Integrated Household Living Conditions Survey 2005

Data Collection Methodology in Urban and Rural Areas
The data collection procedures for EICV 2005 will be similar to those for the previous survey, described in Dr. Scott’s report. This methodology was effective for EICV 2000-2001 and will improve the comparability between the results of the two surveys.

Each sample urban ZD will be enumerated by one enumerator during a cycle of 33 days. The sample urban ZDs will be divided into 10 cycles throughout the year. The 9 sample households in each ZD will be divided into 3 groups of 3 households each. The enumerator will visit one group of 3 sample households each day, the next group of 3 households will be visited the following day, etc. This sequence will be repeated so that each household is interviewed every third day. This 33-day cycle in the urban ZDs will ensure that the payday is covered for salaried employees.

In the case of the rural stratum, a team of four enumerators and supervisors will be assigned to each province. The year will be divided into 10 data collection cycles of 32 days each, and 4 sample rural ZDs will be enumerated within each province during one cycle . The cycles will be further subdivided into two periods of 16 days each for the data collection in each sample rural ZD. The sample rural ZDs will be paired each cycle so that a group of two enumerators can be assigned to each sample ZD. The 12 sample households in each rural ZD will be divided into two groups of 6 households each, corresponding to the workload of one rural enumerator during the 16 day period. Each enumerator will visit a group of 3 sample households every other day.

Distribution of Sample ZDs by Data Collection Cycle within the Year
In order for the EICV 2005 to represent seasonality in the household income and expenditure data, it is important to have a sample that is representative across space and time throughout the 12-month data collection period. Ideally, a nationally-representative sample of ZDs should be enumerated each cycle. Sometimes it may be necessary to adjust the set of sample ZDs to be enumerated each cycle to take into account logistical considerations, weather conditions, etc. In the case of EICV 2000-2001, the distribution of the sample segments over time appears to have been mostly determined by operational considerations, which affected the geographic representativeness of the sample each cycle. This can result in a potential bias when there are geographic differences in seasonality.

For EICV 2005 it is recommended to select a representative subsample of ZDs to be assigned to each cycle for data collection. Therefore the 620 sample ZDs can assigned cycle codes from 1 to 10 systematically. In this way a nationally-representative sample of 62 sample ZDs (10 in Kigali-Ville, 8 other urban and 44 rural) can be assigned to each cycle. Then adjustments can be made to these cycle assignments based on logistical considerations, weather conditions, etc. Given the greater dispersion of a nationally-representative subsample of ZDs each cycle, this may increase the cost of the field operations slightly, but this investment will improve the seasonal representativeness of the EICV 2005 sample. In the case of the other urban stratum, the sample ZDs in each town should be spread evenly across the four quarters of the survey year as much as possible.