Abstract

The Census of Agriculture and Fisheries (CAF) is a major statistical activity of the government. It is a large scale undertaking by the National Statistics Office (NSO) geared towards the collection and generation of statistics on country’s agriculture and fisheries section. Moreover, the CAF is primarily considered to provide reliable sampling frames for various agricultural surveys. The availability of a quality agricultural sampling frame will contribute extremely to the improvement of current agricultural surveys. This study presented some of the ideas and activities that are being undertaken in the design of an appropriate sampling strategy for use in the 2002 CAF. Several options were presented, namely, MUN, SRS2, SRS4, SRS8, SYS4, SYS8. Among these sampling strategies, MUN performed quite well in generating provincial estimates. It must be noted, that this is conditional to the province (type of agricultural activity), and structural variable considered. However it is not generally satisfactory in generating municipal level estimates. The results also indicate that indeed large resources would be required if one wishes to generate precise municipal level estimates.

.