Forum OpenACS Q&A: survey / simple-survey status for 4.6?

I am testing the survey and simple-survey packages for 4.6 at the
moment. They look very much alike in their fuctionality. Simple-
survey is maintained by Don and survey by Dave (the design
experience😉...

I've heard or read that for one Malte is working on an improvement of
survey or simple-survey while dotLRN has survey within its objective.
What is the current status? Shouldn't we deprecate on of the two?

Collapse
Posted by Dave Bauer on
David,

Survey is the survey package that was commissioned by Sloan for dotLRN. It should replace simple-survey. Its too bad aD chose to name packages wimpypoint-slim, simple-survey etc.. instead of just using a logical name that would make sense for the future, and make it easier to upgrade.

Survey is inspired partly by the work done my Malte for ACES, but we had to do most things a little differently to meet the requirements for Sloanspace. There aren't any scripts to upgrade from a simple-survey installation to survey, but I think we could create a way to bring surveys over from simple-survey to the new survey package.

Collapse
Posted by Matthew Geddert on
Along the same lines, it sounds like survey/simple-survey are thus pretty complete. I was wondering if anybody has a discussion about or is planning to add "grading" functionality to survey. If not, i may want to start working on that soon (after Michael Steigman and I finish porting/improving the events package from 3.x - to 4.x). Dave, would you be the person to talk to about adding this type of functionality to survey? I think it would significantly improve the usefulness of the survey package.

What i mean by "grading" is the following. If you have a multiple choice survey with a number of questions - possibly broken down into a number of sub-survey categories - grading would allow you to:

  • it would allow teachers to administer multiple choice quizzes/tests/practice tests via survey that are automatically graded!
  • it would allow people to take personality tests/quizzes, where you analyze responses in categories, and derive something about a person from each of those categories based on the score - and then you could give a response to the person per category. I.e. you seem more angry than normal, you are a thinking person (instead of a sensing person), etc.
  • Even if you don't give the analysis to the web user - you could generate useful reports for the site administrators, about the responses given by complex surveys (i.e. the average response for category 1 was 23.4, and category 2 was 32.7, etc.). I guess this would be sort of like the automatic generation of graphcs/pie charts.
Collapse
Posted by Matthew Geddert on
Let me make this clear - I don't think I did that - I am willing to do the work and ad this functionality to survey if it isn't already in progress... this is not a request to have somebody else do something for me, rather it is an inquisition into whether or not this has already been thought about, and if it is in the works, and if not how I should go about adding this functionality (unless you think this doesn’t belong in survey and I should create my own package).
Collapse
Posted by Stan Kaufman on
Matthew, I built such a beast (based on surv-simple but involving a rewrite of the data model, admin UIs, etc) in 3.2.5. I'm porting it over to 4.x since it does stuff that I need (particular for health status measures). It wouldn't be very easy to merge my package with Dave's survey at this point, and I understand that the OACS leadership is appropriately concerned with having too many duplicated, similar packages. Still, I'd be happy to help collaborate in some fashion or other, since I agree it is an important capability for a survey package to have.

If you're interested, you can see my package in action if you look at the Demos section at http://www.cvoutcomes.org/

Collapse
Posted by Don Baccus on
and I understand that the OACS leadership is appropriately concerned with having too many duplicated, similar packages.
Well ... my personal opinion is that we don't want a bunch of duplicated, similar packages "officially supported" by the OpenACS project. I personally don't see any reason to discourage people from writing what their heart desires. On the other hand I doubt the community wants to encourage the proliferation of similar packages as long as there are a bunch of things that are missing entirely still in OpenACS 4.x. Fill in the missing holes rather than repeatedly polish the same stone ...

There's going to be a lot of talk about this and related topics when we get into planning 4.7. Public talk.

Whatever is decided in terms of how we might support distribution of packages that aren't part of the "official family" I can't imagine that there won't be some level of support. Something like Ben's proposed MIST tool, something minimal like a repository for user contributed packages at openacs.org, something we've not thought of yet ... whatever it might be, it will be *something*. We want to encourage folks to write packages and to make them visible and available to users, whether or not they make it into the "officially supported" suite of packages.

Collapse
Posted by Dave Bauer on
On survey in particular. We definitely want scored or graded survey capability. Lets coordinate this. I am sure that Sloan and other dotLRN users will be interested in this, so lets work to get a proposal done for public discussion.
Collapse
Posted by Carl Robert Blesius on
We will definitely have a need for this and other features. You can count me in on this one (I might be able to find some additional funding locally for this as well), although my hands are tied till the 23rd of October.
Collapse
Posted by Ola Hansson on
I'm very happy that people bring up the idea of scored/graded surveys. It shows, I think, that there is a general demand for this feature which appears to be more or less essential to an e-learning environment.

We (Polyxena) felt the need for something (exactly) like this while we were evaluating the Curriculum module from 3.x. (BTW, we've started working on expanding Curriculum and are looking at making it a 4.x package.)

(The Curriculum module keeps track of work items or elements (urls) that are part of a course (a curriculum). It describes the involved elements, and, more importantly, acts as a progress meter by showing a progress bar at the bottom of each page. The elements in this visual curriculum bar get checked as the user visits the corresponding url.)

One piece of functionality the original Curriculum module lacks (among other) is the ability to control that the user has actually read and understood the document before moving on to another element in the curriculum. We immediately thought of the survey module as already being able to serve Curriculum with that functionality. Since it doesn't yet do that, we would be very interested in working with you on developing such graded surveys.

Ola, making Curriculum a 4.x package would be great! Eventually combining it with survey would even be better. In addition to the the obvious (and really cool) ways it could be used in dotLRN, it would make it easy to create tailored health/medical messages using the web. Examples: interactive web applications that help people quit smooking, loss weight, exercise, etc. - just something I have been thinking about.
Collapse
Posted by Matthew Geddert on
I'm glad to hear so many people wanting to contribute to a "grading" system for survey. I would like to volunteer to write out a spec doc for these additions/modifications/re-write, which I will then post here on openacs.org for community feedback. If anybody would like to help contribute to the initial proposal please contact me via email. I'm really busy right now, so I won't be able to start working on this till late next week.
Collapse
Posted by Staffan Hansson on
The ideas that have been ventilated in this thread all sounds very interesting to me. In fact, some of the issues, the didactical ones, are even within my field of expertise, for a change. But, as usual, I had trouble getting an overview of the big picture, like what is the extent of the survey project and what are the major aspects of it?

So I did what I usually do, I collected my thoughts in writing. The text I produced mainly offers a rough structure of thought, and the details (especially the technical) are to be taken with a grain of salt. Will you guys please have a look at it and see if you agree or disagree, and in what way, with my description of the extension of the Survey module that we are all contemplating.

Visit it now and you'll find that I've thrown in an extra text for you as a bonus; the exclusive first hand draft presentation of the Curriculum module we guys over at Polyxena are developing. (It's not as completed as the draft implies, though.) Of course, the same goes for that piece of work; if you have issues with it, please ventilate those in this forum.

Please look upon these thoughts as my (low-level) contribution to the initial proposal for the Survey spec doc that Matthew volunteered to provide. I took the liberty of putting it up for public scrutiny instead of sending him a private email. I just needed to paint my own mental map of what we are about to do, and the main thing for me is that my mental map resembles yours. If it doesn't, I'd like to know about it before I get lost.

Collapse
Posted by Dave Bauer on
I hope somewhere in the new curriculum package we enable addition of related packagage through service contracts. This seems the best way to allow a package such as survey to let the curriculum package know when a survey/test has been completed.
Collapse
Posted by Ola Hansson on
Dave,

I hope so too. It seems appropriate and, besides, it would give me a reason to finally learn about service contracts.

Generally speaking, what would the reasoning be behind the decision to let one package define a contract with operations and to let the other own its implementations, and not the other way around? (Instingtively I feel that Survey should own the contract, but I can't really say why.)

Also, when I compare the sc code for forums with, e.g., News I find that Forums uses "bindings" and News does not. What are they good for?

Collapse
Posted by Dave Bauer on
Ola,

I think curriculum would define the operations and for a package to "work" with curriculum, it would have to provide an implementation of that operation for its objects.

I am probably not the best person to explain it.

Are you talking about the search service contract for new and forums? They should work pretty much the same. The only difference is that forums has to define the triggers to insert items into the search queue because it does not use the content repository. News uses the content repository and doesn't need the triggers.

Collapse
Posted by Matthew Geddert on
I just wanted to let everybody know that I unexpectedly have to fly to Canada this week for a funeral, so I will likely not be able to get my proposal for "survey package expansion" up on this bboard until Monday the 21st (instead of late this week as I had previously committed to).

Staffan, thanks for submitting the document you wrote. This will certainly help me to create my proposal. I can tell you that after my initial reading, I agree with you in large part, and that we are roughly on the same page.

Collapse
Posted by Staffan Hansson on
Matthew: That's good news - and bad news. Thanks for telling us the bad news first.

All of you: I figure my lay-up of the problem implies that the Survey extensions we are discussing are of two quite distinct sorts:

  • The Survey Stats functionality is an inductive operation within Survey, suggesting to me that in this respect Survey will be the client and the graph software the service provider.
  • The Graded Survey functionality is a deductive operation within Survey, suggesting to me that in this respect Survey will be the service provider and Curriculum the client.
These were my gut feelings. They may very well be wrong, though, and that's one of the issues I was hoping we could figure out somehow.
Collapse
Posted by Dave Bauer on
Staffan,

Sounds right.

But first, graded surveys should be implemented by the survey packge, or possibly an additional pacakge (but probably not) independent of integration with curriculum.

That is, I should be able to run a graded or scored survey without using the curriculum package.

Collapse
Posted by Staffan Hansson on
Dave: My thoughts exactly.
Collapse
Posted by Michael Feldstein on
Regarding interoperability with the curriculum module, you might want to look at SCORM and the IMS Content Packaging standard from which it is derived.
Collapse
Posted by Staffan Hansson on
If I understand SCORM correctly, it's a (well-established?) standard for exporting a curriculum and its meta-data to an XML file for an external LMS (Learning Managent System) to import. Thus enabling online learning websites to offer their courses to external parties.

This is very interesting functionality indeed. If it's for real, it seems that it would be worth some time and energy (and dare I suggest money) to set up the tools for such SCORM compliance, in order to facilitate interoperability between online learning websites.

I didn't get the impression that SCORM is needed for the internal operations on the OpenACS system, though. I may have misunderstood SCORM, of course; is it in fact supposed to be a total, completely integrated, solution for the LMS? If not, the direct interoperability between Curriculum and Survey is a service contract thing after all, I suppose.

I makes me wonder, though: How far is this here Curriculum package, enforced with the graded survey functionality, from being an LMS?

Collapse
Posted by David Siktberg on
There was a very lively, extensive, and often contentious discussion about SCORM and LMSs on the Web-Based Training / Online Learning Listserv this past summer.  The context was implementing object oriented technology for training content.  What I took away is that SCORM is well intentioned, but not fully implemented or usable, though I did not follow up with an examination of SCORM details to confirm that.  I think you would be well advised to check this out.

You can subscribe to the listserv at http://www.trainingplace.com/source/thelist.html

and then request an archive listing.  Look under the headings "Learning Objects" and "RLO, LMS, SCORM, etc."

If that does not work, I can cull these items from my email folder with WBTOLL messages and forward them to anyone interested.

Collapse
Posted by Matthew Geddert on
I am about to leave for my flight, so this will be short. I agree with Dave that things like grading in survey should be independent of other packages. I'll be thinking about the integration with things like ciriculum on the plane. I have been thinking that Survey would have a number of special functionalities that could be enabled/disabled either by the use of parameters or service contracts. It is my hope that survey could stand alone and provide all the functionalities we've talked about and some more complicated ones i have been pondering without dependence on anything but the ACS core... I'll provide justification for that in the paper i am working on. i've gotta go.
Collapse
Posted by Caroline Meeks on
Actual compliance with IMS would not be a realistic short term goal. However, its a wonderful source of complied knowledge and experience.

Specifically the IMS Question & Test Interoperability Specification is likely to be an excellent source of use cases and a useful reference for defining a datamodel.

Talking the same language is the first step towards interoperablity so I think we should also use IMS to help us define our vocabulary and name our columns and variables.

Collapse
Posted by Caroline Meeks on
When you get a chance please take a look at the IMS Simple Sequencing Specification for ideas on how curriculum might work.

If you think of curriculum as a work-flow/sequencing tool, in the old 3x Curriculum there was no branching, just one order to the links. The success criteria was viewing the url. Simple and useful. The Polyxena Curriculum document describes a system with one order to the links and success criteria of passing a test.

Other use cases include, pre-tests, reviews if the user fails the test, i.e. branched sequencing. Also some lessons might be passed without tests (just viewing the material), some only by instructor approval, and maybe some with some other criteria, like posting to a bboard. Thus we need the ability to specify different success criteria.

I am totally in favor of implementing only a subset of the general case functionality, with the subset being driven by the needs of whoever is doing the work. But it would be good to put references to what is not being implemented into the design documents as well as hooks and notes to allow the next developer to add on easily.

We put a little of our vision on how we saw Survey moving into the future in our RFP. I uploaded it into the .LRN extranet on SloanSpace. Email me if you need access and don't have an account.

Collapse
Posted by Staffan Hansson on
David and Caroline: That's quite a homework you gave me... No, seriously, I appreciate it. I'll just go through the information you pointed me to and I'll get back here with my thoughts on it as soon as I'm done.

Thanks for sharing.

Collapse
Posted by Matthew Geddert on
I finished up my initial proposal, you can view it on this thread: https://openacs.org/bboard/q-and-a-fetch-msg.tcl?msg_id=0006i3&topic_id=11&topic=OpenACS
Collapse
Posted by Staffan Hansson on
OK, I'm back after a week ruled by Murphy's Law (everything that could go wrong did so). Between headaches, Ola and I grabbed the opportunity to study the IMS Simple Sequencing Specifications. And we came to the conclusion that Caroline seems pretty on top of it; they are indeed valuable sources of inspiration and implementation. At the same time, reading up on some of the older openacs.org posts, we did sense that general skepticism toward simply accepting SCORM as the final solution that David had sniffed out elsewhere. We may keep in mind that there is also the way of web services to explore before trying to export curriculums, or sequences. But, generally speaking, we liked what we saw.

As has been suggested, the exact solution for implementing a full-fledged LMS lies in the (not too distant) future, and we haven't looked for it now. What we have been looking for are any already established standards of concepts and terms, and IMS certainly seems an authoritative source of such standards. We're quite happy to use the IMS Simple Sequencing Specifications as guidelines in the future development of the Curriculum package.

I've put up our preliminary plan for implementation of the Curriculum package in several phases at our site. It's also a call for any interested educational institutions to contact us and start off the actual implementation of Curriculum. We're basically ready to go.