[DDI-SRG] Inclusion of an existing CodeScheme into a hierarchical coding structure

[Wendy Thomas]

see below

On Thu, 24 Jan 2008, Joachim Wackerow wrote:

> I see a requirement of an inclusion of an existing coding scheme into a
> hierarchical coding structure. Use case:
>
> Building a total for all ISCO occupation codes for a ncube. ISCO has
> four nested levels in a hierarchical coding scheme. It doesn't include a
> code for the overall total. An overall total can make sense when using
> ISCO in a ncube. A new code for the total can be invented. All the ISCO
> codes must be nested below this top-level code. This is currently only
> possible by copying the whole ISCO coding scheme to this place. It isn't
> possible to include by reference an existing code scheme at the code
> level. This is currently only possible on the coding scheme level.
>
> The suggestion is to add a choice below code (code or a reference to a
> coding scheme) instead of just a code (like now).
>
> The whole structure of CodeSchemeReference below CodeScheme can be then
> used below Code in alternative to a nested Code.
>
> Example:
>
> <l:Code levelNumber="1" isDiscrete="false">
>   <l:CategoryReference>
>     <r:ID>ISCO_total</r:ID>
>   </l:CategoryReference>
>   <l:Value>10000</l:Value>
>   <CodeSchemeReference>
>     <r:ID>ISCO_CodingScheme</r:ID>
>   </CodeSchemeReference>
> </l:Code>
>
> The code with the value 10000 represents all nested codes, here all
> codes existing in the ISCO coding scheme. This is here the overall total.
>
> The CodeSchemeReference allows also to restrict the included coding
> scheme by levels or codes.

According to CodeSchemeReferenceType the full CodeScheme is referenced and 
there is no option for limiting inclusion. The option for using only parts 
of a coding scheme is found in Variable/CodeRepresentation which is 
extended by CodeSubsetInfo. This includes the IncludedLevel, 
IncludedCodeRerference, etc.

I think restrictions should only occur at the variable level not at the 
CodeScheme level. By importing only a section of a larger scheme and then 
editing it, I lose its relationships to the the larger scheme. However, I 
still think we need to be able to extend a code set by adding codes and 
adjusting level designations when needed. More on this in this evenings 
email.

Wendy


>
> When this would be possible the question raises, are the levels still
> appropriately numbered. The documentation on levelNumber of Level says:
> "Designated identifier for the level; generally increases as the level
> of indention for nesting increases". This is not really clear: is it
> just an identifier or is it a ranking number system?
>
> In the case of an identifier it would not cause a problem. In the case
> of a ranking number system the top-level code has to have a lower level
> number than the level below. Then it would be more flexible to invent a
> element which reassigns new numbers to existing coding schemes.
>
> Overall it seems to be easier just to change the documentation in terms
> of clarity that it is just an identifier and that - if using increasing
> numbers - this wouldn't express some actual hierarchy.
>
> Any comments?
>
> Achim
> _______________________________________________
> DDI-SRG mailing list
> DDI-SRG at icpsr.umich.edu
> http://www.icpsr.umich.edu/mailman/listinfo/ddi-srg
>

Wendy L. Thomas                          Phone: +1 612.624.4389
Data Access Core Director		 Fax:   +1 612.626.8375
Minnesota Population Center              Email: wlt at pop.umn.edu
University of Minnesota
50 Willey Hall
225 19th Avenue South
Minneapolis, MN 55455