Reference Sheet V
As you engage and connect (Stage 1) and understand risk and resilience (Stage 2), you will discover some important differences in the community. As you do so, you will need to help the community to capture its different voices in a way that permits comparison. One way to do this is through your choice of sampling. A sample is a subset of a whole population, information from which can enable accurate conclusions to be drawn about the whole population. Sampling (the process of selecting a sample) is necessary whether data employ random (probability) samples for quantitative methods
(such as a survey) or purposeful (non‑random) samples for qualitative methods (such as interviews or focus groups). Sampling is a technique that makes it possible to identify a representative subset of a population when you cannot communicate with all its members. While there is no special formula for sampling for a resilience assessment, two common approaches are described below:
Purposive sampling (social groups) is when you knowingly determine and select groups from whom you need data. Purposive sampling involves participants who are selected with a specific purpose in mind, not randomly. One purpose that aligns perfectly with Red Cross Red Crescent values is diversity sampling. This technique aims to capture the widest relevant diversity in a given community, thereby ensuring that all voices are heard. The first step is to establish what diversity exists in the community. To do this, identify groups early in the data collection process that are less visible and more marginalized. These might include women, immigrants, youth, the elderly, people with disabilities, or ostracised groups (such as certain castes).
Square sampling (geography): One way to ensure you have a balanced sample is to ask the resilience team or community leaders to invite at least 20 people to the assessment exercise (10 women and 10 men). Ideally, you should take a map of the community and divide it into 10 squares of roughly the same size. Invite one man and one woman from each square. That way, you will avoid selecting only the well‑connected or friends of volunteers. After all, the sample should be as representative of the wider community as possible (see Resilience Dashboard).
If you are conducting a more formal quantitative survey, you may use stratified sampling to select participants at random from each of the strata or subgroups you wish to survey.
You can also break your data collection into groups by applying the same method or tool with identified subgroups separately. A seasonal calendar, for example, will look quite different if you first ask fishermen to describe their year and then ask female farmers. While it is not always necessary to repeat each session completely with each group, it is important to capture an appropriate range of voices, giving particular attention to those who are most vulnerable.
Sampling may sound demanding, but it is at least as important for qualitative as for quantitative methods. You need to develop a thoughtful sampling strategy (method and tool) for every data collection session you conduct. You will also need to be able to convince others that those who provide your primary data accurately represent the groups or perspectives you are interested in.
For more technical support see Project/Programme monitoring and evaluation (M&E) guide (IFRC 2011), pp. 36‑38, Annex 2 of which lists other useful resources. IFRC’s Rapid Mobile Phone‑based (RAMP) survey (2012) provides valuable guidance, both on using mobile phones to collect data and on practical sampling and surveys.