[DDI-SRG] EduDDI in Germany

Tue Feb 17 10:55:10 EST 2009

See below

Wendy Thomas wrote:
> A few quick comments aside from this is REALLY FANTASTIC news!
> 
> See below...wlt
> 
> On Tue, 17 Feb 2009, Joachim Wackerow wrote:
> 
>> Yesterday I was at the second meeting of EduDDI and some good news:
>>
>> EduDDI was formed last December to build a framework for using DDI3 at
>> several educational research institutes in Germany. The driving people
>> are Michel Knigge und Martin Mechtel from the IQB in Berlin (Institute
>> for Educational Progress), who attended the workshop last autumn in
>> Dagstuhl. One other important institute is the DIPF (German Institute
>> for International Educational Research). One goal is to build resource
>> packages specific for the needs of educational research.
>>
>> It was mentioned that NEPS is planning to use DDI3. The National
>> Educational Panel Study (NEPS) is a new major panel study in Germany.
>> http://www.uni-bamberg.de/en/neps/
>>
>> A student is working in the Diplomarbeit (like master thesis) on the
>> subject standards/DDI.
>>
>>
>> Some questions/issues were raised at the meeting:
>>
>>
>> - Test scales
>>
>> The data collection module does not cover enough the needs for test
>> scales. Example: tasks of the PISA study. IQB would be interested to
>> discuss this requirement after some further experiences with DDI.
>>
> We do have a bug already listed regarding dealing with anchored scales. 
> We should certainly discuss the types of scales they need to be able to 
> express prior to addressing this bug.
I had the impression that they would like a whole specific structure for 
that. But I cannot tell details currently. I asked Michel for some basic 
papers so I can understand their needs.

> 
>>
>> - Controlled vocabularies/Genericode
>>
>> They are interested to use Genericode lists in an extensive way. Not all
>> places in DDI3 - they are interested - in support this possibility. In
>> generalizing this approach: I think it would be a good idea to support
>> the mechanism of adding a Genericode list to every machine-actionable
>> string value of elements/attributes. This way specific allowed values
>> can be defined in a flexible way dependent from the
>> institution/domain/application. As already discussed and in development:
>> the DDI Alliance can publish Genericode lists (controlled vocabularies)
>> for some elements where a general, domain-independent CV can make sense.
>>
>> This would cause a change on every element/attribute which is defined as
>> xs:string and probably additionally elements/attributes which are
>> defined as InternationalStringType, TypedStringType, and 
>> OneCharStringType.
>>
>>
> I think we should be somewhat careful on this as it may not always make 
> sense to have a controlled vocabulary option for all of these string 
> classes. We should investigate further.
Yes, this needs some investigation.

I have the impression that the currently sparse use of Genericode should 
be improved for a lot of string classes, even when a general use CV does 
not make sense. I think it would be good practice to have in reusable a 
string type which can be just a string or alternatively an externally 
defined genericode list. This string type can then be used by many 
machine-actionable string elements.

Especially in terms of application development would that be a help. The 
the allowed values are not hidden in the application but are in a 
external and reusable Genericode list.
Similar thinks we discussed long time ago.

> 
> 
>> - Universe, machine-actionable value
>>
>> They asked also for a way to use values - which are mentioned in
>> CodeSchemes/CategorySchemes - in Code of Universe.
>> Example: a survey on all German Bundeslaender has a variable Bundesland
>> with a related CodeScheme/CategoryScheme. In further surveys specific to
>> only one Bundesland the Universe should be defined (and controlled)
>> according to the CodeScheme/CategoryScheme of the other survey.
>>
> Universe when geographic can be associated with one or more codeschemes 
> using GeographicStructure and GoegraphicLocation. Did this go beyond 
> geographic related universes?
Yes, I had the impression. The Bundesland was just an example. I'll ask.

It is not clear to me how an geographic universe can be associated with 
a CodeScheme?

> 
> 
> 
>> Universe in conceptual component has a child MachineReadable ("A
>> machine-readable statement of the universe, expressed as a Coding."),
>> which is defined as r:CodeType ("Structures the expression of a coding
>> in the data collection process."). This element is used several times in
>> data collection, where it does make sense. There seems to be something
>> weird. The attribute programmingLanguage does not really make sense for
>> a universe.
>>
>> If MachineReadable would be defined as l:CodeType, then it would meed
>> the above mentioned requirement. Was this the original intention?
>>
> We are currently reviewing the use of CodeType and CommandType and this 
> should be wrapped in. l:CodeType was specifically defined to support 
> l:Code in l:CodeScheme and has nothing to do with relaying a machine 
> readable (actionable) means of relaying how a universe is constructed. 
> The intent here was to provide a location to define in a generic code 
> (e.g. MathML) or another language how a universe was defined. For 
> example the members of a parent universe that have certain 
> characteristics as defined by values of certain varaibles. At least that 
> was the original intent.
Their intention is to use a value from a controlled vocabulary which is 
related to a CodeScheme/CategoryScheme. Currently this approach could be 
only defined in an application. The question is, does it make sense to 
put this relationship into DDI?

> 
> 
>> Anyway, this issue seems to need some discussion.
>>
>>
>> - Universe of a variable, which is defined in a resource package.
>>
>> Reusable variables should stay in resource packages. When using this
>> variables from a VariableScheme by VariableReference it doesn't seems to
>> be possible to define a specific universe. This is only possible below
>> Variable in the resource packages, where it doesn't make sense, because
>> the variable should be reusable, i.e. reusable for different universes.
>> Currently a workaround with VariableGroup is possible. VariableGroup can
>> have a UniverseReference and can use VariableReference to point to the
>> Variable in the resource package.
>>
>> It seems that UniverseReference (now used in Variable) should also be
>> also usable in VariableReference and VariableGroupReference.
>>
> yep...definately a problem.
Actually this reminds me on some discussion with Dan Gillman in Montreal 
after presenting the ISCO resource package. He saw some fundamental 
issues to put a variable in a resource package (with ISO 11179 in mind). 
One question would be: does it make sense to put just the 
CodeScheme/CategoryScheme into a resource package?

> 
>>
>> Achim
>>
>>
>> -- 
>> GESIS - Leibniz Institute for the Social Sciences
>> Postal address: P.O. Box 122155, 68072 Mannheim, Germany
>> Visiting address: B2 1, 68159 Mannheim, Germany
>> Phone: +49 (0)621 1246 262
>> Fax: +49 (0)621 1246 100
>> E-mail: joachim.wackerow at gesis.org
>> www.gesis.org/en/institute/
>> _______________________________________________
>> 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

-- 
GESIS - Leibniz Institute for the Social Sciences
Postal address: P.O. Box 122155, 68072 Mannheim, Germany
Visiting address: B2 1, 68159 Mannheim, Germany
Phone: +49 (0)621 1246 262
Fax: +49 (0)621 1246 100
E-mail: joachim.wackerow at gesis.org
www.gesis.org/en/institute/