnaires by post, the
forms are now being delivered by enumerators and later collected by them.
B. Improving efficiency
Survey efficiency is gained through the proper management of resources. All survey managers
should be well trained and should be vested with the authority to respond quickly to unexpected field
conditions—adjusting logistics and personnel accordingly. Extra salary incentives that encourage
fieldwork over central office work are detrimental to both the efficiency and the accuracy of the
surveys. The extra time spent in the field not only lessens the amount of time spent in cleaning and
summarizing the data, it allows more data to be collected, thus adding even more to the burden of
work to be done in the central office. Staff working on the agricultural data have not yet systematically
collected, entered, analyzed, and published data as one continuous and coherent process. The emphasis
on data collection has pulled resources away from analysis—the outcome being piles of unprocessed
data. Reducing data collection to the essentials will go a long way toward solving the central office
staffing problems, but incentive structures need to be modified to encourage office work in Lusaka
on analysis and publication.
C. Improving timeliness
The inability to publish timely reports has been most obvious in the CSO commercial farm
surveys. Many of the reports are published well after the year in which the data were collected. One
reason for delay is the imbalance between field and central office staff discussed earlier.
Reorganization may be necessary to assure that each survey unit has a data entry and computing staff
capable of handling the current data collection activities. The speed with which data are handled can
be greatly enhanced with proper data entry/computer editing programs. Modern microcomputer
processing is available to overcome many of the time and accuracy problems of out-of-date systems.
Data summary and analyses can also be made more timely through the proper addition of computer
techniques.
Late reports are also often the result of ill-defined priorities. If the data managers do not have
a clear sense of priorities, it is easy to allow outside requests for data to overcome what should be
ongoing data series. To maintain proper focus on the priorities, a clear mandate must be agreed upon