Michelene T. H. Chi and colleagues have demonstrated the value of one-on-one interactions. Interestingly peer-to-peer collaboration is effective. This is a more constrained version of team collaboration that may be easier to implement giving good results for more moderate effort. --Tom Wason

Knowledge convergence and collaborative learning

Heisawn Jeong and Michelene T. H. Chi

Instructional Science

Springer Netherlands

Volume 35, Number 4 / July, 2007

287-315 Abstract This paper operationalized the notion of knowledge convergence and assessed quantitatively how much knowledge convergence occurred during collaborative learning. Knowledge convergence was defined as an increase in common knowledge where common knowledge referred to the knowledge that all collaborating partners had. Twenty pairs of college students collaborated to learn a science text about the human circulatory system. Comparisons of individual pre-test and post-test performance revealed that students shared more knowledge pieces and mental models after collaboration. Although the amount of convergence was modest, analyses showed that collaborative interaction was responsible for the increase in common knowledge. The increase in common knowledge was observed in knowledge that was never stated in the learning text as well as in knowledge that was explicitly presented in the text. The amount of convergence was related to interaction such that real pairs shared more knowledge than nominal pairs, and more interactive pairs shared more inferred knowledge than less interactive pairs. Collaborative dialogues and learning artifacts (e.g., drawings) also indicated that common knowledge was constructed during collaboration. Possible reasons for the discrepancy between the impression of strong convergence assumed in the literature and the results of this study are discussed along with the need to develop a more comprehensive understanding of the notion that includes its process, outcome, and sources of convergence.

TY - JOUR
JO - Cognitive Science: A Multidisciplinary Journal
PB - Psychology Press
AU - Chi, Michelene T. H.
AU - Siler, Stephanie A.
AU - Jeong, Heisawn
AU - Yamauchi, Takashi
AU - Hausmann, Robert G.
TI - Learning from human tutoring
SN - 0364-0213
PY - 2001
VL - 25
IS - 4
SP - 471
EP - 533
AB - Human one-to-one tutoring has been shown to be a very effective form of instruction. Three contrasting hypotheses, a tutor-centered one, a student-centered one, and an interactive one could all potentially explain the effectiveness of tutoring. To test these hypotheses, analyses focused not only on the effectiveness of the tutors' moves, but also on the effectiveness of the students' construction on learning, as well as their interaction. The interaction hypothesis is further tested in the second study by manipulating the kind of tutoring tactics tutors were permitted to use. In order to promote a more interactive style of dialogue, rather than a didactic style, tutors were suppressed from giving explanations and feedback. Instead, tutors were encouraged to prompt the students. Surprisingly, students learned just as effectively even when tutors were suppressed from giving explanations and feedback. Their learning in the interactive style of tutoring is attributed to construction from deeper and a greater amount of scaffolding episodes, as well as their greater effort to take control of their own learning by reading more. What they learned from reading was limited, however, by their reading abilities. © 2001 Cognitive Science Society, Inc. All rights reserved.
UR - http://www.informaworld.com/10.1207/s15516709cog2504_1
ER -

The first question I'll pose on every thread is, "is there sufficient demand and current practice to adopt this into SCORM?". In this case, I don't have an opinion on the answer.

From what I understand of HLA, doesn't it do something similar to what is proposed here? I could be wrong, but it might be worth investigating if we go down this path.

Overall, I think the proposal raises some interesting points about the "how" synchronous learning with SCORM can be accomplished.  That said, I have questions, if not really concerns, on some elements of what's proposed here: 
 

"Many educators are moving away from the typical lecture-style teaching, and on to more
interactive, engaging learning methods such as team-based learning"

In what contexts?  Academic settings may be asking for this, but in the corporate space, I am seeing a demand for improvements in asynchronous instruction.  When it comes to the synchronous, people would track if it was easy to do so, but the learning leaders I'm in contact with in the corporate space don't care about the tracking -- they care more that a) the synchronous session is technically feasible; and b) that it is convenient, efficient and saves money. 

Later, with regards to the Intermediary Object, I was wondering if the Intermediary Object couldn't just be a sub-function of the run time, when running courses in a synchronous "state" or mode?  It might be easier for educators and content developers if they didn't have to build this intermediary object custom to each experience.

One element that I'm not sure about is the function for: 

Generate default data for users that are not online...

...this intermediary object could also be programmed to contain 'filler' data for
one or two of the user roles (in this case only three SCOs) so that if only one learner was
using the system, the object could randomly create data to fill in for the other two roles.
This would allow flexibility for the learners to go through as a full team, or simply as a
single player put into the simulated TBL environment.

I wonder how this could be implemented easily by content developers.   I can accept the advantage of an "NPC" in an interactive simulation (like the idea of playing with/against the computer), but applied to something less engaging... and specifically with regards to the term "random" -- where's the learning model in that?  It's technically possible.  I would argue that it may not be pedogically appropriate.

My one concern with the paper is specifically with regard to the Assessment model presented:

...each learner would second receive a score from his peers regarding his
performance within the SCO. This score would be derived from learners rating their peers
according to a specific set of criteria...

Synchronous assessment with peer evaluation -- but without moderation or facilitation?  The model suggested here *may* have applications in academia, but in just one example in the corporate world -- where pay is tied to performance on an assessment -- a wide range of legal issues surface with the model.

I can see the value in such assessments given proper context, but this doesn't appear to be universally applicable, so I have to question the need.  The bigger question then is -- will SCORM 2.0 be ported for "themes" based on the different communities of interest that need specific functionalities, or are we going to design a one-size-fits-all SCORM 2.0?

I think this submisstion addresses an interesting use-case. It does not really come up in the market I am familar with which is corporate/enterprise e-learning. However, I have the following thoughts:

 1) In the section Assessment/Individual Assessment, a scheme is described for computing an individual score by factoring in scores recieved from his peers. I just wonder in practice if accurate measures of the individual can be derived this way. Are the peers versed enough in the subject matter or scenario to give a good evaluation of the individual? Are the peers impartial (postively or negatively) enough? 

2) From a technical perspective, the LMS (or training environment) will need to manage multiple teams taking the same TBL-based content. Individuals could also be part of more than one team, where each team is taking the same or different TBL-based content. These are combinations to keep in mind to fully support this use-case. This feels like logic that would be layered on top of the LMS's sequencing engine.

There is great value in performance-based team learning and group exercises. I'm sure most reading this would agree that learners of today would rather engage in this type of group learning activity than listening to a lecture.

"Currently, courses designed using SCORM can integrate some elements of TBL. For example, chat functions can be added to SCOs in order to provide collaboration and peer review by multiple learners. The exchange of ideas can serve very useful for learners as they interact with and glean insights from one another throughout the SCOs. This chat function, however, is dependent upon learners synchronously proceeding through the SCOs. "

RE: This is possible with the SCORM of today. While this could be integrated into a sco, it might make more sense to be an independent service or feature provided by the LMS. The learners could be provided instructions to open a new broswer window and launch the chat client while the sco is still running.

 "Another possible application of TBL within the current SCORM environment would be to have the learners progress through the SCOs as a team. This is not ideal as most likely team members would have to all congregate around a computer to complete the training."

RE:  This is possible with SCORM today. I have witnessed classroom situations where an instructor leads a class of 30+ learners through SCORM content. Even though it was intended to be experienced as self-paced they go through the content together (each having their own workstation) and discuss their learning. This is more of a "blended" learning strategy, but could most likely falls under the confines of TBL. This would allow each learner to receive their own scores rather than trying to create a team report.

Summary: I do like the proposal for calculating a score based on both individual performance and team performance. Perhaps this could be accomodated in the CMI data model by writing to a new element that is dedicated to storing information about a learner's performance when part of a team exercise. Supporting such things as TBL would require much more detailed implementation instructions for both the content developer and a LMS. With the exception of calculating the scores automatically from a group, this type of activity can be supported today. In fact, many LMS applications have support for more than just self-paced training. Instructors can enter in scores for many types of training in a LMS. However, calculating group scores is  not standardized and could be addressed in SCORM.