[DDI-ADG] Aggregate Data Notes

[J Gager]

Here are my notes on the aggregate data discussion so far:
 
1.  There is a need to describe an attribute similar to the way that a
variable is defined.  An attribute is distinguished from a variable, in
that it only serves to describe an observation (or set of observations),
and is not something actually measured.  If the value of an attribute is
changed, the actaul phenomenon being measured is not changed.  For
instance if you change the attribute if you take our example of
population by region and age, there is the attribute of unit multiplier,
which in one case might be hunder thousands.  If you change this, to say
ten thousands, you are not changing what was actually measured.  If you
were to change gender, you are actually changing what was measured.
Some common attributes, such as unit of measure already exist in the
DDI, however, there is a need to define custom attributes that may not
already exist.  This additon will allow for that.
These attriubtes can be be coded (containing category values) or uncoded
(being string, numeric, etc.)  Note that in SDMX uncoded attributes can
specify max length, fix length, alpha, numeric, or alphanumeric - if we
want to align, we should consider something similar.
 
2.  In the nCube structure, there is a need to attach these attributes
to an nCube (similar to series level attributes in SDMX), to a dataItem
in the physical structure (that is to an actual observation - so if we
include a structure for transporting data, we also need to include this
ability here), or to an entire data file (similar to a data set level
attribute in SDMX).
 
3.  Need to include a way to transport actual data in the instance, as
opposed to the data file - probably similar to the SDMX Generic Data
structure - we will discuss this more on Tuesday.
 
Thanks,
 
J
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.icpsr.umich.edu/pipermail/ddi-adg/attachments/20050826/5476234a/attachment.html