When you receive data either as part of a data scramble or in the field, all aspects of the acronym CUSTARD should be questioned and recorded as some form of metadata, so that it can be referred to later on. CUSTARD stands for:
- Coordinate system
- Units
- Source
- Triangulate
- Absent values
- Restrictions
- Date
Coordinates
If coordinates are provided, find out what form they take. Are they decimal degrees (DD, e.g. -1.283333, 36.816667, degrees minutes seconds (DMS, e.g. 1° 17′ 0″ S, 36° 49′ 0″ E), or something else?
Units
What is the unit of measurement? For example, is it individuals, families, buckets, tonnes? It is also important to find out how big the unit is. In the case of families – what is the average size of a local family, for buckets – how big is the bucket, e.g. 2 litres, 5 litres, 10 litres etc?
Source
Where has the data come from, who collected the data and how was it collected, e.g. a census or assessment? These questions are important to ask as they may give an indication of the quality and validity of the data. You may also wish to get more data or ask further questions from the source at a later date. It is important to record the contact details of the source so that you can get back to them if necessary.
Triangulate
Look at the data and ask ‘does it look right?’. Compare it with other data you may have and see if it makes sense. If it doesn’t, then ask the source questions about it.
Absent values
Look for any absent values and check what an ‘empty cell’ means. Is it meant to be empty, or is the data missing?
Restrictions
This information will primarily come from the source, but it is important to understand the restrictions of the data. There may be licensing issues that need to be resolved or limitations based on a specific licence. It could be that the user has permission to use and create products from the data, but they can’t share it with other organisations. Commercial or protection issues may restrict the use and sharing of data.
Date
Find out when the data was collected or created. This will give an indication of how old it is and its currency. Population data, for example, is often based on a census that may happen every 10 years, so if you receive census data that is eight years old it’s likely the population has changed significantly in that time. At any acute phase of an emergency response situational data, such as the affected population, could be changing many times a day, and having a date and time stamp is important to report.