Markup language for questionnaires

[Stian Håklev]

In my work with educational research, I often encounter questionnaires. I
wonder if anyone know about existing or propose markup-languages for
writing questionnaires? I did a bit of searching, but found only some
academic papers and some mentions of heavy XML frameworks from the 1990's -
nothing that looked very current or useful.

My eventual vision would be to have something very light-weight, perhaps
Markdown-based, and be able to generate both web and paper questionnaires.
(I don't know if there's any good questionnaire websites that have APIs for
ingesting or importing question setups?)...

The second step would be able to automatically generate some R code, for
example, to parse the incoming data from the web questionnaire service. I
like Google Forms, but I always end up having to write a bunch of boiler
plate, to change field names, convert things to ordered factors, etc. I
should be able to specify in the questionnaire markup file that something
is an ordered factor (like a Likert-type Not at all, somewhat, neutral
etc), and get the data cleanup for free...

Another advantage of a simple text-based format is that it would make it
much easier to share and fork, diff etc questionnaires. There is a huge
amount of standard questionnaires in educational science, for example, but
many of them are encumbered with Copyright and high fee payments. And even
if they are not, you are likely just to get them as PDF and having to
retype them. I'd love a Github repo full of open sourced validated
instruments for testing for example physics knowledge, and being able to
fork one these, make a few changes, and right away see what has been
changed, generate paper and online questionnaires, automatically clean up
the data etc...

So yeah, lot's of ideas. I'd love to hear if anyone else finds this
interesting, or if you could point me in the direction of people who are
already working on this kind of stuff.

Stian

[Thomas Leeper]

There are two standards worth looking at here:

1) Data Documentation Initiative: http://www.ddialliance.org/ This is the
current international standard for survey preservation, which includes both
survey data and metadata. I would say it is borderline impossible to
generate by hand. I've had the ambition to create an R package to generate
it, but haven't had the time. There are a couple of R packages floating
around for generating DDI from SPSS
(http://www.icesi.edu.co/CRAN/web/packages/spssDDI/index.html) and
Stata (https://github.com/ddionrails/r2ddi) files, but nothing to generate
them from scratch or really do anything else with them.

2) Triple-S: http://www.triple-s.org/. This is probably the 90's era
standard you've encountered. It's only sophisticated enough to handle
relatively basic surveys, but seems to be widely supported by a lot of
software. You might be able to write this by hand and/or craft a simple
package to do it for you.

Additionally, if you use Qualtrics, you should take a look at the features
for survey import, which include a decent markdown-esque markup language
(http://www.qualtrics.com/university/researchsuite/advanced-building/advanced-options-drop-down/import-and-export-surveys/#PreparingATXTFileInAdvancedFormatForImporting).
Perhaps that would be the model for a more general survey markup language
that could be automatically converted to DDI, Triple-S, or whatever other
format.

This is definitely something I've thought about working on, especially the
idea of generating R code to process survey results based on a
questionnaire markup document. But I want to try to avoid the XKCD
standards problem (http://xkcd.com/927/). So, if you have ideas for how to
make this work, I'd love to discuss more.

Post by Stian HÃ¥klev
In my work with educational research, I often encounter questionnaires. I
wonder if anyone know about existing or propose markup-languages for
writing questionnaires? I did a bit of searching, but found only some
academic papers and some mentions of heavy XML frameworks from the 1990's -
nothing that looked very current or useful.
My eventual vision would be to have something very light-weight, perhaps
Markdown-based, and be able to generate both web and paper questionnaires.
(I don't know if there's any good questionnaire websites that have APIs for
ingesting or importing question setups?)...
The second step would be able to automatically generate some R code, for
example, to parse the incoming data from the web questionnaire service. I
like Google Forms, but I always end up having to write a bunch of boiler
plate, to change field names, convert things to ordered factors, etc. I
should be able to specify in the questionnaire markup file that something
is an ordered factor (like a Likert-type Not at all, somewhat, neutral
etc), and get the data cleanup for free...
Another advantage of a simple text-based format is that it would make it
much easier to share and fork, diff etc questionnaires. There is a huge
amount of standard questionnaires in educational science, for example, but
many of them are encumbered with Copyright and high fee payments. And even
if they are not, you are likely just to get them as PDF and having to
retype them. I'd love a Github repo full of open sourced validated
instruments for testing for example physics knowledge, and being able to
fork one these, make a few changes, and right away see what has been
changed, generate paper and online questionnaires, automatically clean up
the data etc...
So yeah, lot's of ideas. I'd love to hear if anyone else finds this
interesting, or if you could point me in the direction of people who are
already working on this kind of stuff.
Stian
--
http://reganmian.net/blog -- Random Stuff that Matters