Three sales training seminars are tested for effectiveness; size
of organization is a blocking factor.  A total of four organizations
were chosen, one from each of four size groups.  At each organization,
15 salespeople were assigned at random to 3 groups of size 5.  The
3 groups were assigned to one of the three seminars.  After 1 month,
the relative improvement in each person's monthly sales was
measured.  

What is the model for this experiment?  What is the experimental
unit and what is the observational unit? What is the name of this
design?  Would it be proper to test for an effect due to organization
size?

This is essentially a randomized complete block design.  Size of
organization is supposed to be the blocking factor but it's
actually aliased with organization.  Viewing organization as an
effect, the design looks similar to a random effects model (ignoring
assignment of treatment to group).  The experimental units were the
groups of size 5 while the observational unit was the individual. I'd
discourage testing organization size because of the confounding.