...
Data is more and more collected through mobile data collection systems, such as Kobo and ODK: this allows easier data management. In the Follow up on data collection / Mobile data collection folder on top of the page, you will find guidance on how to set up and use mobile data collection systems.
...
- Unit of measurement, which is the level the data is collected at (e.g. individual, household, institution/infrastructure, community, area). This will have an influence on the type of data collected: the higher the level, the less reliable is the answer of the interviewee.
- Data-collection methods (e.g. direct observation, key informant interviews, focus group discussions, community discussions, key-informant interviews, household interviews, etc.): in the Design the methodology folder above can be found a table detailing pros and cons of the different data collection methods.
- Sampling methods, or in other words the criteria you will use to select the respondents. There are two main types of sampling: probability sampling – in which respondents are selected randomly and every person in the sampling frame has the same chance of being selected, and non-probability sampling – in which respondents are not selected randomly. Probability samplings are much more resource-intensive but can generate statistically significant findings, while non-probability samplings are often lighter in terms of resources but generate findings that are indicative only. For this reason, there is always a tradeoff between representativeness of findings and cost/time constraints.
...
Ensuring close follow up during the collection phase will improve the quality and timeliness of data. It is key that progress and challenges of data collection is regularly monitored. To achieve this, a matrix can be set up to track the number of forms that have been submitted, the areas that have been completed and the issues hampering progress. You need to check and clean data as soon as they come through to spot inconsistencies and follow up with the enumerators.
In the Follow up on data collection folder above you will find templates of tracking matrices and data cleaning tools.
...
Analysis should aim not only at describing the situation (for instance, where and who lacks safe water), but also at explaining the causes (for instance, lack of improved water points), interpreting the effects (for instance, linking presence of AWD with lack of safe water) and anticipating possible evolutions (for instance, the potential increase of child mortality rate in certain areas). Another key aspect is the implementation of cross-sectorial analysis based of WASH data or data from other relevant sectors, such as nutrition, health, education, etc.
In the Analyze the data folder above can be find documents that describe possible approaches towards these different levels of analysis.
...
Findings should be disseminated in a timely and effective way. Different types of information products can be considered, including factsheets, maps, web-platforms, reports, etc. depending on the audience and the resources available. In the Share information folder can be found templates as well as example of information products from past assessments.
Information products should be shared both with the primary audience through the coordination platforms channels (coordination meetings, MailChimp, SendinBlue, social media, etc.), and the broader humanitarian community, thought platforms such as HumanitarianResponse.info, ReliefWeb (see key external weblinks at the bottom of this page), etc. It is important to share the anonymized, cleaned dataset on the Humanitarian Data Exchange (HDX, see key external weblinks) – the main humanitarian online data sharing platform, so that other people can have access to data and run their own analysis.
...