Enquête Statistique Agricole 1992

A complete list of villages and relative population updated to 1984 was used as primary sampling unit frame. Exhaustive enumeration of holdings in each sampled village was done, and the complete list was used as secondary sampling unit frame. A two-stage sampling design was used; the primary sampling units, the villages, were drawn with unequal probabilities and repetitions permitted (each village was drawn with probability proportional to its population). In weighing the available resources with the required degree of precision, a sampling size of 10% was considered adequate; consequently, the number of villages sampled was 319 out of a total of 3191. A fixed number of five secondary sampling units, the traditional sector holdings, were randomly selected with equal probability and without repetitions from all sampled villages. All villages drawn twice had 10 holdings drawn or, better, all villages drawn n times, had 5n holdings drawn. The total number of holdings sampled in the country was 1595. This sample design was independently applied to the six departments of the country; in each department, the sub prefectures were considered as strata. The number of villages selected in each sub prefecture was proportional to the population of the same sub prefecture.