Mechanics of Undergraduate Research at Liberal Arts Colleges

1
Mechanics of Undergraduate Research at Liberal
Arts Colleges Lessons Learned

• Amruth Kumar
• Ramapo College of New Jersey
• Mahwah, NJ 07430-1680
• amruth_at_ramapo.edu

2
Experience

• Intelligent tutors for Computer Science I,
  Programming Languages
• Developing a new tutor
• Large scale software development, integration,
  testing, documentation
• Extending the feature set of tutors
• Artificial Intelligence, HCI
• Evaluating a tutor
• Computer Science education research

3
Experience

• 1996 2000
• What can I do for you?
• Independent Study
• 2000 2006
• What can you do for me?
• Grant-supported
• 2007 -
• What can you do for me altius, citius, fortius?
• Honors project

4
The Ideal Candidate

• Right mix of ability and motivation
• Success 10 inspiration 90 perspiration
• Ideal student 40 ability 60 motivation
• The 4.0 student on a 3-year degree program
• Show me how to do it and I will do it student

5
Recruiting

• Pro-active Approach select students
• Reactive Publicize, wait for students to
  approach you
• ?
• Proactive better than reactive
• 2 1 Ratio
• Recruited Retained
• Retained Participate
• Participate Excel

6
Guidance

• Highly regimented
• Weekly meetings
• Step-by-step instructions
• Vs
• Loosely structured
• Schedule, reporting, tasks, etc.
• ?
• One size does not fit all
• Can lose a student with the wrong strategy!

7
Managing Expectations

• Contract drawn up upfront
• Helps the student measure progress
• Written in students words!
• Detailed weekly log
• Records locus of thinking, rationale for
  decisions, hours worked
• Conference deadlines as deadlines

8
Managing Expectations

• Development
• Short cuts in software design, implementation
• Software never comes out right on the first try!
• Evaluation
• Inadequate features, interfaces, services
• Good enough is not good enough!
• Writing up the results
• Write rewrite cycles
• Focusing literature search
• Presentation
• More than reciting the paper
• Practice makes perfect

9
What you can do for me research

• Development
• Core system on a need to know basis
• Design for reusability
• Implementation not done until integrated
• Vet everything, leave nothing to chance
• Evaluation
• Access to previous protocols, data formats,
  results
• Writing up the results
• Database of references

10
Building a community of researchers

• Friday evening jam sessions with pizza coke
• Good for beginning students
• Morale booster
• May not work once students mature
• Participation in professional conferences
• Participants mature - the experience,
  recognition and networking
• Just as influential on other students as the
  participants stories and legends

11
Conclusion

• Convincing them they need to re-implement the
  solution 20-40 hours
• Cajoling them to clean up the software so that
  their work can be tested 2 3 revisions
• Reminding them of conference deadlines 5 6
  email messages
• Getting them to write a professional paper 8-10
  revisions
• ?
• Seeing the appreciation of research dawn on their
  faces
• PRICELESS!