8.2.2 Checking the plausibility of the data
When? During the deployment phase of any MDC survey on the field and especially at the beginning, it is highly recommended to check the plausibility of the collected data on a regular basis (i.e. at the end of each day of the survey).
Why? Making a regular checking of the collected data can allow to identify errors or bias in order to correct them before the end of the survey. This step is therefore complementary to the Data Cleaning that is often necessary at the end of the survey.
Among other, checking the plausibility of the data allows:
- To identify potential errors among the collected data
- To ensure that all enumerators understood the different questions and answers correctly and rectify the situation for the rest of the survey if required
The ability to assess the plausibility of the collected data greatly depends on the knowledge of the field and project specificities. However a few features can systematically be checked:
- GPS coordinates: if collected within the survey, using the KoBo maping feature allow to ensures that forms have been completed in the right locations
- Duration of form filling: checking the time spent on a form (difference between start and end times) may sometimes provide information about surprisingly short or long forms (not necessary means that something went wrong)
- Weird answers: If it appears clear that some answers look obviously weird in comparison to other or that a given enumerator always provides another answer that all the others, it may be useful to ensure that the question and possible answers were well understood.
The ability to assess the plausibility of the collected data greatly depends on the knowledge of the field and project specificities. The more the person in charge of the checking has a good understanding of the project and context, the more likely he/she will be able to detect illogical, impossible or strange answers.
For example, the plausibility of the collected data can easily be checked online in real-time on the KoBoToolbox account under the DATA section of the deployed project and then by selecting Table.