[DDI-users] Re:Adding tags to the DDI (Andrew L Arnott)

[I-Lin Kuo]

Just a comment for the DDI instrumentation sub-committee.

There are two approaches to marking up this kind of information:

1) The existing DDI approach that Sanda is suggesting places the markup within
the <var> element itself.
2) Andrew's approach which places the markup outside the <var> element.

>From a technical point of view, I think the first kind of approach is inferior
-- it clutters up the variable with tags that (I feel) don't really belong in
<var>. In the case where multiple <var>s share the same question text or pre
question text or instructions or conditional branching, each variable must be
marked up with the identical redundant information. Andrew's approach of
collecting all the variables into a <vargrp> and then marking up the vargrp is
my preferred approach, but this has to be done more systematically.

Quoting Sanda Ionescu <sandai at icpsr.umich.edu>:

> Correct me if I'm wrong, but I think that Andrew is basically asking for a 
> way to mark up in DDI skip patterns (if response to Vx is code "1" then go 
> to Vxx; or, Vxx was aked of people who answered "1" to Vx).
> 
> I have repeatedly asked for a modification to the DDI which would allow an 
> accurate rendition of these patterns.
> 
> Right now, there are two ways in which we can mark-up such patterns: 1) by 
> using the <forward> and <backward> elements within <qstn>; technically, 
> these are meant to back- or forward-reference other questions, but they can 
> contain PCDATA, so an explanatory text can be inserted too (I am personally 
> not too fond of this option).
> 2) by using the internal link element (<Link>)  one can actually 
> cross-reference between the value "1" in variable Vx (category level) and 
> the entire variable Vxx (variable level). Category, and, respectively, 
> variable IDs would be used  in the Link's "refs" attribute.
> 
> I hope this helps.
> Sanda.