A data cube displays data in two groupings: measures and dimensions.

• Measures: These are the data elements that can be counted. Things like a count of research awards, or the sum of budget expenditures.
• Dimensions: Some people refer to these as the "slicers" or "breakdowns". These are the data elements that allow you see a fact sliced by a particular dimension. For example, you can look at the count of research awards by Department using the Financial Organization Dimension.

However, it's very important to know that not all Measures can be sliced by all Dimensions. This makes sense conceptually. In the ResearchAdminData cube for instance:

• You can see a count of research proposals sliced by the proposal IDs because all proposals have proposal IDs.
• You cannot see a count of awards sliced by the proposal IDs in the proposals dimensions because not all awards have proposal IDs.

Conceptually, and in the data, the measures don't have a relationship to that dimension.When a user joins a measure to a dimension that doesn't have any underlying relationship, the cube will return erroneous results, like a table full of duplicated values. This doesn't mean the cube is broken, it just means those two elements can't be joined. Our partners in the Office of Research and Information Services (ORIS) put together this handy document to demonstrate this concept: RAD Cube Facts and Dimensions.

Note: The above document from ORIS uses the term "facts" instead of "measures". In this case, the terms "facts" and "measures" are interchangeable.