I have to say I am a little surprised with a long elaborate reply.
It's a pat on the back in one way.
I guess at this point of time, I have to clarify what I intended to do.
(1) Manipulating peer effects to produce desired outcomes was not the
objective. The objective was to treat the negative peer effects as
contagious diseases and have timely interventions at appropriate
nodes. Putting it another way, the intention here is to model the
spread of apathy and trace it back to it's sources so that we have
good information upon which action can be taken.
(2) Measuring procrastination is a way to measure the negative peer
effects. Procrastination has got direct relation with our interests,
our kneejerk reactions to taking up a challenge (if it is too
difficult, we wouldn't attempt it - or push it back.
http://en.wikipedia.org/wiki/File:Challenge_vs_skill.svg)
(3) Bottom up interventions are a sure shot way to target improvements
in classrooms. For example, you can not change the syllabus/challenge
levels/incentives and expect such top down changes to trickle down
into new habits formed or new things learned. So, in the era of
computers, bottom up interventions can be carried out with much ease.
(3.1) How do models help here? Models might help here in identifying
the students who would most readily learn new things. Perhaps the
absence of certain negative habits, or presence of positive attitudes
can help us judge who would pick things up easily, and then insist
within limits of modesty (this might be done by student bodies, like
perhaps a Guidance and Counseling Unit) that they finish up these
courses. We cannot expect such one-off sessions to spiral upward, but
atleast we are not letting students spiral downwards.
Consider the following:
(1) We have certain observations of the student life, including
habits, learning abilities, and the spread of apathy.
(2) We know that crystallizing these in models would be much more
beneficial, because it is merely a way to get our knowledge being
expressed more clearly. It does not introduce artificial/crude
elements. It mentions inaccuracy whenever it cannot explain/predict
something.
(3) One probable use of the model could be to search for
interventions. That, however, does not mean that our sole purpose is
to seek out interventions. The purpose is to crystallize our
knowledge/observations into models. Once that is done, we can always
think of ways in which that model can be used. However, seeking out
possible interventions is not a way to go.
My suggestion is that we discuss these effects on the Wiki and begin,
after deliberation, the modeling of these systems.