[DDI-users] Questions related to QuestionGrid
Wendy Thomas
wlt at umn.edu
Fri Nov 21 11:16:27 EST 2014
Below
On Thu, Nov 20, 2014 at 10:46 AM, Wackerow, Joachim <
Joachim.Wackerow at gesis.org> wrote:
> Wendy
>
>
>
> Thank you for the quick answer.
>
>
>
> I have still a couple of comments and questions, see below.
>
>
>
> Achim
>
>
>
>
>
>
>
> *From:* ddi-users-bounces at icpsr.umich.edu [mailto:
> ddi-users-bounces at icpsr.umich.edu] *On Behalf Of *Wendy Thomas
> *Sent:* Mittwoch, 19. November 2014 19:23
> *To:* Data Documentation Initiative Users Group
> *Subject:* Re: [DDI-users] Questions related to QuestionGrid
>
>
>
> see below
>
>
>
> On Wed, Nov 19, 2014 at 11:45 AM, Wackerow, Joachim <
> Joachim.Wackerow at gesis.org> wrote:
>
> I looked at a use case of the ESS (European Social Survey) round 6
> questionnaire. Now I have some questions and comments related to
> QuestionGrid.
>
> With some of them I’m pushing probably the limits of the current approach
> in 3.2.
>
> Any thoughts on the questions and comments would be appreciated.
>
>
>
> 1. Resulting variables
>
> One use case is that each question (i.e. row) in a grid results in a
> variable. The variable can only reference the whole QuestionGrid, but not
> the individual question. This is not really satisfying.
>
> Are there any other possibilities?
>
>
>
> This is the reason for In/Out Parameters and generation instructions that
> support the distribution of looped questions or grid response arrays. The
> variable can reference a generation instruction and define the binding as
> well as the item in the array that it uses. See examples created at
> Dagstuhl training this year.
>
> *[JW] I’m not sure which examples you mean. The whole approach with In/Out
> Parameters for the purpose to make a relation from a variable to a piece of
> a question grid seems to be quite complicated. Please elaborate on this,
> also regarding the related (3) with question number.*
>
> *(the source questionnaire is still in PAPI, other modes may use other
> numbers)*
>
> I've attached the set of example for InOutParameters
>
>
>
> 2. Reuse of single questions in a grid
>
> A single question in a grid (or a subset of a grid) could be reused in
> another grid or as single question (this happens sometimes in new waves of
> the same survey). But questions in a grid are realized as
> CodeScheme/CategoryScheme.
>
> A single code could be referenced to reuse the Code/Category as question
> in another context, but this seems to be awkward. Did I miss something?
>
>
>
> Is the "question" a part of dimension CodeList or a response domain?
> each can be reused as needed. Note that any use of a CodeList can be
> limited to a level, range or object.
>
> *[JW] I think I meant the use case where real/complete questions are in
> the rows of the grid. An individual question could be used as a single
> question in another context.*
>
> *Maybe this is then not a real grid but a collection of multiple question
> items (like in 3.1). What should then be used in 3.2 for this purpose?*
>
>
>
I happened to be on the Data Capture call today as they needed a convener.
We talked about this a bit as well as issues of differentiating within a
question text what is "introductory" but necessary to the question and what
is the "inquiry". As to the above, if each line is an understandable
question in itself then my assumption is that this is a sequence of
question items. If they are understandable questions then are they really
bound by a common single question or an instruction to say rate them on a
scale or something like that.
If the "questions" are really specifications of a single question that is
required to make sense then in 3.2 they should be a grid, using the
codelist with categories to specify. The first case above (real
understandable questions) should really be done as sequences in both 3.1
and 3.2 as the MultipleQuestionItem in 3.1 was designed to handle only very
simple girds, not as a means of bundling (sequencing) questions. Of course
people used it however they wished.
> 3. Reuse of QuestionGrid
>
> How can QuestionGrids be reused and a question number be assigned to each
> individual question (in a row)?
>
>
>
> Once again, the same way any question is reused. The path of the data can
> be defined by In Out parameters, a Generation Instruction and a Variable
> defining the binding between the out parameter of the question construct
> and the in parameter of the instruction.
>
>
>
> 4. Alternative to QuestionGrid
>
> In the use case (mentioned below), a QuestionGroup could be another
> option. Then the issues 1-3 from above can be solved. The semantics are a
> little different.
>
> A workaround could be to use a QuestionGrid which refers by BasedOnObject
> to the QuestionGroup.
>
>
>
> Question Groups are administrative objects and cannot be used as the
> content of a QuestionConstruct.
>
> *[JW] OK. How can then MultipleQuestionItem (3.1) be replaced beside with
> QuestionGrid*
>
>
>
> If its was used only to bundle questions in a sequence then it should be
handled as a sequence. This was, in fact, raised during the meetings
discussing the new question structures which included ABS, INSEE, and
others. The answer then as now is that the sequencing of questions has
always been expressed as the control construct Sequence.
> 5. Dimensions in a grid
>
> I looked at the example (slide 14, 15) in the EDDI13 presentation “To 3.2
> or Not to 3.2? That is the Question” at
> http://www.eddi-conferences.eu/ocs/index.php/eddi/EDDI13/paper/view/117.
>
> The slide can show only a XML fragment. Now I’m wondering how “Older than
> Patient” and “Younger than patient” are realized, when “Number still alive”
> would be rank 1 (is rank 2 not the appropriate solution?).
>
>
>
> Related question: How can it be described when one dimension has a
> different type of response domain than the grid? A GridDimension can have
> only a CodeDomain.
>
> *[JW] Sorry, I made apparently two topics with (5). Do you have comments
> on this?*
>
> The CodeDomain represents the content of the dimension in the same way
that a variable is used to describe an NCube dimension. The actual
response. Actual response domains for the data captured can be assigned to
individual cells in the grid so that I can collect multiple types of data
from a single grid (code, numeric, text, etc.). I'm sure I have examples
from INSEE but I'm on vacation today so I'll have to dig them out later.
> 5. Misleading semantics regarding questions in a grid?
>
> The individual questions in a grid are not realized as QuestionItem but as
> Code/Category. The semantics seem to be wrong. Why was the use of
> QuestionItem not possible? This would have solved the issues 1 and 2 from
> above.
>
>
>
> The question is asked regarding a number of objects (defined in a
> CodeList/CategoryScheme or roster.
>
>
>
> 6. Mix of representational and structural information?
>
> I’m not really clear about this. But my impression is that there is not a
> clear distinction between structural and representational information.
>
> There seem to be multiple levels (in the use case below):
>
> - The atomic level of an item (in the grid), like a question or an
> answer category.
>
> - The compound object of question and answer categories.
>
> - The question sequence (in a grid) with the same answer
> categories.
>
> - Some representational/rendering information (for example the
> answer categories are not repeated but only mentioned at the top). This
> could be differently arranged in different modes.
>
>
>
> There are those looking at 4 that say all grids could be defined as loops.
> We need to look at this further as the response of INSEE staff to this
> suggestion (who use grids extensively) was a definite shudder.
>
> This is probably more an issue for DDI 4 and will need some further
> analysis.
>
>
>
> Reference:
>
> ESS round 6 questionnaire
>
> Page 5, starting with “CARD 5 Using this card, please tell me on a score
> of 0-10 how much you” in the document at:
>
>
> http://www.europeansocialsurvey.org/docs/round6/fieldwork/source/ESS6_source_main_questionnaire.pdf
> .
>
>
>
> Achim
>
>
>
>
> _______________________________________________
> DDI-users mailing list
> DDI-users at icpsr.umich.edu
> http://lists.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 umn.edu
>
> University of Minnesota
>
> 50 Willey Hall
>
> 225 19th Avenue South
>
> Minneapolis, MN 55455
>
> _______________________________________________
> DDI-users mailing list
> DDI-users at icpsr.umich.edu
> http://lists.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 umn.edu
University of Minnesota
50 Willey Hall
225 19th Avenue South
Minneapolis, MN 55455
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.icpsr.umich.edu/pipermail/ddi-users/attachments/20141121/c8667b0d/attachment-0001.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: InOutBind.zip
Type: application/zip
Size: 580105 bytes
Desc: not available
Url : http://lists.icpsr.umich.edu/pipermail/ddi-users/attachments/20141121/c8667b0d/attachment-0001.zip
More information about the DDI-users
mailing list