The panel consensus technique was designed for use
in large organisations (e.g. a military service) with a capability for
generating a large number of ideas (perhaps 4-500 or more) that would
then need to be narrowed down (Taylor, 1972).
There is no time is built in for research, it is assumed
that due to large number of people involved that the necessary knowledge
is available, therefore sensible decisions can be made based on discussion
and voting. Originally (1972), when it was described, it required a
lot of clerical and administrative support, and must have been a very
cumbersome process, implying a many-layered hierarchy; indeed the method
read like an awesome explanation for delayering! However, if repeated
nowadays, much of it might be computer and network based within a much
flatter structure, making it much simpler operationally.
The underlying picture is that of progressive filtering
through a series of selective funnels (c.f. the idea of a series of
hurdles that underlies Progressive
Hurdles).
The early phases engage large numbers of less skilful
people using fairly straightforward methods to remove less suitable
options, leaving small numbers of high-powered people to deliberate
in more sophisticated ways on the resulting short-lists.
To begin: Each panel is staffed
by a neutral administrator who looks after the paper-work, checks
time-keeping, helps with weighting calculations, etc. and there are
also an overall controller and administrator.
Idea generation phase: 24 hours are given to individuals
with some knowledge of the problem, to come up with ideas. Each problem
is presented in a comprehensive (up to 2 pages), standardised way
(Title, Problem statement, Key points of the idea, Description of
how it might be implemented). Strict anonymity is preserved. For the
latter phases to make sense, this phase needs to generate at least
4-500 ideas.
Screening Phase: The 4-500 ideas
are divided up randomly between 15 screening panels of 15 people each,
carefully chosen for their shared familiarity of the field. Each panel
is given 3-4 hours to reach consensus about the best five of the ideas
allocated to it, working via a discussion and by assigning each idea
a value on a five-point rating scale. This results in a short-list
of 75 (15 x 5) ideas to pass on to the next phase.
Selection Phase: 3 Further panels,
each of 5 middle managers selected for their expertise in the field,
are given identical sets of clean copies of these 75 ideas. In much
the same way as the previous phase though possibly with more analysis
the selection panels endeavour to reach consensus. Again, each has
3-4 hours to reach consensus about what it considers the 5 best ideas,
though this time they have to write statements justifying their choice.
There may well, of course, be duplicates amid the resulting 15 (3
x 5) lists, as the three panels are working independently in parallel.
Refining Phase: One panel of 5 highly
experienced upper-middle managers takes these 15 ideas and narrow
them down to a final short-list, with cases justifying their choices,
additionally they may simplify, develop or combine ideas as long as
their basic material remains intact.
Decision Phase: A further panel
of five top managers come to a decision on their preferred option
to pursue and how it shall be implemented