Saturday, December 27, 2008

Promoting Research Group Spirit and Peer Student Support

Posted on Saturday Oct 06, 2007

Earlier I didn't emphasize much on research group spirit. Recently I realized its importance and tried some measures to promote research group spirit.

I found that UIUC's Prof. Jiawei Han's several measures in his data mining research group could be valuable to borrow. I borrowed them recently in my group.

1. Allow students to volunteer to take on some services in the group. In the past, I (as the advisor) took on most of the services in the group including maintaining the group web pages, coordinating the group meetings, etc. Then students might feel like being managed without feeling to own the research group. In addition, I am too busy in doing these types of things and the students don't learn how to organize things or manage things: an important skill in their future career.

In the research group, early this semester I asked students to volunteer to take on various roles in the group:

*. Group Webmaster (news, group Web page, pictures, etc)
*. Group meeting coordinator
*. Server system administrator
*. Industry/visitor coordinator
*. Conference and journal review coordinator
*. Research proposal coordinator
*. Social activity coordinator

I found this mechanism works pretty well. For example, recently when a visitor from industry gave a guest lecture in my course when I was out of town, I asked the industry/visitor coordinator to organize student meetings with the visitor by introducing our research and doing demo; the whole process was organized by the coordinator with help from other students. The process went well and the students can also improve their independent skills: when the advisor is not around (in the future after they graduate, their advisor won't be around!), they can still successfully carry out things.

But I still need to figure out a way to encourage students to send emails in our group mailing list, whose emails are primarily sent by myself.

2. Acknowledge and honor those students who made great achievements in research so that these students can feel being recognized and other students can learn from these students and try to catch up. Jiawei Han's group honors the best-performing students each semester after students submit their research performance summary for the semester. Recently our research group also held voting among students (each one vote) and myself (with two votes, as suggested by one student, saying that my judgment would be more comprehensive). In the end, we voted one golden award winner and two silver award winner (with the same number of votes).

3. Besides borrowing Jiawei Han's measures, I also tried to promote peer support among the students in the group. Earlier the whole group activities centered around me, including reviewing their paper drafts, giving feedback on their research, etc. I would hope to set up a peer support system so that students can help each other and learn from doing so. Since some time ago I encouraged students to do proof reading each other's papers, and help each other. I will think of more other measures in promoting peer support.

4. As a routine practice in many research groups, asking students to present their own work or other related work by other researchers is quite valuable. Earlier I used the group meeting time slots to go round-table debriefing and I found it not that worthwhile in spending time. Nowadays, instead, in each group meeting, each student makes a presentation and then other students and I give feedback either on the content or presentation skills. Again, in this way, the group meetings shift from being dominated or driven by myself to being managed by students themselves.

I will think of more other measures in promoting peer support and group spirit. If you have any comments, you are welcome to discuss here.


  1. I do not think Mechanism 2 is a good idea. For one thing, if the research group is not very large, everyone can see each other's progress and it should be clear to everyone that who is the best. For another, there are many issues affecting research. Having no result does not mean the student was not working hard. Ranking students will frustrate the low-ranked students, who probably have already been frustrated for producing no result.

    Yingfei Xiong

  2. There is a possible solution for Yingfei's concern. Assume there are 10 students in a group. After collecting the votes, the counting should be done privately by the advisor, and at the end, only the top 2 ranks should be announced instead of disclosing all the vote counts and all the ranks (1-10) in public among students. That way, everyone can be motivated by the top two performers. Other 8 students wont know their exact rank, but will still be motivated to achieve the top 2 spots next semester. If all ranks are disclosed, it might be a bit frustrating for the 10th ranker!

    A different concern exists however. We have assumed there are 10 students. A perfect ranking algorithm exists to accurately rank the students - "only-advisor-votes-and-students-wont"! this algorithm is perfect because advisor knows the performance of each and every student and can easily rank the students precisely. Student voting (even with an extra vote for the advisor) may not always yield a perfect ranking because as a student one might not exactly know how much effort other students have put the previous semester. So, what are the advantages of student voting methodology (an approximate algorithm), when a perfect algorithm (advisor voting) already exists? Transparency cannot be an advantage because we all trust our advisor anyways!

    Mithun Acharya

  3. Note that in practice we don't rank all students (giving scores to each student) but asking students to vote the best two in their opinion.

    Indeed some research topics can take a while to generate results/papers. Students and I vote based on our overall impression, not solely based on results/papers. It is possible that a student with a single important, high-impact paper could win over another student with several ok papers.

    Indeed, it may have side effect like frustrating those students who cannot get awarded. But I think these students shall turn the frustration into incentive for doing better and making greater progress. For these students, living in their comfort zone without incentive wouldn't be a better alternative.