[DDI-users] RE: Adding tags to the DDI

Andrew L Arnott Relate_Webmaster at byu.edu
Tue Jan 20 11:36:32 EST 2004


Yes, thank you Richard.  That is exactly what I am looking for.  Although
the simple skip pattern support currently in the DDI is not sufficient.  My
ideas?  I'll include a few. Below them I will include how I am working
around the problem for now (for the interest of others trying to do what I
am doing also in the meantime).  I am pleased to know that the DDI committee
is working on this.

 

A schema that supports something like this would be nice: 

<varGrp ID="VG8F1" type="subject" var="V5 V6 V7">

  <labl>Kindness</labl>

  <conditional>

    <!-- The following conditionals would specify that this 

      -- varGrp is to be displayed and asked of the user 

      -- IF (V1 == 5) OR (V2 >=3 AND V3 == 2) -->

    <if var="V1" equals="5"/>

    <if var="V2" greaterOrEqualTo="3">

       <if var="V3" equals="2"/>

    </if>

  </conditional>

</varGrp>

 

It is a simple schema that allows complex AND/OR structures, and various
ways to compare values (such as =, >, >=, etc.)  Specifically, I prefer this
model of the varGrp specifying what it is dependent on, rather than
individual var's specifying what varGrps depend on them.  All varGrp's,
whether they contain other varGrp's or var's directly, should support this
syntax.  

 

Now, what am I doing in the meantime?  Well, I'm misusing a tag to store the
question skipping information.  Here is a code snippet of what I'm doing:

<varGrp ID="VG8F1" type="subject" var="V5 V6 V7">

  <labl>

    Kindness

  </labl>

  <defntn>

    v1 >= 4

  </defntn>

</varGrp>

That's right.  I'm using straight text in a "defntn" tag to store the code,
and my application code interprets this syntax and produces the desired
result.  Much less than desirable, but it gets the job done, and we don't
use the defntn tag for anything else right now.

 

Those are my thoughts so far.  I have it working as is.  But if anyone has
better ideas for implementing this, please let me know.  As soon as DDI
comes up with a standard on this, I'll jump on the wagon. I really like the
DDI spec.

 

Andrew L Arnott

Web Programmer

The RELATE Institute

School of Family Life

Brigham Young University

  _____  

From: Richard C. Rockwell [mailto:richard at ropercenter.uconn.edu] 
Sent: Tuesday, January 20, 2004 9:24 AM
To: Andrew L Arnott; Data Documentation Initiative Users
Subject: Re: Adding tags to the DDI

 

Andrew, as I read your message, it seems to me that you are asking a
question entirely different from the two answers that have (so far) been
posted.  Your question seems to me to arise from survey research rather than
from information science or even from the details of the DDI and XML.

I think that you are asking about what survey researchers know as "skip
patterns," in which some characteristic of a question (which is almost but
not entirely equivalent to a "variable") is contingent upon the response to
another question.  The most common characteristic concerns whether the
question is asked or not.  

An example is found in a specific GSS series of questions.  The first
element in the series is the question "Are you a citizen of [the United
States]?".  If the answer is "Yes," the interview then proceeds to "Were you
born in [the United States]?".  If the answer to the question is "No" then
and only then is this question asked "What citizenship do you hold?".  Then
"No" respondents are also asked about their country of birth, returning to
the normal flow of the questionnaire.

Skip patterns can be considerably more complicated than this, including
contingency on the values of two or more variables.  Even the question text
can be contingent on the answer to a preceding question, although so far we
don't see that frequently.  

At the moment the DDI handles only the simplest of skip patterns, with "come
from" and "go to" elements.  This might suffice, Andrew, to meet your needs
of the moment.

The Expert and Steering Committtes are aware that much more is needed in
order for the DDI to be capable of documenting the increasingly complex
instruments used in various forms of Computer Assisted Interviewing.  Your
ideas concerning what a future release of the DDI should be capable of
documenting would be most welcome.

("At 02:10 PM 1/19/2004 -0700, Andrew L Arnott wrote:

I m fairly new to DDI, and moderately new to XML schemas.  I am using the
DDI 2.0 schema (XSD) file to validate incoming DDI files on the web.  My
problem is that DDI (apparently) does not have tags to make some variables
conditional on the values of other variables.  I have tried to write another
schema and do something like this:

<ddi:var ID= V1 &>

            <myExt:if var= v5 equals= 1 />

            <labl>&</labl>

</ddi:var>

 

But the DDI schema doesn t allow for unknown elements, I guess.  Can anyone
suggest how I can get around this problem?  If you know a different approach
to my problem, I d be very open to it.  

 

Thanks,

Andrew Arnott
_______________________________________________
DDI-users mailing list
DDI-users at icpsr.umich.edu
http://www.icpsr.umich.edu/mailman/listinfo/ddi-users

--------------------------------------
Richard C. ROCKWELL
Executive Director, The Roper Center & 
Institute for Social Inquiry
Professor of Sociology
University of Connecticut
341 Mansfield Road, U-164
Storrs, CT 06269-1164  USA
V +1 860 486-4440
F +1 860 486-6308
Richard.Rockwell at uconn.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lion.icpsr.umich.edu/pipermail/ddi-users/attachments/20040120/926d3755/attachment-0001.htm


More information about the DDI-users mailing list