Rwanda - Rwanda Seasonal Agriculture Survey 2017

In order to provide the basis for conducting sample surveys based on complete coverage of the farm level, and as a better way of collecting agricultural data and finding better precise survey estimates, upgraded seasonal agricultural survey(USAS) used a Multiple-Frame Sampling (MFS) methodology by which, area frame was constructed and survey sample was drawn from it. Apart from that, a list frame of large-scale farmers (LSF), with at least 10 hectares of agricultural holdings, was done to complement the area frame just to cover crops mostly grown by large scale farmers and that cannot be easily covered in area frame.

For area frame design, the selection process was a two-stage process which begins with the first stage at the district level with every agriculture stratum (1.1, 2.0 and 3.0) divided into PSUs of 100 Ha (in strata1.1 and 2.0) and 500 ha (in stratum 3.0). The strata are broken into PSUs following visible boundaries and each PSU is given a unique PSU identity number. Based on the sample allocation to each district stratum the first-stage of the sample selection process uses a probability proportional to size (PPS) sampling method to select the number of PSUs equal to each stratum.

Then each of the selected PSUs is divided into secondary sampling units (SSUs) of approximately 10 ha for strata 1.1 and 2.0 and approximately 50 ha for stratum 3.0, and then randomly one SSU is selected from each sampled PSU and each is called a sample segment. A total number of 960 segments were selected countrywide using random sampling and the sample covers all 30 districts to ensure a representative sample at district level.

Finally, selected segment is subdivided into 1000 grid squares of 100 square meters each. To identify the plots/farms within the segment to be enumerated, a random sample of 5% grid squares was selected in each sample segment resulting in a nationwide point-sample total of 51,390 grid squares. A total number of 201 large scale farmers was identified and enumerated to complement area frame methodology.


Sampling weights were calculated for each stratum in each district considering the total number of segments in the stratum and the sample size in the specific stratum.