So the difference between calculated measures and calculated attributes is that one creates a measure, and one creates an attribute.
This is a critical difference because in a HANA model, you can only aggregate (SUM/MIN/MAX/COUNT) on measures. So if you need to consume your objects as an aggregated data type, you need to select MEASURE. It will then appear as a formula in the semantic layer.
If however you want to group by the object in your model then you need to use attribute.
Note that if you use a measure when you expect an attribute, you can get incorrect aggregations and therefore incorrect numbers.
If you use an attribute when you expected to use a measure, you can get an explosion in data volumes because you are not aggregating.
John