Closed UteH closed 7 years ago
Karsten pointed out that we have to define what kind of relations can occur and what kind of relations these specific variables have
I also think that we have to find a more general definition of what should be flagged as related variable. In my view currently the following types occur in our data (feel free to complete the list :-) ):
The current DZHW-data has not the complexity level which would cause any problem when assigning these relations (apart from the harmonization over time series, which is currently not in our scope). But in the future the study designs will gain complexity and we will have to find ways to handle it (think about a 10-wave-multi-cohort-study). I think we should have this in mind when we think about data edition functions of the MDM.
(No Interface Update. An alternative interface will be added by an another issue)
All derived variables are contained in one dataset => Reuse the pattern of panelIdentifier e.g.: phd2014-ds1-bsch04c_g1 When displayed via a link we should order by indexInDataSet
I think we talked about this several times but I don't remember why we did not do this, so sorry in advance:
I think users will appreciate it if variables that only differ by suffix were conntected, e.g.: Someone looks at the variable questionnaire and is interested in "1. Qualifikation: Hochschulart" which is bstu09_a, thus not accessible. As users do not (yet) know how the variable suffix works it could be very helpful to display other variables that come from bstu09_a (and are accessible) which are:
Could these be "cross-listed" in all these variables in "related variables"?