In this article
1. Explaining multi-level data
When a survey includes questions about people as well as questions about events those people have been involved in, the data from that survey can be presented in two levels. One level represents the respondents (people), and the other level represents the events (such as consumption occasions).
When data is presented this way, it is possible to look at all the relationships between the characteristics of the people (such as age, sex, or income) and the characteristics of their “events” (such as when a drink was consumed, what brand was it, or where was it consumed).
The data are stored as records which are connected to one another through links. (Unique identifiers) Learn more.
In this hierarchical database model, each child record has only one parent record, whereas each parent record can have multiple child records.
A multi-level database may have many more “event” records than “person” records because each person can, for example, have had many drinks.
2. How are counts different?
The unit of count is what each record in the data represents. Whether it be a respondent (person), or an occasion, etc. Data may be collected at multiple levels, for example, one of the files contains information about respondents and a separate file contains information about their consumption occasions.
At the people level, each record represents a respondent.
- 30% of people said Cool Nut is their favorite brand.
At the occasions level, each record represents a consumption occasion.
- 66% of people have consumed Cool Nut.
- Cool Nut was consumed in 43% of the consumption occasions.
- Across all occasions, 55% of the total volume consumed can be associated to Cool Nut.