Harnessing the Power of Peers
Karen Lochbaum, Joseph Psotka, and Lynn Streeter
Abstract
Over the past year, several hundred Army officers have discussed tactical and military management scenarios using
an electronic collaborative learning environment. The scenarios were
either "Think Like a Commander" scenarios
developed at Ft. Leavenworth or "Tacit Knowledge of Military
Leadership" scenarios developed jointly by the
Army Research Institute and Yale University. A standard threaded
discussion platform was modified to allow
participants to automatically find semantically similar notes and
relevant reference material in an online library
using Latent Semantic Analysis, a machine learning algorithm for text
understanding.
Greater equality of participation has long been known to be an outcome
of electronic discussion groups with
anonymity enhancing this effect. These effects were replicated here as
well in discussions involving officers of the
same rank. However, we also found that electronic discussions produced
higher quality initial and final solutions to
complex military scenarios than comparable face-to-face discussions.
The quality of the responses was graded by
military experts and by an automated grading program developed by
Knowledge Analysis Technologies. The
automatic grading system exhibited reliability as high as the military
experts.
Two forces appear to drive the superior performance and greater
learning evidenced in the online environment: (1)
peer pressure-knowing that others will read and comment on one's
solution produces more thoughtful and
complete responses even under conditions of anonymity; (2) learning
from peers-reading, reacting and
commenting on each other's notes produces a superior final solution
over face-to-face discussion. Lower ranking
officers (i.e., Lieutenants and Captains) demonstrated greater learning
from the online discussion than did Lt.
Colonels, whose initial responses were nearly optimal at the outset.
We discuss these results as well as planned enhancements to the online
environment to create an even more effective
e-learning environment.
Full Paper (PDF)
|
