Multi-Tier Framework Energy Survey, 2022

Sample design and selection

The energy survey included a household survey, a public institution survey, and a survey of refugee camps and adjacent host communities. For each survey, different sampling frames were used to select representative samples.

(1) Household Survey
The sampling frame for the household survey was the complete list of residential households from the Rwanda Population and Housing Census Report 2012 (RPHC2012) provided by the National Institute of Statistics. It excluded households located within 5km from the refugee camp, and it comprised information on Rwandan residential households grouped by Province, District, Sector, Cell, and Village.

The sample size for the household survey was designed to obtain estimates with high precision for the main indicators targeted by the survey. Since the Rwanda Demographic and Health Survey (2019/2020) indicated that in Rwanda, 46% of households have access to electricity, the value was used in the calculation of the sample size and considered as the benchmark indicator value. The minimum sample size required for the household survey was calculated using the formula provided in the document Rwanda Energy Survey (2022) Sampling Strategy and Weighting provided under Resources.

Stratification for the household survey was done considering province, urban, and rural area as strata to increase the efficiency of the sample design. An equal sample allocation was applied to allocate households between urban and rural areas and later a square root allocation was applied to allocate the sample size between provinces

Sample selection: In the first stage, 223 villages were selected using probability proportional to size (PPS), and in the second stage, after listing all residential households in the sampled villages, 18 households were selected using systematic random sampling methods (it was anticipated that 50% of electrified households and 50% of non-electrified households be sampled from each sampled village for interview).

(2) Public Institutions Survey
For the public institutions survey, given that the sampling frame is not available, all education and health centers located in the area where the household survey took place were interviewed. The sample size for the public institutions survey depended on the number of institutions found in villages selected for the household survey at the time of survey administration.

The sample size for the public institutions’ survey corresponded to all educational institutions and health centers found in the sampled villages. In other words, the final sample size for the public institution survey was determined based on the number of villages selected for the household survey.

(3) Refugee Households and Host Community Surveys
The refugee household survey was conducted in the five refugee camps in Rwanda located in different areas. The sampling frame for this survey was the exhaustive list of households in the five refugee camps. Additionally, adjacent host communities, within 5km of the refugee camps, were surveyed.

Sample Size: The sample size for the refugee camp and hosting communities was determined using the same formula used to calculate the minimum sample size for the household survey. The only difference is that since the indicator of interest (P) is unknown (proportion of households connected to the national grid), the prevalence rate was assumed to be 50% to maximize sample size and relative standard error to 3.6%. Therefore, the minimum sample size was 1,700 households. The total population of refugee camps and host communities was 264,316 of which the population of refugee camps represented 10% and the population of hosting communities represents 90%. To design an effective sample size for each category, 40% of the sample was drawn from the refugee camp and 60% from hosting communities. In other words, 700 households were selected for interview in refugee camps and 1,000 households in hosting communities; this implies a relative standard error (RSE) of 5.7% and 4.7% respectively.

Sample Selection: For the refugee household survey, in the first stage, 39 segments from refugee camps were selected using probability proportion to size (PPS), and in the second stage, a systematic random sampling technique was applied to select 18 refugee households from the sampled segment. For host communities survey, 56 villages were selected using probability proportional to size (PPS) and at the second stage, 18 households from each sampled village were selected using systematic random sampling technics.

For more information, see the document Rwanda Energy Survey (2022) Sampling Strategy and Weighting provided under Resources.