Rwanda - Rwanda Seasonal Agriculture Survey 2013

The sample was composed of two categories of respondents: Agricultural Operators and Large Scale Farmers (LSF). For the category of Agricultural Operators, the 2013 RSAS benefited from a dual frame sampling design (called Multiple Frame Survey MFS). For the category of LSF, everyone has been enumerated.

The 2013 RSAS used imagery from RNRA with a very high resolution of 25 centimeters. During the construction of the area sampling frame, the entire land area of Rwanda was subdivided into 10 non-overlapping land-use strata defined by proportion of cultivated land or other land-use characteristics. Three of these strata were chosen to be used for the survey since they were composed of agricultural land. Thereafter, agricultural land strata were delineated into Segments within the Primary Sampling Units (PSU) with identifiable physical boundaries. For agricultural operators, the segments were the Second Stage Sampling (SSU) units.
The 2013 RSAS covered 327 segments, spread throughout the country during the two main agricultural seasons (A and B) and 251 segments during the Season C both in mountains and marshlands areas.

The survey covered 15,441 Agricultural Operators and 562 LSF in season A; 15,730 Agricultural Operators and 503 LSF in Season B; and 1,412 Agricultural Operators in Season C. In Season C, LSF were not covered.
Each selected PSU having a size of 200 - 400 Hectares was subdivided into Second Stage Sampling Units (SSUs) of around 20 Hectares each, following natural boundaries. Note that for Stratum 3 PSUs, a segment had a size of around 50 Hectares.
In every selected PSU, one SSU (or Segment) was randomly selected for data collection purposes.

It is important to note that in each of agricultural season A and B, data collection was undertaken in two phases. Phase I was mainly used to collect data on demographic and social characteristics of interviewees, area under crops, crops planted, rainfall, livestock, etc. Phase II was mainly devoted to the collection of data on yield and production of crops.

Regarding the selection of respondents, from the list of LSF in Phase I of Season A or Season B, all 562 LSF were enumerated. The LSF were engaged in either Crop farming activities only or Livestock farming activities only or both Crop and Livestock farming activities. For Agricultural Operators(being the Small Scale Farmers within the Segment), every selected Segment was firstly screened using the Screening Form. That means enumerators accounted for every plot inside the segment. All Tracts were either Agricultural (cultivated land and fallow land) or Non-Agricultural Land (water, forests, roads, rocky and bare soils and buildings).

During Phase I, a complete enumeration of all farmers having agricultural land and operating within the selected Segment was undertaken by using a Farm Questionnaire. In Phase II, 25% of the Agricultural Operators undertaking either Crop farming activities only or both Crop and Livestock farming were selected and interviewed using a Farm Questionnaire for this phase.In Phase II, Season A, a sample of 1,799 Agricultural Operators was selected using the method of Probability Proportional to Size (PPS) in each district.

In Phase II, Season B, a sample of 1,941 Agricultural Operators was selected as follows:
a) 1,545 Agricultural Operators were selected of which 1500 were from Strata 1 and 2 selected at district level, and 45 Agricultural Operators were from Stratum 3 selected at country level (they were mainly from Nyagatare, Gatsibo and Kayonza districts). Again the PPS method was used, area under crops being the measure of size in each district; also
b) 500 Agricultural Operators were selected using area under crops in each district. Due to the previous selection explained in a) above only 396 were retained due to the removal of duplicates. This second sample gave weight to major crops and thus increased representativeness of crop yield in the districts.

In Season C, a Screening Form was used to undertake a complete enumeration and account for every plot inside the segment on which land use was taking place. From a list of Agricultural Operators having agricultural land and cultivating Season C Crops:
a) A 10% sample of operators in Marshlands was selected for data collection that combined farm inputs, expenditure and production questions;
b) A complete enumeration in Mountain sites was undertaken for data collection that combined farm inputs, expenditure and production questions.


The response rate for Seasonal Agriculture Survey is 98%.


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