[DDI-users] Using Groups to manage frequent questionnaire versions
Wendy Thomas
wlt at pop.umn.edu
Wed Nov 11 11:19:41 EST 2009
Dear Klaus,
This came up frequently in a recent Expert Workshop at Dagstuhl and will
be the subject of 2-3 of the Use Cases arising out of it. Basically there
are a number of approaches to grouping in terms of how to determine what
the primary tree structure will be. From the discussions at Dagstuhl there
are a few primary considerations:
1) What is the most stable information? Least stable information?
You will want those parts that remain the most stable near the top of the
tree and the least stable near the bottom. For example, in the integrated
ISSP example we had, the content of the modules (topical
questions/variables) were the primary grouping (first level of subgroups)
because this was the most stable content and primary approach for the
researchers who would be served. The next level was Country (geography),
then time (least stable).
2) What is the primary approach to the data? In the above example it was
topical module. However, in another example using the ISSP, the concern
was to manage the production process within a single country so that the
focus was on time rather than topic. Both approaches work, but each
facilitates a different usage.
3) In the case you mention below, if I am understanding it, you want to
capture the development process of a questionnaire as well as the final
questionnaire used in data collection. You could capture the development
process (change in content) either by grouping as you noted or by
versioning the content of your question schemes and control contstruct
schemes.
I think it would help to see an example of what you are trying to capture.
I think you are saying you have the following situation, but I'm not
clear.
Wave I: Start with Original content
Versioning during design process resulting in Wave I: content X
Wave II: Start with Wave I: content X content
Versioning during design process resulting in Wave II: content Y
Wave III: Start with Wave II: content Y content
Versioning during design process resulting in Wave III: content Z
Etc.
Is this correct? Do waves have different content (other than revisions due
to development) such as topical changes?
As an aside, please note that there is additional structure being
developed to capture the questionnaire development process which you
probably want to follow.
Wendy
On Wed, 11 Nov 2009, Klaus Pforr wrote:
> Hello,
>
> I'm beginning to plan to document a complex panel study
> (http://www.pairfam.uni-bremen.de/index.php?id=2&L=1) with DDI and have
> a question about how to use groups. It is certainly reasonable to
> organize the different waves as children of one higher level group,
> where the changes in each wave are overriders of a default version.
>
> In our project we have the problem of organizing many versions in the
> questionnaire design process of each wave. Is it reasonable to organize
> these versions also with groups?
> I think about placing the things that are generally agreed on as group
> level info and the version changes at the child-level.
>
> Best regards
>
> Klaus Pforr
> --
> __________________________________
>
> Klaus Pforr
> MZES AB - A
> Universität Mannheim
> D - 68131 Mannheim
> Tel: +49-621-181 2801 (nachmittags)
> fax: +49-621-181 2803
> URL: http://www.mzes.uni-mannheim.de
>
> Besucheranschrift: A5, Raum A312
> __________________________________
> _______________________________________________
> DDI-users mailing list
> DDI-users at icpsr.umich.edu
> http://www.icpsr.umich.edu/mailman/listinfo/ddi-users
>
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
More information about the DDI-users
mailing list