How to select households in a cluster
All households in the selected cluster must be identified, and each should have an equal probability of being selected. To ensure this, the listing teams should follow the instructions and complete the household listing form (electronic or paper-based) provided in the Generic household listing form online tool. In some situations, only households with a specific population group (for example children 6–23 months of age) will be selected for the survey, in which case information to identify these households needs to be included in the household listing. Only those households will be included on the list used for the selection procedures described below.
As mentioned in Module 6: Selecting clusters, most cluster surveys sample the same number of households in each cluster. This is generally conducted using systematic random sampling of the predetermined number of households from an ordered household list to obtain wide geographic distribution.
If the cluster mapping and household listing have been conducted in advance, then household selection needs to be done centrally once the listing forms are completed. If it has not been done in advance, the same process will be conducted in the field using the listing developed immediately prior to the survey sample collection.
Step 1: In the final column of the Household listing form, “Survey HH number,” start with the number 1 and assign a sequential number to each household listed that meets one of the three following criteria:
- Occupied residential dwellings
- Households that refused to cooperate during household listing; or
- Households whose occupants were temporarily absent during household listing
Leave the cell blank if the dwelling unit is not occupied or the structure is not a residential structure. For each cluster, the number assigned to the last household on the list corresponds to the total number of households for that cluster.
Step 2: Record this total number of households in the MICS template for systematic random selection of HHs1 (“MICS Household Selection Template”). If the cluster was selected after segmentation, record the proportion that the selected segment represents in the PSU/EA in the “Proportion of the selected segment” column. See the Mapping houshold listing and segmentation online tool for this information. If no segmentation was carried out, leave the value of 1 in the column “Proportion of the selected segment”.
Step 3: The MICS template for systematic random selection of HHs will automatically generate the numbers for those households to be interviewed in the survey for each cluster. The selected households should be indicated on the Household listing online form by circling the corresponding number in the “Survey HH number” column. If household selection is done in the field and if it is culturally acceptable, where possible, mark the number on the door frame of the structures selected, using a marker or chalk.
Other approaches are sometimes used to identify survey households, such as having a random starting household and then selecting households in a specified direction, or using the “next nearest household,” as is frequently done in Expanded Program on Immunization (EPI) surveys. This is not recommended for micronutrient surveys because it results in a less-dispersed random selection and increases the design effect (specifically, it decreases the likely diversity of selected households).